Creating an Ontology for the user profile Method and applications Maria golemati, Akrivi Katifori, Costas Vassilakis, George Lepouras, Constantin Halatsis model user profiles; this standard ontology will facilitate the A User profiling is commonly employed nowadays to communication between applications and serve as reference ity and other user-centrmi to support personalization, i This work presents such an ontology for modeling user hindering thus application interoperability at the user profile profiles. The purpose was to create a general yet extendable level,increasing the amount of work to be done and the ontology that will be able to adapt to the needs of every ossibility of errors or omissions in the profile model. This work application, maintaining at the same time a general common aims at creating a user profile ontology that incorporates structure so as to satisfy portability and communication concepts and properties used to model the user profile Existing between different applications. After a brief overview of literature, applications and ontologies related to the domain of existing work in the area of profiling in relation with user context and profiling have been taken into account in order ontologies, the method for creating the user profile ontology is to create a general, comprehensive and extensible user modeL. This ontology can be used as a reference model, in order to presented, followed by a presentation of the ontology itself. alleviate the aforementioned issues. The model, available for Examples of the application of the model in two domains are download, is exemplified through its application in two different provided in the following section. The last section presents the areas, personal information management and adaptive conclusions and briefly outlines future work visualization IL RELATED WORK Index terms--user profile ontology, user modeling, context In the last few years the need for software systems to utomatically adapt to their users has been recognized in many L INTROdUCTION application areas and the research on user profiling and The continuing progress in network technologies and data context has spread into many disciplines which are concerned with the development of computer systems that are to be used mounts of documents.The need for more effective by heterogeneous user populations [8. Context in [131 information retrieval has lead to the creation of the notions of categorized in human user context and surroundings context he semantic web and personalized information management, and may also be categorized according to persistence areas of study that take advantage of the semantic context of (permanent and temporary) and evolution (static and the presented information and the user to facilitate the dynamic). Elaine Rich [15] identifies a three dimensional information storage and retrieval process. The notion of user space of user models: I)canonical vs individual user model profiling has been introduced in order to record the user 2)explicit vS implicit user model and 3)long-term vs short- context and personalize applications so as to be tailored to the rm user mode user needs Another important issue is that a user might be found in Ontologies have been proven an effective means for various contexts. Thus, a context-aware system has to infer modeling digital collections and user context. They can be a which context the user is in a given moment in time, and domain related to a specific area of interest and be used for According to [23], a user model contains all information browsing and query refinement Ontologies model concepts that the system knows about the user. It is generally initialized and relationships in a high level of abstraction, providing rich either with default values or by querying the user. Users in grouped in"stereotypes”, like "woman”or semantics for humans to work with and the require "computer scientist,, according to particular characteristics formalism for computers to perform mechanical processing which are application specific Using an ontology to model the user profile has already An overview of methods for building a user profile are een proposed in various applications like web search 19) presented in [15] and [16]. User mode godellin and guidelines are presented in[8], concentrating on modelling of [12] and personal information management [7]. However, up user knowledge, plans, and preferences in a domain. It focuses to this point, ontologies modeling user profles are on stereotype(as opposed to individua) profiles. The need for application-specific a profile that supports reasoning is also stressed out in [15] specifically for a particular domain. Taking into account the The goals listed above can be achieved through the use of continuing incorporation of ontologies in new applications there is an emerging need for a standard ontology that will Ontologies. Ontologies in the form of hierarchies of user
Creating an Ontology for the User Profile: Method and Applications Maria Golemati, Akrivi Katifori, Costas Vassilakis, George Lepouras, Constantin Halatsis Abstract— User profiling is commonly employed nowadays to enhance usability as well as to support personalization, adaptivity and other user-centric features. Insofar, application designers model user profiles mainly in an ad-hoc manner, hindering thus application interoperability at the user profile level, increasing the amount of work to be done and the possibility of errors or omissions in the profile model. This work aims at creating a user profile ontology that incorporates concepts and properties used to model the user profile. Existing literature, applications and ontologies related to the domain of user context and profiling have been taken into account in order to create a general, comprehensive and extensible user model. This ontology can be used as a reference model, in order to alleviate the aforementioned issues. The model, available for download, is exemplified through its application in two different areas, personal information management and adaptive visualization. Index Terms—user profile, ontology, user modeling, context I. INTRODUCTION The continuing progress in network technologies and data storage has enabled the digitization and