《电子商务 E-business》阅读文献：A framework of online shopping support for information recommendations

Expert Systems with Applications 37(2010)6874-6884 Contents lists available at Science Direct Expert Systems with Applications ELSEVIER journalhomepagewww.elsevier.com/locate/eswa A framework of online shopping support for information recommendations Wen-Shan Lin,, Nathalie Cassaigne Tzung-Cheng Huan Information Systems, National Chiayi University, No 580, Sinmin Road, Chiayi 600, Taiwan, ROC b School of Informatics. The University of Manchester, P.O. Box 88, Manchester, M60 1QD, UK Graduate School of Tourism and Leisure, National Chiayi University, No 580, Sinmin Road, Chiayi 600, Taiwan, ROC ARTICLE INFO A BSTRACT Keywords The growth of e-commerce has caused problems with personalized recommendations. Although several Electronic commerce attempts have made to improve or automate the retrieval and filtering of such information, no gener E-marketing strategy oping support(Iss) framework links the semantic context of online shopping with shoppers purchases in order to improve the efficiency of online shopping support. Through the application of knowledge-modeling, this paper selects a college population to empirically investigate and establish the relationship between e- marke ing terms and shoppers'buying behavior. General online shopping and the online book purchases are selected to validate the generic framework. Two hypotheses are tested: (1)e-marketing terms are impo tant in influencing shoppers'decisions: and (2) shoppers behave differen espect to different types of buys. Experimental results indicate that shoppers perceive the impo of e-marketing terms differently whilst shopping online. Six types of shoppersare classified: (1) general-purpose, (2)security. concerned, (3)value. (4)fashionable, (5) time-sensitive, and(6)service-oriented. Results and future research opportunities are discussed. This paper serves as a basis for improving online information search shopping pu e 2010 Elsevier Ltd. All rights reserved. 1 Introduction searching information for supporting shoppers. Therefore, this pa- per applies the knowledge-modeling approach to model shoppers While the World Wide Web continues to gain in popularity, shopping behavior. We claim that the e-marketing strategies ap- accessing relevant information online is becoming difficult. Web plied by vendors to sell or promote their products online influence contents are characterized by large numbers of diverse information shoppers shopping-oriented searches. sources. In e-commerce, shoppers seek buying recommendatio This paper improves the knowledge of online shopping supports through search tasks of intelligent shopping supports(ISS), or by adopting a user-centric approach. The research objectives are: agents. It therefore becomes necessary for ISS to understand what (1)to identify and present the most influential e-marketing terms shoppers want and, at the same time, what information exists on in shoppers' decision-making: (2)to model online shoppers'deci- line. This poses challenges to ISS: not only are human minds diffi- sion-making styles; (3) to experimentally validate the research cult to understand, but the information available online is also framework among college students. ubiquitous and heterogeneous(Anupam, Hull, & Kumar, 2001 The framework was experimentally validated for cases of gen- shoppers face two main limitations. The first is that they do not its generosity for online shopping. Book purchases are selected be ave enough knowledge about the veracity of their search results cause this is a relatively mature online market(Labitzke, 1999)cat- ue to a lack of awareness about the information that commercial egorized Brenner, Zarneknowb, wittig, 1998 )as a tangible and web sites are required to provide the second is that shopping sup- homogeneous type of product. by buying this kind of tangible ports rely on firm attributes (e.g. price comparison) which may not product online, textual contents of online bookshops gain impor represent all factors taken into shoppers'considerations As a re- tance in affecting shoppers' decision-making. The paper concludes sult, shoppers do not have enough support when making their buy- with discussions and suggestions for future work. knowledge of search while shopping online to assist IsS in proposed research hypotheses. Section 3 presents research dea ing decisions. In other words, there is a need to acquire shoppers the literature and justifying th and methodology. The statistical results are addressed and ex Corresponding author. Tel: +886 5 2732895: fax: +886 5 2840929. plained in Section 4. Finally, the discussions and conclusions are E-mail address: wslin@mailncyu. edu. tw (W.-S Lin). stated in Session 5 0957-4174 front matter o 2010 Elsevier Ltd. All rights reserved. oi:10.1016eswa201003034

A framework of online shopping support for information recommendations Wen-Shan Lin a,*, Nathalie Cassaigne b , Tzung-Cheng Huan c aDepartment of Management Information Systems, National Chiayi University, No. 580, Sinmin Road, Chiayi 600, Taiwan, ROC b School of Informatics, The University of Manchester, P.O. Box 88, Manchester, M60 1QD, UK cGraduate School of Tourism and Leisure, National Chiayi University, No. 580, Sinmin Road, Chiayi 600, Taiwan, ROC article info Keywords: Electronic commerce E-marketing strategy Intelligent shopping support (ISS) abstract The growth of e-commerce has caused problems with personalized recommendations. Although several attempts have made to improve or automate the retrieval and filtering of such information, no generic framework links the semantic context of online shopping with shoppers’ purchases in order to improve the efficiency of online shopping support. Through the application of knowledge-modeling, this paper selects a college population to empirically investigate and establish the relationship between e-market￾ing terms and shoppers’ buying behavior. General online shopping and the online book purchases are selected to validate the generic framework. Two hypotheses are tested: (1) e-marketing terms are impor￾tant in influencing shoppers’ decisions; and (2) shoppers behave differently with respect to different types of buys. Experimental results indicate that shoppers perceive the importance of e-marketing terms differently whilst shopping online. Six types of shoppers’ are classified: (1) general-purpose, (2) security￾concerned, (3) value, (4) fashionable, (5) time-sensitive, and (6) service-oriented. Results and future research opportunities are discussed. This paper serves as a basis for improving online information search for shopping purposes. 2010 Elsevier Ltd. All rights reserved. 1. Introduction While the World Wide Web continues to gain in popularity, accessing relevant information online is becoming difficult. Web contents are characterized by large numbers of diverse information sources. In e-commerce, shoppers seek buying recommendations through search tasks of intelligent shopping supports (ISS), or agents. It therefore becomes necessary for ISS to understand what shoppers want and, at the same time, what information exists on￾line. This poses challenges to ISS; not only are human minds diffi- cult to understand, but the information available online is also ubiquitous and heterogeneous (Anupam, Hull, & Kumar, 2001; Domingue, Martins, Tan, Stutt, & Pertusson, 2002). Nevertheless, shoppers face two main limitations. The first is that they do not have enough knowledge about the veracity of their search results due to a lack of awareness about the information that commercial web sites are required to provide; the second is that shopping sup￾ports rely on firm attributes (e.g. price comparison) which may not represent all factors taken into shoppers’ considerations. As a re￾sult, shoppers do not have enough support when making their buy￾ing decisions. In other words, there is a need to acquire shoppers’ knowledge of search while shopping online to assist ISS in searching information for supporting shoppers. Therefore, this pa￾per applies the knowledge-modeling approach to model shoppers’ shopping behavior. We claim that the e-marketing strategies ap￾plied by vendors to sell or promote their products online influence shoppers’ shopping-oriented searches. This paper improves the knowledge of online shopping supports by adopting a user-centric approach. The research objectives are: (1) to identify and present the most influential e-marketing terms in shoppers’ decision-making; (2) to model online shoppers’ deci￾sion-making styles; (3) to experimentally validate the research framework among college students. The framework was experimentally validated for cases of gen￾eral online shopping and online book buy, thereby demonstrating its generosity for online shopping. Book purchases are selected be￾cause this is a relatively mature online market (Labitzke, 1999) cat￾egorized (Brenner, Zarneknowb, & Wittig, 1998) as a tangible and homogeneous type of product. By buying this kind of tangible product online, textual contents of online bookshops gain impor￾tance in affecting shoppers’ decision-making. The paper concludes with discussions and suggestions for future work. Section 2 begins by reviewing the literature and justifying the proposed research hypotheses. Section 3 presents research design and methodology. The statistical results are addressed and ex￾plained in Section 4. Finally, the discussions and conclusions are stated in Session 5. 0957-4174/$ - see front matter 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.eswa.2010.03.034 * Corresponding author. Tel.: +886 5 2732895; fax: +886 5 2840929. E-mail address: wslin@mail.ncyu.edu.tw (W.-S. Lin). Expert Systems with Applications 37 (2010) 6874–6884 Contents lists available at ScienceDirect Expert Systems with Applications journal homepage: www.elsevier.com/locate/eswa

W-S Lin et aL Expert Systems with Applications 37(2010)6874-6884 68 2. Related works and hypotheses and product, and does not take into account the integrated product and service information with respect to e-marketing strategies ap- 2.1. Online shopping support plied by online vendors to promote and sell their products. Shoppers have demanded more personalized information deliv- 2. 2. E-marketing strategies ery services, such as intelligent shopping agents, to obtain informa- tion about products (Ringland Duce, 1987). Moreover, the In order to model e-marketing strategies, the conventional mar- behavior of an intelligent agent should emulate human behavior, keting strategies(Benman Evans, 2004: Mariotti& Gobbi, 2001): such as reasoning and problem solving ackson, 1999). In our case, place, people, product, promotion, price and process are consid shopping agents require intelligence and knowledge to process and ered. However, not all of the conventional marketing strategies find online information to meet shoppers' needs and styles. West can be applied to online markets due to limitations and differences (1991)asserts that one of the search functions for dynamic Inter- addressed in the literature relative to electronic commerce( Chan net search is to match sellers'offers with buyers' preferences and Krant, 2000: Gallmanm, 1996: Labitzke, 1999). Several studies y Lau, Hofstede, and Bruza(2000) who states that intelligent claim that these six strategies are applicable with modifications hopping agents' memory should hold profiles of shoppers prod-(Chaffey, Mayer, Johnston, Ellis-Chadwick, 2003: Clow Baack uct requirements as well as knowledge of a particular product do- 2002). The framework of e-marketing strategies proposed in this main. Karacapilidis and Moraitis(2001)proposed a web-based paper (Table 1)indicates the relationship between this framework ommerce shopping support in which customers can delegate and conventional marketing strategies. Six categories of e-market tasks to their personal software agent. However, the shopping ng strategies are identified In this framework, the main amend- port only recognizes the criteria of performance(e.g. costs and ment is the replacement of "people"by computer-mediated safety issues). Similarly, the knowledge base of the shopping electronic commerce(CMEC); virtual shop assistants are provided. support, named Alice(Domingue et al, 2002)is limited to price The term CmEC was coined by researchers to denote online com The framework of e-marketing strategies. Strategies Price Conventional marketing Physical shop Product Price E-marketing strategy Place, the web stores Computer-mediated electronic Product Promotion Price Process, e.g. onal shopping cart Place. the web store CMEC Product Promotion nce Process (02)presented in preferred languag web store front by displaying ste map or (08)sale quick check-out product(s) discounts (014)order the ava ll ntry if it is stated products O9)Free shipping formation is giv 010)special items of privacy pol 011)new prod review about (o8.7)list price recommended shipping option support 019)FAQ sectio (o18)gt service(020)retum polio icture is displaye Fig. 1. A hierarchy of e-marketing terms