dissemination of huge amounts of documents. The need for more effective information retrieval has lead to the creation of the notions of the semantic web and personalized information management, areas of study that take advantage of the semantic context of the presented information and the user to facilitate the information storage and retrieval process. The notion of user profiling has been introduced in order to record the user context and personalize applications so as to be tailored to the user needs. Ontologies have been proven an effective means for modeling digital collections and user context. They can be a very useful tool, because they may present an overview of the domain related to a specific area of interest and be used for browsing and query refinement. Ontologies model concepts and relationships in a high level of abstraction, providing rich semantics for humans to work with and the required formalism for computers to perform mechanical processing and reasoning. Using an ontology to model the user profile has already been proposed in various applications like web search [9], [12] and personal information management [7]. However, up to this point, ontologies modeling user profiles are application-specific, with each one having been created specifically for a particular domain. Taking into account the continuing incorporation of ontologies in new applications, there is an emerging need for a standard ontology that will model user profiles; this standard ontology will facilitate the communication between applications and serve as reference point when profiling functionalities need to be developed. This work presents such an ontology for modeling user profiles. The purpose was to create a general yet extendable ontology that will be able to adapt to the needs of every application, maintaining at the same time a general common structure so as to satisfy portability and communication between different applications. After a brief overview of existing work in the area of profiling in relation with ontologies, the method for creating the user profile ontology is presented, followed by a presentation of the ontology itself. Examples of the application of the model in two domains are provided in the following section. The last section presents the conclusions and briefly outlines future work. II. RELATED WORK In the last few years the need for software systems to automatically adapt to their users has been recognized in many application areas and the research on user profiling and context has spread into many disciplines which are concerned with the development of computer systems that are to be used by heterogeneous user populations [8]. Context in [13] is categorized in human user context and surroundings context and may also be categorized according to persistence (permanent and temporary) and evolution (static and dynamic). Elaine Rich [15] identifies a three dimensional space of user models: 1) canonical vs. individual user model, 2) explicit vs. implicit user model and 3) long-term vs. shortterm user model. Another important issue is that a user might be found in various contexts. Thus, a context-aware system has to infer which context the user is in a given moment in time, and consequently adapt the system to that context [14] [15]. According to [23], a user model contains all information that the system knows about the user. It is generally initialized either with default values or by querying the user. Users in some cases are grouped in “stereotypes”, like “woman” or “computer scientist”, according to particular characteristics which are application specific. An overview of methods for building a user profile are presented in [15] and [16]. User modeling issues and guidelines are presented in [8], concentrating on modelling of user knowledge, plans, and preferences in a domain. It focuses on stereotype (as opposed to individual) profiles. The need for a profile that supports reasoning is also stressed out in [15]. The goals listed above can be achieved through the use of ontologies. Ontologies in the form of hierarchies of user
interests have been proposed in [9]. Gauch at al. [10] also characteristics and not the dynamic ones, like his her current proposed a system that adapts information navigation based position on a user profile structured as a weighted concept hierarchy Gruber's design criteria [32(clarity, coherence, The user may create his/her own concept hierarchy and use it extensibility, minimal encoding bias, minimal ontological for browsing web sites. A user model can also be built using commitment) were taken into account during the creation an ontology schema. Razmerita et al. [11] presented a generic process. In order to create a simple yet extensible and ontology-based user modelling architecture applied in the adaptable model, user profile information models maintained context of a Knowledge Management System by various applications, like 34 were gathered and In the field of ontology design, efforts have been made by examined; general ontologies like the ones presented in [291 several research groups to facilitate the ontology engineering were also taken into account. process, employing both manual and semi-automatic methods. At this point no automatic concept extraction has been used Semi-automatic methods focus on the acquisition of as the information in the available profile models did not ontologies from domain texts. In [2], for example, a contain high level concepts but rather instances of possible framework is proposed with this objective, it incorporates concepts and slot names Consequently, the ontology designers several information extraction and learning approaches, in team proceeded by analyzing the semantics of the profile order to face the discovery of relevant classes, their models and suggesting concepts that would adequately model organization in a taxonomy and the non-taxonomic them. Table I exemplifies this procedure by presenting how relationships between classes. Comprehensive surveys of certain information from the IcQ [34 user profile were existing methodologies can be found in [3 and [4]. mapped to ontology constructs Throughout the ontology creation process, the designers may TABLE I. ICQ USER PROFILE EXCERPT AND ITS