2. Related works and hypotheses 2.1. Online shopping support Shoppers have demanded more personalized information deliv￾ery services, such as intelligent shopping agents, to obtain informa￾tion about products (Ringland & Duce, 1987). Moreover, the behavior of an intelligent agent should emulate human behavior, such as reasoning and problem solving (Jackson, 1999). In our case, shopping agents require intelligence and knowledge to process and find online information to meet shoppers’ needs and styles. West (1991) asserts that one of the search functions for dynamic Inter￾net search is to match sellers’ offers with buyers’ preferences and by Lau, Hofstede, and Bruza (2000) who states that intelligent shopping agents’ memory should hold profiles of shoppers’ prod￾uct requirements as well as knowledge of a particular product do￾main. Karacapilidis and Moraitis (2001) proposed a web-based e￾commerce shopping support in which customers can delegate tasks to their personal software agent. However, the shopping sup￾port only recognizes the criteria of performance (e.g. costs and safety issues). Similarly, the knowledge base of the shopping support, named Alice (Domingue et al., 2002) is limited to price and product, and does not take into account the integrated product and service information with respect to e-marketing strategies ap￾plied by online vendors to promote and sell their products. 2.2. E-marketing strategies In order to model e-marketing strategies, the conventional mar￾keting strategies (Benman & Evans, 2004; Mariotti & Sgobbi, 2001): place, people, product, promotion, price and process are consid￾ered. However, not all of the conventional marketing strategies can be applied to online markets due to limitations and differences addressed in the literature relative to electronic commerce (Chan & Krant, 2000; Gallmanm, 1996; Labitzke, 1999). Several studies claim that these six strategies are applicable with modifications (Chaffey, Mayer, Johnston, & Ellis-Chadwick, 2003; Clow & Baack, 2002). The framework of e-marketing strategies proposed in this paper (Table 1) indicates the relationship between this framework and conventional marketing strategies. Six categories of e-market￾ing strategies are identified. In this framework, the main amend￾ment is the replacement of ‘‘people” by computer-mediated electronic commerce (CMEC); virtual shop assistants are provided. The term CMEC was coined by researchers to denote online com￾Place, the web store CMEC Product Promotion Price Process (O1) offered product(s) (O2) presented in preferred language (O3) web site is located in a specific country if it is stated (O4) full contact information is given (O5) the availability of physical shop (O6) assortments of products (O7) the availability of privacy policy (O8) sale: today's special/ time-limited sale/ bundle sale/ online-only deals/ special offers/ discounts (O9) Free shipping (O10) special items (O11) new product (O12) recommended products (O13) online payment option / quick check-out (O14) order checking facility (O15) the friendliness of web storefront by displaying site map or navagational tool (O16) customer serivce/online support (O17) International shipping option (O18) gift service (O19) FAQ section (O20) return policy (O21) customer review about product (O22) product picture is displayed (O8.6) Sale price (O8.7) list price Fig. 1. A hierarchy of e-marketing terms. Table 1 The framework of e-marketing strategies. Framework Strategies Place People Product Promotion Price Process Conventional marketing strategy Physical shop Shopkeepers Product Promotion Price Process E-marketing strategy Place, the web stores Computer-mediated electronic commerce, e.g. navigational tools Product Promotion Price Process, e.g. shopping cart W.-S. Lin et al. / Expert Systems with Applications 37 (2010) 6874–6884 6875

W.-S. Lin et al/ Expert Systems with Applications 37(2010)6874-6884 Environment stimul (e. g, advertisement, interpersonal observation INPUTS Attentional and perceptual filter ation of wants C Memery(menton needs, necessary actions, outcomes perience, beliefs, Brand attitudes OUTPUTS Brand purchase intentions Social, economic, cultural, political impeding/ Fig. 2. Consumer information processing process [ Source: Baker, 1999) merce( Chaffey et al., 2003; Kiang, Raghu, Shang, 2000). This pa- available to shoppers. The process consists of the mental treatment per interprets CMEC as a vital differentiation between general and of this data as the consumer stores it, links it with existing ideas e-marketing strategies in the sense that the friendliness of web and memories, and evaluates its relevance to his or her goals tores(e.g. their navigational or search tools)is provided on a com The outputs are the consumers opinions about this data(e.g. an puter-mediated basis in electronic commerce ntention to buy or to postpone buying). The central information The term"customer service"has traditionally categorized brick process is most relevant to our study. In general, consumers partic and mortar shops as"process, "offering services at any point in the ipate in the information processing process. shopping process. However, it is included here under the categor A diagram is proposed to analyze buying behavior regarding dif- place, "since displaying service-relevant information helps shop- ferent types of products( Fig 3). Baker(1999)identifies four m pers understand the services offered by the web stores. Therefore, ket initiators in relation to customers'different buys customers can locate this information at their convenience before shopping. Another point is the category of"price. "No e-marketing(1)Complex consumer buy occurs when the consumer is highly term is directly linked to this category: however, it does affect volved and perceives the product as discontinuous. For shoppers buying behavior as relevant to their budget. The second this type of product, shoppers go through a cognitive learn- point needing clarification is Internet security. No direct e-market ing process (i.e. information search, brand evaluation, ing term is related to this concept. However, a level of web site detailed post-adoption appraisal). security is defined by the level of information provided about the (2)Dissonant buying occurs when the consumer is highly commercial side(e.g. full contact information, the availability of nvolved but sees no significant differences among the privacy policy, or the availability of a physical shop). Therefore rands, and buys the product in a hurry. Such consumers oth this"price"category and the concept of shop from a secur re likely to seek alternative brands that will meet their web site are included in the framework of shopping support Shop expectations pers'different buys are presented below. (3) Habitual buy occurs when there is no significant difference Based on the framework of e-marketing strategies, a hierarchy mong brands and the consumer is minimally involved in of e-marketing terms perceived by shoppers as important is cre- the purchase Extended information processing is unneces- ated based on the observations of well-known web sites(Fig. 1 ary, and experience is the safest guide Twenty-two e-marketing terms are considered important in influ (4) Variety-seeking buy occurs when there is little involvement encing a shopper's buying decisions and brand proliferation. The consumer chooses something To test the importance of e-marketing terms in affecting shop- new to relieve boredom pers'behavior in the decision-making process, Hl is proposed: Such behavior is typical of low-involved innovators. H1: E-marketing terms are important in influencing shoppers Nevertheless, shoppers in physical stores are categorized 2.3. Consumers' different types of buys ferent mixtures of product information are provided to meet online shoppers'needs. To take this notion into our study, different styles Consumer buying decisions are cognitive processes: an i of online shoppers search and evaluate information presented in tual sequence of thinking, evaluating and deciding Fig. 2 web shops by judging the perceived level of importance of e-mar rizes these processes(Baker, 1999). In Fig. 2, the inputs are keting terms. In order to classify types of online buys, the literature in traditional marketing is examined. Several types of shoppers are IListsofrecommendedwebsitesforbuyingproductsprovedbyhttplidentifiedanddocumentedasthemarketsegmentationstrategies (Ghosh, 1994: Liu Wei, 2002; Stephenson Willett, 1969: Swait

merce (Chaffey et al., 2003; Kiang, Raghu, & Shang, 2000). This pa￾per interprets CMEC as a vital differentiation between general and e-marketing strategies in the sense that the friendliness of web stores (e.g. their navigational or search tools) is provided on a com￾puter-mediated basis in electronic commerce. The term ‘‘customer service” has traditionally categorized brick and mortar shops as ‘‘process,” offering services at any point in the shopping process. However, it is included here under the category ‘‘place,” since displaying service-relevant information helps shop￾pers understand the services offered by the web stores. Therefore, customers can locate this information at their convenience before shopping. Another point is the category of ‘‘price.” No e-marketing term is directly linked to this category; however, it does affect shoppers’ buying behavior as relevant to their budget. The second point needing clarification is Internet security. No direct e-market￾ing term is related to this concept. However, a level of web site security is defined by the level of information provided about the commercial side (e.g. full contact information, the availability of privacy policy, or the availability of a physical shop). Therefore, both this ‘‘price” category and the concept of shop from a secure web site are included in the framework of shopping support. Shop￾pers’ different buys are presented below. Based on the framework of e-marketing strategies, a hierarchy of e-marketing terms perceived by shoppers as important is cre￾ated based on the observations of well-known web sites1 (Fig. 1). Twenty-two e-marketing terms are considered important in influ￾encing a shopper’s buying decisions. To test the importance of e-marketing terms in affecting shop￾pers’ behavior in the decision-making process, H1 is proposed: H1: E-marketing terms are important in influencing shoppers’ decisions. 2.3. Consumers’ different types of buys Consumer buying decisions are cognitive processes: an intellec￾tual sequence of thinking, evaluating and deciding. Fig. 2 summa￾rizes these processes (Baker, 1999). In Fig. 2, the inputs are the data available to shoppers. The process consists of the mental treatment of this data as the consumer stores it, links it with existing ideas and memories, and evaluates its relevance to his or her goals. The outputs are the consumer’s opinions about this data (e.g. an intention to buy or to postpone buying). The central information process is most relevant to our study. In general, consumers partic￾ipate in the information processing process. A diagram is proposed to analyze buying behavior regarding dif￾ferent types of products (Fig. 3). Baker (1999) identifies four mar￾ket initiators in relation to customers’ different buys: (1) Complex consumer buy occurs when the consumer is highly involved and perceives the product as discontinuous. For this type of product, shoppers go through a cognitive learn￾ing process (i.e. information search, brand evaluation, detailed post-adoption appraisal). (2) Dissonant buying occurs when the consumer is highly involved but sees no significant differences among the brands, and buys the product in a hurry. Such consumers are likely to seek alternative brands that will meet their expectations. (3) Habitual buy occurs when there is no significant difference among brands and the consumer is minimally involved in the purchase. Extended information processing is unneces￾sary, and experience is the safest guide. (4) Variety-seeking buy occurs when there is little involvement and brand proliferation. The consumer chooses something new to relieve boredom. Such behavior is typical of low-involved innovators. Nevertheless, shoppers in physical stores are categorized according to their styles. Vendors target styles of shoppers in order to meet shoppers’ needs and maximize profits. In other words, dif￾ferent mixtures of product information are provided to meet online shoppers’ needs. To take this notion into our study, different styles of online shoppers search and evaluate information presented in web shops by judging the perceived level of importance of e-mar￾keting terms. In order to classify types of online buys, the literature in traditional marketing is examined. Several types of shoppers are identified and documented as the market segmentation strategies (Ghosh, 1994; Liu & Wei, 2002; Stephenson & Willett, 1969; Swait Social, business, cultural, political and economic environment Environment stimuli (e.g., advertisement, interpersonal observation) Attentional and perceptual filter Interpretation, formation and evaluation of wants, needs, necessary actions, outcomes Brand beliefs Brand attitudes Brand purchase intentions Social, economic, cultural, political impeding/ facilitating conditions Response (purchase/rejection) INPUTS CENTRAL PROCESSING OUTPUTS Short-term memory Long-term memory Experience, beliefs, attitudes goals, evaluative criteria, etc. Fig. 2. Consumer information processing process [Source: Baker, 1999]. 1 Lists of recommended web sites for buying products proved by http:// www.bizrate.com. 6876 W.-S. Lin et al. / Expert Systems with Applications 37 (2010) 6874–6884

ems with Appli 68 High Involvement Complex Buying Dissonant Buying Adaptors Innovators Habitual Buying Variety-Seeking ow Involvement Fig 3. Decision styles of market initiators(Baker, 1999). Shopper shopping based market segmentatio Market segmentation approach Author hopping oriented segments in physical shops Stephenson and willett(1969) he co he recreational shopper The price-bargain shopper The store-loyal shopper Fashion orientation segments in womens apparel market in physical shops Ghosh(1994) ality conscious Value oriented segments in physical shops Swait and Sweeney(2000) shion enthusiasts Timid and uninvolved Surfer and searcher segments in web stores This paper (1) Security-concerned shopper (4)Time-sensitive shopper (5)Service-oriented shopper able 3 Shoppers decision-making styles. Group Decision style ation needs Search criteria ites designed to diverse audiences with necessary pieces of Check information richness ation Searcher Security-concerned Responsive web sites Check the reliability of web site ast value Fashionable shopper information Time-sensitive shopper Useful web sites with high ordering efficiency and well organized contents ng time for first-time user should short Service-oriented Web sites with high helpfulness in terms of customer service and functional Check the level regarding to responsiveness of shopper features web sites Sweeney, 2000). Table 2 illustrates the literature and the pro- 2. 4. E-marketing semantic terms and their importance ecting posed online buys in our study Styles of online shoppers are seg- shoppers'behavior mented by the main factors that orient their buying processes fashion, value, and quality. The types of buys reflect the buying Different sets of e-marketing strategies are hypothesized as purposes. With references to the literature, this paper proposes having certain values in affecting shoppers' different types of buys two groups of online shoppers: surfers and searchers (Table 3). By relating the frameworks of e-marketing strategies to shoppers The surfer, a general-purpose shopper, prefers browsing online types of buys, a cross-framework is derived that details the e-mar information but generally does not intend to purchase. The search- keting terms for shoppers that engage in different types of buys ers study the specific information or conditions provided by web (see table 4). ites before making a purchasing decision. Five styles of shoppers in point, the general-purpose buy occurred when are identified: security-concerned, value, fashionable, time-sensi- shoppers look for web sites that offer online payment options tive shoppers, and service-oriented and shopping carts. Shoppers prefer to shop on web sites that of- In order to test and identify styles of shoppers, H2 is developed: fer customer service and full explanations. Table 5 depicts the contextual terms shoppers may look for while conductin H2: Shoppers behave differently with respect to different types ent types of buys. For the general-purpose buy, shoppers of products web sites that have full contact information customized

& Sweeney, 2000). Table 2 illustrates the literature and the pro￾posed online buys in our study. Styles of online shoppers are seg￾mented by the main factors that orient their buying processes: fashion, value, and quality. The types of buys reflect the buying purposes. With references to the literature, this paper proposes two groups of online shoppers: surfers and searchers (Table 3). The surfer, a general-purpose shopper, prefers browsing online information but generally does not intend to purchase. The search￾ers study the specific information or conditions provided by web sites before making a purchasing decision. Five styles of shoppers are identified: security-concerned, value, fashionable, time-sensi￾tive shoppers, and service-oriented. In order to test and identify styles of shoppers, H2 is developed: H2: Shoppers behave differently with respect to different types of products 2.4. E-marketing semantic terms and their importance in affecting shoppers’ behavior Different sets of e-marketing strategies are hypothesized as having certain values in affecting shoppers’ different types of buys. By relating the frameworks of e-marketing strategies to shoppers’ types of buys, a cross-framework is derived that details the e-mar￾keting terms for shoppers that engage in different types of buys (see Table 4). As a case in point, the general-purpose buy occurred when shoppers look for web sites that offer online payment options and shopping carts. Shoppers prefer to shop on web sites that of￾fer customer service and full explanations. Table 5 depicts the contextual terms shoppers may look for while conducting differ￾ent types of buys. For the general-purpose buy, shoppers look for web sites that have full contact information, customized options, Complex Buying Habitual Buying Dissonant Buying Variety-Seeking High Involvement Adaptors Innovators Low Involvement Fig. 3. Decision styles of market initiators (Baker, 1999). Table 2 Shopper decision-making styles for online shopping based market segmentation studies. Market segmentation approach Author Types of shoppers Shopping oriented segments in physical shops Stephenson and Willett (1969) The convenience shopper The recreational shopper The price-bargain shopper The store-loyal shopper Fashion orientation segments in women’s apparel market in physical shops Ghosh (1994) Price conscious Value conscious Quality conscious Value oriented segments in physical shops Swait and Sweeney (2000) Fashion enthusiasts Style seekers Classics Timid and uninvolved Surfer and searcher segments in web stores This paper Surfer: (1) General-purpose shopper Searcher: (1) Security-concerned shopper (2) Value shopper (3) Fashionable shopper (4) Time-sensitive shopper (5) Service-oriented shopper Table 3 Shoppers’ decision-making styles. Group Decision style Information needs Search criteria Surfer General-purpose shopper Web sites designed to diverse audiences with necessary pieces of information Check information richness Searcher Security-concerned shopper Responsive web sites Check the reliability of web sites Value shopper Best value Check value of produce Fashionable shopper Attractive and entertaining web sites Check the up-to-date information Time-sensitive shopper Useful web sites with high ordering efficiency and well organized contents Check the friendliness of information displayed The learning time for first-time user should be short Service-oriented shopper Web sites with high helpfulness in terms of customer service and functional features Check the level regarding to responsiveness of web sites W.-S. Lin et al. / Expert Systems with Applications 37 (2010) 6874–6884 6877

W.-S Lin et aL/ Expert Systems with Applications 37(2010)6874-6884 information about price, a product picture, promotion news and 3. Research method online payment options. While looking for this information, shop pers make a buying decision based on these e-marketing 3.1. Method terms General online shopping and book shopping are chosen for this study to test shoppers' search and decision-making behavior. 2.5. General online shopping and online book shopping Books are the most popular products over the Internet, as shown in GVUs WWw User Survey(Source: Georgia Tech Research Cor- Generalonlineshoppingandonlinebookbuyvalidatetheproporation,Usa,foundathttp://www.gvu.gatech.edu/user_surveys). posed framework. Table 6 lists the e-marketing terms that shop- Setting the buying scenario for general online shopping and online rs may perceive as important. Each term has a number for a book purchasing can gather generic and specific information about corresponding e-marketing strategy. search behavior Table 4 Cross-framework between frameworks of e-marketing strategies and shopper different buys. CMEC Promotion Proces eting strategies and different types of buys eral-purpose buy S3: Value buy √ √√√√√ 66: Service-oriented buy Table 5 The cross-framework between e-marketing strategies and shoppers' decision-making styles. Place Promotion Process S1(03)Local web site (o2)Preferred (o1)The availability (08. 1-8.5)Today s (O8.6)List (13. 1)Online payment of preferred product time-limited sale/ bundle sale/ price/(OS.7) nly deals/special offers/ (04 Contact information (o15)Friendliness 014 With order ecking feature (o5)With physical shop 012)Recommended produc assortment of (O7)With privacy and security availability of the (o11) New prod tact information (o1)The availability (OS)Seasonal sales (O8.6)Sale (014)with order checking feature (O7) With privacy and security assortments of availability of the recommendation sign 3(03)Web site that is located in (o1)The availability (oS)Special offers (o17)International (O11)Ne 010)Special 022)The availability of the The av ce design (o18)With gift service/ gift coupo S6 (020) With (o19)The availability (o14)With order of FAQ section checking feature (016) Customer services

information about price, a product picture, promotion news and online payment options. While looking for this information, shop￾pers make a buying decision based on these e-marketing terms. 2.5. General online shopping and online book shopping General online shopping and online book buy validate the pro￾posed framework. Table 6 lists the e-marketing terms that shop￾pers may perceive as important. Each term has a number for a corresponding e-marketing strategy. 3. Research method 3.1. Method General online shopping and book shopping are chosen for this study to test shoppers’ search and decision-making behavior. Books are the most popular products over the Internet, as shown in GVU’s WWW User Survey (Source: Georgia Tech Research Cor￾poration, USA, found at http://www.gvu.gatech.edu/user_surveys). Setting the buying scenario for general online shopping and online book purchasing can gather generic and specific information about search behavior. Table 4 Cross-framework between frameworks of e-marketing strategies and shopper different buys. Place CMEC Product Promotion Price Process E-marketing strategies and different types of buys S1: General-purpose buy pp p p pp S2: Security-concerned buy p p p p p S3: Value buy p p p p S4: Fashionable buy p p S5: Time-sensitive buy p p p S6: Service-oriented buy p p p Table 5 The cross-framework between e-marketing strategies and shoppers’ decision-making styles. Place CMEC Product Promotion Price Process S1 (O3) Local web site (O2) Preferred language (O1) The availability of preferred product (O8.1–8.5) Today’s special/ time-limited sale/bundle sale/ online-only deals/special offers/ discounts (O8.6) List price/(O8.7) sale price (13.1) Online payment option (O4) Contact information (O15) Friendliness of interface design (O18) With gift service/gift coupon (O9) Free shipping (O14) With order checking feature (O5) With physical shop (O6) Good assortment of products (O12) Recommended product (O7) With privacy and security policy (O20) The availability of the product’s picture (O11) New product (O12) Seasonal sales S2 (O4) Contact information (O1) The availability of preferred product (O8) Seasonal sales (O8.6) Sale price (O14) With order checking feature (O7) With privacy and security policy (O6) Good assortments of products (O21) With customer review, guarantee seal, or recommendation sign (O22) The availability of the product’s picture (O20) With return policy S3 (O3) Web site that is located in specific country (O1) The availability of preferred product (O8) Special offers (O17) International shipping S4 (O11) New products (O10) Special items (O22) The availability of the product’s picture S5 (O15) The availability of navigational/search tool, e.g. site map/ index: the friendliness of interface design (O12) Highlighted/ recommended products (O14) The availability of shopping cart/quick check-out (O18) With gift service/gift coupon S6 (O20) With return policy (O19) The availability of FAQ section (O14) With order checking feature (O16) Customer services 6878 W.-S. Lin et al. / Expert Systems with Applications 37 (2010) 6874–6884

W-S Lin et aL Expert Systems with Applications 37(2010)6874-6884 Proposed important e-marketing terms shoppers will perceive. General online shopping Online book shopping (2)web sites are written in preferred language (2)book stores that are written in preferred language 3)web sites are served in specific country where servers are located as local web sites 4) full contact informati (full contact information 5)the availability of physical shop, ie a physical shop is available in the high (5)the availability of physical shop, ie a physical shop is available in the high (6) there are good assortments of products, ie listed items are plentiful (6)there are over three ries of books available (7) the availability of privacy policy 7) the availability of privacy policy (8) sales are available (8)sales are availabl 9)free is available lable (10) spec (10) special books are listed or highlighte (11)new products are listed or highlig (11)new books are listed or highlighte (12)recommended products are listed or highlighted (12)recommended books are listed or highlighted (13)online payment option is available, ie shoppers can pay for the purchase(s)(13)online payment op ailable, ie shoppers can pay for the purchase(s) simply giving their bank cards' details 14)order checking facility is availabl 14)Order checking facility is available (15)web sites interface is user friendly (15)book shops' interface design is user friendly, i.e. it is easy to find information (16)customer service section is available 16)shopper service section is available (17)if it is not a local web site, international shipping is available (17)if it is not a local web site, international shipping is available (18)gift service is available (18)gift service is available (19) the availability of frequently asked questions section(FAQ) (19)the availability of frequently asked questions section(FAQ) 20) the availability of return policy 21) the availability of shopper review (21) the availability of shopper review (22) the availability of product picture (22)the availability of book picture 3. 2. Sampling half (57.5%)of participants were familiar with web site develop- ment, while 36.2% were familiar with the literature marketing An experimental survey is conducted to validate the proposed strategies. Respondents were asked to express their n-mak framework of shoppers sion-making styles. According to ing styles. The higher frequency of decision-making styles for both GVU's WWw User Surveys, 62.4% of Internet users have attended general online shopping and online book shopping were the value or graduated from college. It also reveals that people of 21-25 shopper (29.1% and 35.4%, respectively ) and the time-sensitive years old are strongly represented among Internet users. Therefore, shopper(23.6% and 23.6%) university students are selected as our survey subjects. The ques tionnaires were distributed to 124 undergraduate and graduate 3.4.2. Analysis of variance students. All respondents volunteered to complete the To understand the perceived importance of e-marketing strate- gies on shoppers'decision-making behavior, t-test and Pearson correlation were applied to perform analysis of variance and data 3.3. Constructs and questions Ten participants were invited to join a focus group discussion 4. results r a pilot study based on the results, the research framework of shoppers decision-making styles was acquired and tested by six 4.1. Reliability analysis constructs. Where possible, the constructs were based on the liter ature on marketing and e-marketing frameworks. Original ques To ascertain the degree of reliability of the e-marketing research tions were based on the information system and electronic commercial texts. Questions pertaining to the constructs: (1)place, Cramer, 1999: Katerattanakul Siau, 1999)is tested. There are (2)computer-mediated electronic commerce, (3)product,(4)pro- debates about what is an acceptable level of reliability(Negnevit motion,(5)price and (6)process, were measured on a five-point sky, 2002), but it is accepted that the value of approximately 0.7or above in order to proceed with any further tests Results indicate levels of shoppers' perceived importance, measured by three sets 3.4. Statistical analysis method of e-marketing terms: (1)the total 44 e-marketing terms(2)the 22 e-marketing terms for general online shopping, and ()the 22 The SPSS software package is used for data analysis in the study. e-marketing terms for online book shopping. The results show that The data analysis methods are stated in the followings. the standardized items'alpha, are 0.8883 for the combined study 0.8011for 3.4. 1. Descriptive statistics general online shopping, and 0. 8042 for online book shopping. This This experimental survey was conducted among university stu- demonstrates high instrument-reliability with a high internal con- dents in Manchester, UK over one month. Results of this experi- sistency; therefore further tests are carried out ment likely reflect the student population. Participants are predominantly students(82%)and university staff (12%) The larg- 4.2. Analysis of variance est percentage(57.5%)was between the ages of 20-29 years. Most respondents were experienced online shoppers(85.8%), while al- o investigate whether H1:(e-marketing terms are important most half(49.6%)were very experienced Internet users. More than in influencing shoppers decisions) is supported for the sample

3.2. Sampling An experimental survey is conducted to validate the proposed framework of shoppers’ decision-making styles. According to GVU’s WWW User Surveys, 62.4% of Internet users have attended or graduated from college. It also reveals that people of 21–25 years old are strongly represented among Internet users. Therefore, university students are selected as our survey subjects. The ques￾tionnaires were distributed to 124 undergraduate and graduate students. All respondents volunteered to complete the questionnaires. 3.3. Constructs and questions Ten participants were invited to join a focus group discussion for a pilot study. Based on the results, the research framework of shoppers’ decision-making styles was acquired and tested by six constructs. Where possible, the constructs were based on the liter￾ature on marketing and e-marketing frameworks. Original ques￾tions were based on the information system and electronic commercial texts. Questions pertaining to the constructs: (1) place, (2) computer-mediated electronic commerce, (3) product, (4) pro￾motion, (5) price and (6) process, were measured on a five-point interval scale from very important (1) to unimportant (5). 3.4. Statistical analysis method The SPSS software package is used for data analysis in the study. The data analysis methods are stated in the followings. 3.4.1. Descriptive statistics This experimental survey was conducted among university stu￾dents in Manchester, UK over one month. Results of this experi￾ment likely reflect the student population. Participants are predominantly students (82%) and university staff (12%). The larg￾est percentage (57.5%) was between the ages of 20–29 years. Most respondents were experienced online shoppers (85.8%), while al￾most half (49.6%) were very experienced Internet users. More than half (57.5%) of participants were familiar with web site develop￾ment, while 36.2% were familiar with the literature on marketing strategies. Respondents were asked to express their decision-mak￾ing styles. The higher frequency of decision-making styles for both general online shopping and online book shopping were the value shopper (29.1% and 35.4%, respectively) and the time-sensitive shopper (23.6% and 23.6%). 3.4.2. Analysis of variance To understand the perceived importance of e-marketing strate￾gies on shoppers’ decision-making behavior, t-test and Pearson correlation were applied to perform analysis of variance and data relationship. 4. Results 4.1. Reliability analysis To ascertain the degree of reliability of the e-marketing research framework, Cronbach’s Alpha Coefficient (Cronbach, 1951; Howitt & Cramer, 1999; Katerattanakul & Siau, 1999) is tested. There are debates about what is an acceptable level of reliability (Negnevit￾sky, 2002), but it is accepted that the value of approximately 0.7 or above in order to proceed with any further tests. Results indicate levels of shoppers’ perceived importance, measured by three sets of e-marketing terms: (1) the total 44 e-marketing terms (2) the 22 e-marketing terms for general online shopping, and (3) the 22 e-marketing terms for online book shopping. The results show that the standardized items’ alpha, are 0.8883 for the combined study of general online shopping and online book shopping, 0.8011 for general online shopping, and 0.8042 for online book shopping. This demonstrates high instrument-reliability with a high internal con￾sistency; therefore further tests are carried out. 4.2. Analysis of variance To investigate whether H1: (e-marketing terms are important in influencing shoppers’ decisions) is supported for the sample Table 6 Proposed important e-marketing terms shoppers will perceive. General online shopping Online book shopping (1) the availability of preferred product (1) the availability of preferred book (2) web sites are written in preferred language (2) book stores that are written in preferred language (3) web sites are served in specific country (3) book stores where servers are located as local web sites (4) full contact information (4) full contact information (5) the availability of physical shop, i.e. a physical shop is available in the high street (5) the availability of physical shop, i.e. a physical shop is available in the high street (6) there are good assortments of products, i.e. listed items are plentiful (6) there are over three categories of books available (7) the availability of privacy policy (7) the availability of privacy policy (8) sales are available (8) sales are available (9) free shipping is available (9) free shipping is available (10) special items are listed or highlighted (10) special books are listed or highlighted (11) new products are listed or highlighted (11) new books are listed or highlighted (12) recommended products are listed or highlighted (12) recommended books are listed or highlighted (13) online payment option is available, i.e. shoppers can pay for the purchase(s) by simply giving their bank cards’ details (13) online payment option is available, i.e. shoppers can pay for the purchase(s) by simply giving their bank cards’ details (14) Order checking facility is available (14) Order checking facility is available (15) web sites’ interface is user friendly (15) book shops’ interface design is user friendly, i.e. it is easy to find information on books (16) customer service section is available (16) shopper service section is available (17) if it is not a local web site, international shipping is available (17) if it is not a local web site, international shipping is available (18) gift service is available (18) gift service is available (19) the availability of frequently asked questions section (FAQ) (19) the availability of frequently asked questions section (FAQ) (20) the availability of return policy (20) the availability of return policy (21) the availability of shopper review (21) the availability of shopper review (22) the availability of product picture (22) the availability of book picture W.-S. Lin et al. / Expert Systems with Applications 37 (2010) 6874–6884 6879

W.-S. Lin et al/ Expert Systems with Applications 37(2010)68 of%) is selected to test. It is generally agreed in most discussions or >3) and confidence level less than 90%(P>0.5 mean around roups, the significantly important level of 0.05(confidence level (mean=2.96, p=0.69)are of average importance( ts (26)(Wanninger, 1998 )to include th e standard popula tion mean for medium size samples. A t-test is conducted to eval- late the validity of each e-marketing term in influencing shoppers' 4.3.2. The case of online book buy decisions when choosing an online shop at the confidence level of In online book shopping, 18 e-marketing terms are perceived as 90%. If the mean of importance regarding an e-marketing term is significantly important by shoppers(mean <3)(p <0.05).Among less than 3 out of 5 and its significance level (p value)is less than them, the most important are "the availability of preferred book" 0.05, then the hypothesis is true. The analyses of the two purchase (mean=1.42, p=.000). the availability of online payment option" ases are presented below. These analyses test the level of impor-(mean=1.67. p=.000). the availability of order checking feature tance of each e-marketing term for shoppers when making online (mean=1.69, p=.000)andthe availability of return policy' buying decisions (mean=1.69.p=.000 The less important e-marketing strategies are book stores 4.3. Descriptive statistics which are served local web site"(mean=2.92, p=, 423), availabil- ty of physical bookshop"(mean=3. 16, p=0 145)."gift service In the dimension of e-marketing terms, the mean values are be(mean =3.29, p=0.08), "and"frequently asked question section tween 1.42 and 3. 29. Table 7 shows that both general online buy. (mean=2.88,p=0. 224)" In summary, (1) these 22 e-marketing terms are reliable with ng and online book purchasing reveal that certain e-marketing high internal consistency(alpha value is over 0.7): and (2)they terms are perceived as important in buying decisions. Detailed dis- cussions are presented below. are important in shoppers' decisions. Therefore, HI is accepted. 4.3.1. The case of general online shopping 4.4. Evaluations results of shoppers' decision-making behavior General online shoppers make online purchases for everyday occasions. This part of experiment aims at testing a shopper's per- 4.4.1. Descriptive statistics of shoppers' decision-making behavior ception of the importance of buying something online. It also tests Each style of shopper is selected among participants to deter- the applicability of the framework of e-marketing strategies mine decision-making style. For both general online and online they apply to online shopping. book buying, the value shopper(29.1% and 35.4%) and the time- Results show that 17 e-marketing terms are significant for sensitive shopper(23.6% and 23.6%) have the highest frequency. shoppers(mean <3)(p<0.05). Among these 17 e-marketing terms. For the purchase of tangible products such as books online, 6% the most important are "the availability of product pictures" more shoppers are concerned with value. This may be because 1.52, p=.000).the availability of preferred product" books are tangible and the expected product quality from various 1.58, p=.000),"full contact information"(mean= 1.72, shops is unchanged. Another factor possibly affecting the results and"availability of order checking facility"(mean=1.72, would be the prevalence of university students in the study. The p=000 styles of the value, time-sensitive and security-concerned shoppers In contrast, "the availability of a physical shop"(mean=3. 18, are prevalent among all participants both for general online and p=0.111),recommended products are listed or highlighted" online book shopping. This may be because students are generally (mean=3.00. p= 1.000), availability of gift service"(mean=3.2( eir monetary issues due to the fact of their p= 14), good assortments of products"(mean=2.90, p=0.316). limited incomes. However, it does illustrate the important influ- d availability of frequently asked question section" ences of time and Internet security on shoppers. Table 7 The mean and the significance level of relative importance of the e-marketing strategies. E-marketing strategy E-marketing terms 22 e-marketing terms- the availability of Mean(N= 124) Sig (p value) Mean(N= 124) Sig (p value) Place Customer service section For a non national web site, international shipping Gift servic 224 Web site written in preferred language 2.27 te is served in a specific country, ie. a local web site Full contact information Privacy CMEC nent of products 2.26 194 99 Process Online pay Order checking facility 1.69 Significantly important

groups, the significantly important level of 0.05 (confidence level of 90%) is selected to test. It is generally agreed in most discussions of t-tests (26) (Wanninger, 1998) to include the standard popula￾tion mean for medium size samples. A t-test is conducted to eval￾uate the validity of each e-marketing term in influencing shoppers’ decisions when choosing an online shop at the confidence level of 90%. If the mean of importance regarding an e-marketing term is less than 3 out of 5 and its significance level (p value) is less than 0.05, then the hypothesis is true. The analyses of the two purchase cases are presented below. These analyses test the level of impor￾tance of each e-marketing term for shoppers when making online buying decisions. 4.3. Descriptive statistics In the dimension of e-marketing terms, the mean values are be￾tween 1.42 and 3.29. Table 7 shows that both general online buy￾ing and online book purchasing reveal that certain e-marketing terms are perceived as important in buying decisions. Detailed dis￾cussions are presented below. 4.3.1. The case of general online shopping General online shoppers make online purchases for everyday occasions. This part of experiment aims at testing a shopper’s per￾ception of the importance of buying something online. It also tests the applicability of the framework of e-marketing strategies as they apply to online shopping. Results show that 17 e-marketing terms are significant for shoppers (mean 3) and confidence level less than 90% (P > 0.05). 4.3.2. The case of online book buy In online book shopping, 18 e-marketing terms are perceived as significantly important by shoppers (mean <3) (p < 0.05). Among them, the most important are ‘‘the availability of preferred book” (mean = 1.42, p = .000), ‘‘the availability of online payment option” (mean = 1.67, p = .000), ‘‘the availability of order checking feature” (mean = 1.69, p = .000) and ‘‘the availability of return policy” (mean = 1.69, p = .000). The less important e-marketing strategies are ‘‘book stores which are served local web site” (mean = 2.92, p = .423), ‘‘availabil￾ity of physical bookshop” (mean = 3.16, p = 0.145), ‘‘gift service” (mean = 3.29, p = 0.08),” and ‘‘frequently asked question section” (mean = 2.88, p = 0.224).” In summary, (1) these 22 e-marketing terms are reliable with high internal consistency (alpha value is over 0.7); and (2) they are important in shoppers’ decisions. Therefore, H1 is accepted. 4.4. Evaluations results of shoppers’ decision-making behavior 4.4.1. Descriptive statistics of shoppers’ decision-making behavior Each style of shopper is selected among participants to deter￾mine decision-making style. For both general online and online book buying, the value shopper (29.1% and 35.4%) and the time￾sensitive shopper (23.6% and 23.6%) have the highest frequency. For the purchase of tangible products such as books online, 6% more shoppers are concerned with value. This may be because books are tangible and the expected product quality from various shops is unchanged. Another factor possibly affecting the results would be the prevalence of university students in the study. The styles of the value, time-sensitive and security-concerned shoppers are prevalent among all participants both for general online and online book shopping. This may be because students are generally more concerned with their monetary issues due to the fact of their limited incomes. However, it does illustrate the important influ￾ences of time and Internet security on shoppers. Table 7 The mean and the significance level of relative importance of the e-marketing strategies. E-marketing strategy E-marketing terms General buy Online book buy 22 e-marketing terms – the availability of: Mean (N = 124) Sig. (p value) Mean (N = 124) Sig. (p value) Place Customer service section 1.73* .000 1.88* .000 For a non national web site, international shipping 1.94* .000 2.03* .000 Gift service 3.26 0.14 3.29* .008 Frequently asked questions (FAQ) 2.96 .690 2.88 .224 Return policy 1.65* .000 1.69* .000 Web site written in preferred language 2.27* .000 2.17* .000 Web site is served in a specific country, i.e. a local web site 2.77* .035 2.92 .423 Full contact information 1.72* .000 1.74* .000 Physical shop 3.18 0.111 3.16 .145 Privacy policy 1.97* .000 2.05* .000 CMEC User friendly interface 1.76* .000 1.82* .000 Product Product picture 1.52* .000 2.21* .000 Preferred product 1.58* .000 1.42* .000 Good assortment of products 2.90 0.316 2.50* .000 Special items are listed or highlighted 2.78* .000 2.33* .000 New products are listed or highlighted 2.81* .000 2.26* .000 Recommended products are listed or highlighted 3.00 1.000 2.55* .000 Shopper reviews 2.06* .000 1.94* .000 Promotion Sales 1.91* .016 2.09* .000 Free shipping 1.99* .042 1.89* .000 Process Online payment option 1.73* .000 1.67* .000 Order checking facility 1.72* .000 1.69* .000 * Significantly important. 6880 W.-S. Lin et al. / Expert Systems with Applications 37 (2010) 6874–6884

W-S Lin et aL/ Expert Systems with Applications 37(2010)6874-6884 6881 4.4.2. Analysis of variance payment option is available, ""(14)order checking facility is avail- To investigate H2: (shoppers behave differently with respect to able. "These terms reflect the differences among these six styles of different types of products), one-way ANOVA(Analysis of Vari- shoppers, proving that they are useful in identifying shoppe ance)and cross-tab analysis techniques are conducted. ANOVA is information needs. d to test the six types of decision makers based on shoppers perceptions of importance derived from a confidence interval of 4.5. Cross-analysis between general online shopping and online book 90%. The descriptive summary of general online shopping is shopping presented in Table 8. The largest group consists of value shop- pers(N=37). while fashionable shoppers make up the smallest Paired t-test is applied in analyzing whole samples of two types of purchases, general online shopping and online book shopping. in Results are shown in Table 8 with the mean plots of importance order to compare the perceptions of importance of e-marketing for the six decision types. Among the six groups, eight e-marketing terms. The significant results(Sig. <0. 1)are listed at 90% confidence terms, (2) web site that is written in preferred language "(3)web level ervice in specific country, ""(4)full contact informatior (8)sales are available, (10)special items are listed or higl 1. E-marketing terms of pair (1)"the availability of preferred hted (18)gift service is available, ""(19)the availability of roduct(mean=1.57)versus the availability of preferred book FAQ, "and"(22)the availability of product picture"are perceived (mean=1.43)"(t=2.40, p <0. 1), shows it has significantly dif as significantly important. These six styles are useful in identifying ferent relationship between shoppers' perceptions of impor hoppers'information needs(as presented in Table 9 tance in general online and online book shopping. In this case, The same one-way ANOVA is applied to these subgroups in or- shoppers think"the availability of preferred book"is more er to see if they affected the results at the confidence interval of mportant than"the availability 90% for online book buying. A descriptive summary of different 2. E-marketing terms of pair(6)good assortments of products styles of shoppers for online book buying is presented in Table 10 (mean=2.88)versus more than three categories available in Value shoppers still represent the largest group(N= 45)with the the bookshop(mean=2.50)"(t=3.316, p <0. 1)shows it has sig time-sensitive shopper the second largest group(N= 30). There nificantly different relationship between the shoppers perce are some shifts in shoppers'styles as compared to general online tions of importance in general online shopping and online book shopping. In this case, shoppers think"more than three with the means of im categories available in the bookshop"is more important than roducts icantly important: " (1) the availability of preferred product, ""(2) 3. E-marketing terms of pair( 8)the availability of sales, dis web site that is written in preferred language, ""(5)physical shop counts, or special offers(mean= 1.90) versus the availability Table 8 online shopping Users' decision-making style for General online shopping Users' decision-making style for General online shopping Table g Significantly important e-marketing terms perceived by six styles of shoppers for general online shopping. b D1(N=16)D2(N=24)D3(N=37)D4(N=5)D5(N=30)D6(N=12)sig (2)Web site that is written in preferred language 3)Web site that is located in specific countries 4) Full contact information 2.33 (10)Special item 2.17 (18)Gift service is available (19) the availability of frequently asked question section(FAQ) 2.81 (22)the availability of product picture 1.13 1.67 1.50 <0.1 The perceptions of importance were measured on a 5-point scale ranging from (1)very important to(5)unimportant. b D1= general-purpose shopper, D2= security-concerned shopper, D3= value shopper, D4=fashionable shopper, D5= time-sensitive shopper, D6=service-oriented

4.4.2. Analysis of variance To investigate H2: (shoppers behave differently with respect to different types of products), one-way ANOVA (Analysis of Vari￾ance) and cross-tab analysis techniques are conducted. ANOVA is used to test the six types of decision makers based on shoppers’ perceptions of importance derived from a confidence interval of 90%. The descriptive summary of general online shopping is presented in Table 8. The largest group consists of value shop￾pers (N = 37), while fashionable shoppers make up the smallest (N = 5). Results are shown in Table 8 with the mean plots of importance for the six decision types. Among the six groups, eight e-marketing terms, ‘‘(2) web site that is written in preferred language,” ‘‘(3) web sites are service in specific country,” ‘‘(4) full contact information,” ‘‘(8) sales are available,” ‘‘(10) special items are listed or high￾lighted,” ‘‘(18) gift service is available,” ‘‘(19) the availability of FAQ,” and ‘‘(22) the availability of product picture‘‘ are perceived as significantly important. These six styles are useful in identifying shoppers’ information needs (as presented in Table 9). The same one-way ANOVA is applied to these subgroups in or￾der to see if they affected the results at the confidence interval of 90% for online book buying. A descriptive summary of different styles of shoppers for online book buying is presented in Table 10. Value shoppers still represent the largest group (N = 45) with the time-sensitive shopper the second largest group (N = 30). There are some shifts in shoppers’ styles as compared to general online buying. Results are illustrated in Table 11, with the means of impor￾tance for the six decision types. Eight terms are perceived as signif￾icantly important: ‘‘(1) the availability of preferred product,” ‘‘(2) web site that is written in preferred language,” ‘‘(5) physical shop is available,” ‘‘(6) there are over three categories of books,” ‘‘ (7) the availability of privacy policy,” ‘‘(8) sales are available,” ‘‘(13) online payment option is available,” ‘‘(14) order checking facility is avail￾able.” These terms reflect the differences among these six styles of shoppers, proving that they are useful in identifying shoppers’ information needs. 4.5. Cross-analysis between general online shopping and online book shopping Paired t-test is applied in analyzing whole samples of two types of purchases, general online shopping and online book shopping, in order to compare the perceptions of importance of e-marketing terms. The significant results (Sig. <0.1) are listed at 90% confidence level: 1. E-marketing terms of pair (1) ‘‘the availability of preferred product (mean = 1.57) versus the availability of preferred book (mean = 1.43)” (t = 2.40, p <0.1), shows it has significantly dif￾ferent relationship between shoppers’ perceptions of impor￾tance in general online and online book shopping. In this case, shoppers think ‘‘the availability of preferred book” is more important than ‘‘the availability of preferred product.” 2. E-marketing terms of pair (6) ‘‘good assortments of products (mean = 2.88) versus more than three categories available in the bookshop (mean = 2.50)” (t = 3.316, p <0.1) shows it has sig￾nificantly different relationship between the shoppers’ percep￾tions of importance in general online shopping and online book shopping. In this case, shoppers think ‘‘more than three categories available in the bookshop” is more important than ‘‘good assortments of products.” 3. E-marketing terms of pair (8) ‘‘the availability of sales, dis￾counts, or special offers (mean = 1.90) versus the availability of sales, discounts, or special offers in bookshop (mean = 2.10)” (t = 2.557, p <0.1), shows it has significantly different relation￾Table 8 Descriptive summary of different styles of shoppers. Different types of shopping for general online shopping Users' decision-making style for General online shopping Users' decision-making style for General online shopping Service-oriented sho Time-sensitive shopp Fashionable shopper Value shopper Security-concerned s General-purpose shop Frequency 40 30 20 10 0 Table 9 Significantly important e-marketing terms perceived by six styles of shoppers for general online shopping.a,b D1 (N = 16) D2 (N = 24) D3 (N = 37) D4 (N = 5) D5 (N = 30) D6 (N = 12) Sig. (2) Web site that is written in preferred language 2.31 2.96 2.27 1.40 1.97 2.33 <0.1 (3) Web site that is located in specific countries 2.63 3.33 2.43 2.40 2.80 3.17 <0.1 (4) Full contact information 1.31 1.83 2.03 1.00 1.70 1.75 <0.1 (8) Sales are available 2.06 2.33 1.54 1.60 2.10 1.92 <0.1 (10) Special items are listed or highlighted 2.50 3.13 2.84 2.80 3.07 2.17 <0.1 (18) Gift service is available 3.38 3.46 3.68 2.00 3.03 2.92 <0.1 (19) the availability of frequently asked question section(FAQ) 2.81 3.04 2.97 1.75 3.27 2.50 <0.1 (22) the availability of product picture 1.13 1.67 1.62 1.00 1.50 1.42 <0.1 a The perceptions of importance were measured on a 5-point scale ranging from (1) very important to (5) unimportant. b D1 = general-purpose shopper, D2 = security-concerned shopper, D3 = value shopper, D4 = fashionable shopper, D5 = time-sensitive shopper, D6 = service-oriented shopper. W.-S. Lin et al. / Expert Systems with Applications 37 (2010) 6874–6884 6881

W.-S. Lin et al/ Expert Systems with Applications 37(2010)6874-6884 Descriptive summary of different styles of shoppers. for online book shopping Users' decision-making style for online BOOK shopping Users' decision-making style for online BOOK shopping Table 11 ANOVA between shoppers'perception of importance and six types of types of shopping for the purpose of online book shopping a, b N=19)(N=11)(N=45)(N=2)(N=30) (1) The availability of preferred books D)Physical shop is available, i.e. a physical shop is available in the high street (6) There are over three categories of books 189 2.4 58 87 2.38 227 2.36 150 2.23 (13)online payment option is available, ie shopper can pay for the product by simply 1.79 (14)order checking facility is available 1.63 98 0.1 The perceptions of importance were measured b D1= general-purpose buy, D2= security-concemed buy, D3= value buy, D4= fashionable buy, D5=time-sensitive buy, D6= service-oriented buy ship between the shoppers 'perceptions of importance in gen- books are listed and highlighted in bookshops (mean= 2.56 eral online shopping and online book shopping. In this ca (t=4.947, sig. <0.1), shows it has significantly different relation shoppers think"the availability of sales, discounts, or specia hip between the shoppers perceptions of importance in gen- offers"is more important than"the availability of sales, dis counts, or special offers in bookshop shoppers think"recommended books are listed and highlighted 4. E-marketing terms of pair (9)"free shipping is available in bookshops"is more important than the same e-marketing (mean =2.02)versus the free shipping is available in a book term in general shops. shop(mean=1.90)"(t= 1.771, p <0. 1), shows it has significantly 8. E-marketing terms of pair (22)"pictures of products are avail- different relationship between the shoppers perceptions of able(mean= 1.52) versus pictures of books are available mportance in general online shopping and online book shop- (mean=2. 23)"(t=-8059, sig. <0. 1), shows it has significantly ping. In this case, shoppers think"free shipping is available in different relationship between the shoppers' perceptions of a bookshop"is more important than in general shops. importance in general online shopping and online book shop- 5. E-marketing terms of pair (10)"special items are listed and ping. In this case, shoppers think"pictures of products are avail- highlighted in general shops(mean=2.79)versus special books able"is more important than the same e-marketing term in are listed and highlighted in bookshops (mean=2.34) (t= 4.695, sig. <0.1), shows it has significantly different relation ship between the shoppers' perceptions of importance in gen- Results indicate that shoppers perceptions of importance eral online shopping and online book shopping. In this case, regarding e-marketing terms are significantly different in the two shoppers think" special books are listed and highlighted in cases. Shoppers view the two e-marketing terms of"the availabil- bookshops"is more important than the same e-marketing term ity of sales, discounts, or special offers"and"pictures of products in general shops are available"to be more important whilst shopping in general 6. E-marketing terms of pair (11)"now products are listed and shops. This may be due to the fact that online bookshops appeal highlighted (mean=2.82)versus new books are listed and high- to a relatively mature market (ie. bookshops are quite popular lighted in bookshops(mean= 2. 25)"(t=6. 275, p<0. 1), shows it and a large number of shoppers have shopped for books online). has significantly different relationship between the shoppers With so many online bookshops, the price competition is fierc perceptions of importance in general online shopping and Therefore, the e-marketing term ofthe availability of sales, dis- online book shopping. In this case, shoppers think"new books counts, or special offers"in bookshops is perceived to be less are listed and highlighted in bookshops"is more important than important than in general shops. As books are tangible their qual- the same e-marketing term in general shops. 7. E-marketing terms of pair(12)"recommended products are know exactly which books they are looking for are not concerned listed and highlighted (mean=3.01)versus recommended with the availability of pictures. These reasons explain what may

ship between the shoppers’ perceptions of importance in gen￾eral online shopping and online book shopping. In this case, shoppers think ‘‘the availability of sales, discounts, or special offers” is more important than ‘‘the availability of sales, dis￾counts, or special offers in bookshops.” 4. E-marketing terms of pair (9) ‘‘free shipping is available (mean = 2.02) versus the free shipping is available in a book￾shop (mean = 1.90)” (t = 1.771, p <0.1), shows it has significantly different relationship between the shoppers’ perceptions of importance in general online shopping and online book shop￾ping. In this case, shoppers think ‘‘free shipping is available in a bookshop” is more important than in general shops. 5. E-marketing terms of pair (10) ‘‘special items are listed and highlighted in general shops (mean = 2.79) versus special books are listed and highlighted in bookshops (mean = 2.34)” (t = 4.695, sig. <0.1), shows it has significantly different relation￾ship between the shoppers’ perceptions of importance in gen￾eral online shopping and online book shopping. In this case, shoppers think ‘‘special books are listed and highlighted in bookshops” is more important than the same e-marketing term in general shops. 6. E-marketing terms of pair (11) ‘‘now products are listed and highlighted (mean = 2.82) versus new books are listed and high￾lighted in bookshops (mean = 2.25)” (t = 6.275, p <0.1), shows it has significantly different relationship between the shoppers’ perceptions of importance in general online shopping and online book shopping. In this case, shoppers think ‘‘new books are listed and highlighted in bookshops” is more important than the same e-marketing term in general shops. 7. E-marketing terms of pair (12) ‘‘recommended products are listed and highlighted (mean = 3.01) versus recommended books are listed and highlighted in bookshops (mean = 2.56)” (t = 4.947, sig. <0.1), shows it has significantly different relation￾ship between the shoppers’ perceptions of importance in gen￾eral online shopping and online book shopping. In this case, shoppers think ‘‘recommended books are listed and highlighted in bookshops” is more important than the same e-marketing term in general shops. 8. E-marketing terms of pair (22) ‘‘pictures of products are avail￾able (mean = 1.52) versus pictures of books are available (mean = 2.23)” (t = 8.059, sig. <0.1), shows it has significantly different relationship between the shoppers’ perceptions of importance in general online shopping and online book shop￾ping. In this case, shoppers think ‘‘pictures of products are avail￾able” is more important than the same e-marketing term in online bookshops. Results indicate that shoppers’ perceptions of importance regarding e-marketing terms are significantly different in the two cases. Shoppers view the two e-marketing terms of ‘‘the availabil￾ity of sales, discounts, or special offers” and ‘‘pictures of products are available” to be more important whilst shopping in general shops. This may be due to the fact that online bookshops appeal to a relatively mature market (i.e. bookshops are quite popular and a large number of shoppers have shopped for books online). With so many online bookshops, the price competition is fierce. Therefore, the e-marketing term of ‘‘the availability of sales, dis￾counts, or special offers” in bookshops is perceived to be less important than in general shops. As books are tangible, their qual￾ities can be judged by their product descriptions. Shoppers who know exactly which books they are looking for are not concerned with the availability of pictures. These reasons explain what may Table 10 Descriptive summary of different styles of shoppers. for online book shopping Users' decision-making style for online BOOK shopping Users' decision-making style for online BOOK shopping Service-oriented sho Time-sensitive shopp Fashionable shopper Value shopper Security-concerned s General-purpose shop Frequency 50 40 30 20 10 0 Table 11 ANOVA between shoppers’ perception of importance and six types of types of shopping for the purpose of online book shopping.a,b D1 (N = 19) D2 (N = 11) D3 (N = 45) D4 (N = 2) D5 (N = 30) D6 (N = 17) Sig. (1) The availability of preferred books 1.58 1.27 1.56 2.00 1.23 1.24 <0.1 (2) Book stores are written in preferred language 2.00 2.82 2.36 2.50 1.77 2.29 <0.1 (5) Physical shop is available, i.e. a physical shop is available in the high street 2.63 3.27 3.13 3.00 3.67 2.75 <0.1 (6) There are over three categories of books 1.89 2.45 2.58 2.00 2.87 2.38 <0.1 (7) The availability of privacy policy 1.84 1.27 2.27 3.00 2.03 2.06 <0.1 (8) Sales are available 2.37 2.36 1.70 1.50 2.23 2.38 <0.1 (13) Online payment option is available, i.e. shopper can pay for the product by simply giving his bank card’s details 1.79 1.36 1.91 2.00 1.30 1.75 <0.1 (14) Order checking facility is available 1.63 1.27 1.98 1.50 1.50 1.67 <0.1 a The perceptions of importance were measured on a 5-point scale ranging from (1) very important to (5) unimportant. b D1 = general-purpose buy, D2 = security-concerned buy, D3 = value buy, D4 = fashionable buy, D5 = time-sensitive buy, D6 = service-oriented buy. 6882 W.-S. Lin et al. / Expert Systems with Applications 37 (2010) 6874–6884

W-S Lin et aL/ Expert Systems with Applications 37(2010)6874-6884 cause the differences between general online shopping and online This view is acknowledged by this study to ensure that(1)there book shopping. Therefore, H2 is accepted is a thorough investigation of commercial web sites; (2)e-market ing strategies can be translated into the knowledge of current state of online contextual environment based on the current structure of 5. Discussion and conclusion e-marketing strategies; (3)the structure of e-marketing strategie These 44 e-marketing terms have been tested with the stan can be used to provide a clear structure for adding new terms through constant amendments: (4)intelligent shopping support dards of reliability The results yield a high internal consistency can adapt the knowledge derived from the e-marketing strategies in application to this user experiment (alpha value is over 0. 7 for dynamically to complete search tasks among the commercial total 44 e-marketing terms, 22 terms for general online shopping and 22 terms of online book shopping, respectively ). Therefore, it web site environmen is valid to use this questionnaire to process other evaluative tests nd conclude the research findings. The results for the two re- 5.2. Implications for practitioners search hypotheses are: The results for both cases of general online shopping and online (1)H1:(e-marketing terms are important in influencing shop- book buy show that shoppers perceive the availability of the de- pers'decisions to choose an online shop) is accepted on 17 sired product (e.g. the preferred book)as the most important e- e-marketing terms and 18 e-marketing terms for the two marketing term. In addition, web sites that provide full contad cases of general online shopping and online book shopping. information gain the trust of shoppers. The similar concepts as respectively. To summarize the results for these two case the availability of a product picture, the availability of an order checking facility, an online payment option, product pictures, and ferred product/book, ""(14)order checking facility is avail- a returns policy all possibly have influences on motivating shop- able"are both perceived as very important, among the top pers to buy online. These findings reveal that the critical element marketing terms, while the e-marketing terms in maintaining a good relationship and clear communication be- of"(5) the availability of physical shop. "(18)gift service tween online ps and their customers, the online shoppers in is available, ""(19) the availability of frequently asked ques- EC is making, accepting, and keeping promises by providing full tion section"are all perceived as unimportant. These three information about the online shop and its products, by delivering insignificant e-marketing terms should be marked as such the right product to shoppers, and by offering good post-purchase in the hierarchy of e-marketing terms. For the six types of service, including a clear returns policy buys, results indicate that under different buying scenarios. shoppers perceive e-marketing terms in terms of their 5.3. Limitations of this study and implications for future research importance differently. The e-marketing strategies of (1) place, (2) computer-mediated electronic commerce, 3) Like other empirical research, this study has limitations. First, product, and (4)process are empirically reviewed as statisti- this study focuses on two buying cases: general online shopping cally significant with different types of buys. Eight e-market- and online book buying, so attempts to generalize the results to shoppers:"(7)the availability of privacy pbncy, ses of other products should be done cautiously. Second, results obtained from student subjects per service section is available, "(20)the availability of dents. For other groups of online shoppers, their perceived impor return policy, "(15 web sites' interface is user friendly, nce of e-marketing term under different buy need to be (1)the availability of preferred product,"(22)the avail- verified. Future studies can extend the results of this study by ability of product picture, ""(13)online payment option is investigating other product classes and other groups of consumers available, "and"(14)order checking facility is available (2)H2:(shoppers behave differently with respect to different plied to program-intelligent shopping supports and evaluate their interval for both general online shopping and online In the future, the framework of e-marketing terms needs to be buy. The six types of buys that are presumed by this updated in response to changes in the online market. By doing re reviewed as empirically signific ant in revealing so, it can enhance the online shopping supports and extend it to he differences in buys. sustain sophisticated online web sites. Ongoing efforts can be made to cover higher scalability (e.g. different languages )or differ 5.1. Implications for the academic ent buys, (e.g. online travel shops, online digital camera shops) Online marketers, knowledge engineers, or shoppers can add This research is aimed at bridging the gap between shopping new e-marketing terms. other groups of shoppers can be invited dies of shoppers'decision-making behavior, con- to further verify the decision trees and rules (e.g. industrial practi- tributing to transforming a shopper's knowledge of search and pr tioners, e-marketing marketers). In keeping with the rapidly evolv viding shopping support. By applying the framework of ing e-business, these enhancements and extensions will be crucial marketing strategies, several e-marketing terms revealed their in improving the effectiveness of online information supports. In ortance in affecting shoppers online buying process. Six types of buys are modeled. The framework of e-marketing strategies that line shop s support and evaluate its ability to provide shoppers interact with different types of buys is constructed and tested. It search results reveals that shoppers view a particular set of e-marketing terms as important while searching for online information under a differ- References ent buying scenario. Furthermore, the results of this research can upport developers in expanding intelligent shopping supports to Anupam, V, assist not only in search, but also in negotiation and communica- tion by applying the knowledge of human search and selling ame.html>(a techniques Baker, M.]. (1999) The marketing book. Woburn, MA: Butterworth-Heinemann

cause the differences between general online shopping and online book shopping. Therefore, H2 is accepted. 5. Discussion and conclusion These 44 e-marketing terms have been tested with the stan￾dards of reliability. The results yield a high internal consistency in application to this user experiment (alpha value is over 0.7 for total 44 e-marketing terms, 22 terms for general online shopping and 22 terms of online book shopping, respectively). Therefore, it is valid to use this questionnaire to process other evaluative tests and conclude the research findings. The results for the two re￾search hypotheses are: (1) H1: (e-marketing terms are important in influencing shop￾pers’ decisions to choose an online shop) is accepted on 17 e-marketing terms and 18 e-marketing terms for the two cases of general online shopping and online book shopping, respectively. To summarize the results for these two case studies, e-marketing terms of ‘‘(1) the availability of pre￾ferred product/book,” ‘‘(14) order checking facility is avail￾able” are both perceived as very important, among the top three key e-marketing terms, while the e-marketing terms of ‘‘(5) the availability of physical shop,” ‘‘(18) gift service is available,” ‘‘(19) the availability of frequently asked ques￾tion section” are all perceived as unimportant. These three insignificant e-marketing terms should be marked as such in the hierarchy of e-marketing terms. For the six types of buys, results indicate that under different buying scenarios, shoppers perceive e-marketing terms in terms of their importance differently. The e-marketing strategies of (1) place, (2) computer-mediated electronic commerce, (3) product, and (4) process are empirically reviewed as statisti￾cally significant with different types of buys. Eight e-market￾ing terms are perceived as essential for all styles of shoppers: ‘‘(7) the availability of privacy policy,” ‘‘(16) shop￾per service section is available,” ‘‘(20) the availability of return policy,” ‘‘(15) web sites’ interface is user friendly,” ‘‘(1) the availability of preferred product,” ‘‘(22) the avail￾ability of product picture,” ‘‘(13) online payment option is available,” and ‘‘(14) order checking facility is available.” (2) H2: (shoppers behave differently with respect to different types of products) is accepted based on ANOVA of 90% con- fidence interval for both general online shopping and online book buy. The six types of buys that are presumed by this study are reviewed as empirically significant in revealing the differences in buys. 5.1. Implications for the academic This research is aimed at bridging the gap between shopping support and studies of shoppers’ decision-making behavior, con￾tributing to transforming a shopper’s knowledge of search and pro￾viding shopping support. By applying the framework of e￾marketing strategies, several e-marketing terms revealed their importance in affecting shoppers’ online buying process. Six types of buys are modeled. The framework of e-marketing strategies that interact with different types of buys is constructed and tested. It reveals that shoppers view a particular set of e-marketing terms as important while searching for online information under a differ￾ent buying scenario. Furthermore, the results of this research can support developers in expanding intelligent shopping supports to assist not only in search, but also in negotiation and communica￾tion by applying the knowledge of human search and selling techniques. This view is acknowledged by this study to ensure that (1) there is a thorough investigation of commercial web sites; (2) e-market￾ing strategies can be translated into the knowledge of current state of online contextual environment based on the current structure of e-marketing strategies; (3) the structure of e-marketing strategies can be used to provide a clear structure for adding new terms through constant amendments; (4) intelligent shopping support can adapt the knowledge derived from the e-marketing strategies dynamically to complete search tasks among the commercial web site environment.. 5.2. Implications for practitioners The results for both cases of general online shopping and online book buy show that shoppers perceive the availability of the de￾sired product (e.g. the preferred book) as the most important e￾marketing term. In addition, web sites that provide full contact information gain the trust of shoppers. The similar concepts as the availability of a product picture, the availability of an order checking facility, an online payment option, product pictures, and a returns policy all possibly have influences on motivating shop￾pers to buy online. These findings reveal that the critical element in maintaining a good relationship and clear communication be￾tween online shops and their customers, the online shoppers in EC is making, accepting, and keeping promises by providing full information about the online shop and its products, by delivering the right product to shoppers, and by offering good post-purchase service, including a clear returns policy. 5.3. Limitations of this study and implications for future research Like other empirical research, this study has limitations. First, this study focuses on two buying cases: general online shopping and online book buying, so attempts to generalize the results to other products should be done cautiously. Second, results obtained from student subjects may only represent the population of stu￾dents. For other groups of online shoppers, their perceived impor￾tance of e-marketing term under different buys may need to be verified. Future studies can extend the results of this study by investigating other product classes and other groups of consumers. In addition, the proposed human knowledge of search can by ap￾plied to program-intelligent shopping supports and evaluate their search effectiveness. In the future, the framework of e-marketing terms needs to be updated in response to changes in the online market. By doing so, it can enhance the online shopping supports and extend it to sustain sophisticated online web sites. Ongoing efforts can be made to cover higher scalability (e.g. different languages) or differ￾ent buys, (e.g. online travel shops, online digital camera shops). Online marketers, knowledge engineers, or shoppers can add new e-marketing terms. Other groups of shoppers can be invited to further verify the decision trees and rules (e.g. industrial practi￾tioners, e-marketing marketers). In keeping with the rapidly evolv￾ing e-business, these enhancements and extensions will be crucial in improving the effectiveness of online information supports. In addition, these contextual rules should be applied to program on￾line shopping support and evaluate its ability to provide shoppers with relevant search results. References Anupam, V., Hull, R., & Kumar, B. (2001). Personalizing e-commerce applications with online heuristic decision-making. In Proceedings of 10th world wide web conference, Hong Kong (pp. 296–307), URL: (accessed 7.4.2009). Baker, M. J. (1999). The marketing book. Woburn, MA: Butterworth–Heinemann. W.-S. Lin et al. / Expert Systems with Applications 37 (2010) 6874–6884 6883