TRANSLATION INTO USER ike into account a set of ontology desi PROFILE CLASSES clarity, coherence and extensibility [5] Category Property Modeling in the Profile Ontology IIL. ONTOLOGY CREATION ISSUES Slots of the"Living Conditions As seen form the previous section, ontologies as a notion State mostly limited to taxonomies of user interests. Bearing in Bith of li.ntry have already been introduced in the context of user profiling Slots of the "Person"class mind that for most applications profiling is not restricted to user interests but also encompasses other user characteristics Personal Slot of the"Person"class (such as education, expertise and computer literacy level ), our 0 purpose is to incorporate them in a user profile ontology. This Gender Slot of the"Person"class Not necessary, may be calculated by section,after a brief definition of the ontology concept, presents description of our method for creating the ontology Date of birth of the "Person"class A. Ontology Definition Spoken Languages Could be added as Instances of the As defined in [1], an ontology is a formal explicit “ Education” class description of a domain, consisting of classes, which are the concepts found in the domain(also called entities). Each class User profile models sourced from bibliography were also may have one or more parent classes (is-a or inheritance considered and concepts from these were appropriately links), formulating thus a specialization/generalization adapted and included in the ontology. Information from hierarchy; a class has properties or slots(also called roles or bibliographic sources was exploited for selecting the basic set attributes)describing various features of the modeled class, of upper level classes and restrictions on the slots(also referred to as facets or role Tazari et al [18] suggest the following concepts as descriptions). Each slot, in turn, has a type and could have a important for user profiling: User identity, characteristics, restricted number of allowed values, which may be of simple capabilities, universal preferences, state of the user, types (strings, numbers, booleans or enumerations) or application-specific preferences. Other concepts like current instances of other classes. Classes may have instances, which activity, current terminal, location, motion state and correspond to individual objects in the domain of discourse, orientation are mentioned, but have not been included in this each instance has a concrete value for each slot of the class it ontology as they refer to a dynamic profile. They also propose belongs to. An ontology together with a set of individual a group of parameters concerning personal information(name, instances of classes constitutes a knowledge base birthday, address, bank account, and credit card), general B. Ontology Creation Resource characteristics(physical factors: weight and height, physical disabilities and abilities: reading, speaking and writing) For the creation of the ontology we adopted a top-down education, occupation, interaction-related information, approach; firsty 5t and specialized. The focus of the ontology Interests([91, [101, [20], [23])and preferences[8], [23]are lecting important general concepts, which expertise and user state were later enriched the static profile of the user, his/her more or less permanent considered of particular importance for most applications that
interests have been proposed in [9]. Gauch at al. [10] also proposed a system that adapts information navigation based on a user profile structured as a weighted concept hierarchy. The user may create his/her own concept hierarchy and use it for browsing web sites. A user model can also be built using an ontology schema. Razmerita et al. [11] presented a generic ontology-based user modelling architecture applied in the context of a Knowledge Management System. In the field of ontology design, efforts have been made by several research groups to facilitate the ontology engineering process, employing both manual and semi-automatic methods. Semi-automatic methods focus on the acquisition of ontologies from domain texts. In [2], for example, a framework is proposed with this objective; it incorporates several information extraction and learning approaches, in order to face the discovery of relevant classes, their organization in a taxonomy and the non-taxonomic relationships between classes. Comprehensive surveys of existing methodologies can be found in [3] and [4]. Throughout the ontology creation process, the designers may take into account a set of ontology design criteria, such as clarity, coherence and extensibility [5]. III. ONTOLOGY CREATION ISSUES As seen form the previous section, ontologies as a notion have already been introduced in the context of user profiling. The ontologies used however in relation with user profiles are mostly limited to taxonomies of user interests. Bearing in mind that for most applications profiling is not restricted to user interests but also encompasses other user characteristics (such as education, expertise and computer literacy level), our purpose is to incorporate them in a user profile ontology. This section, after a brief definition of the ontology concept, presents description of our method for creating the ontology. A. Ontology Definition As defined in [1], an ontology is a formal explicit description of a domain, consisting of classes, which are the concepts found in the domain (also called entities). Each class may have one or more parent classes (is-a or inheritance links), formulating thus a specialization/generalization hierarchy; a class has properties or slots (also called roles or attributes) describing various features of the modeled class, and restrictions on the slots (also referred to as facets or role descriptions). Each slot, in turn, has a type and could have a restricted number of allowed values, which may be of simple types (strings, numbers, booleans or enumerations) or instances of other classes. Classes may have instances, which correspond to individual objects in the domain of discourse; each instance has a concrete value for each slot of the class it belongs to. An ontology together with a set of individual instances of classes constitutes a knowledge base. B. Ontology Creation Resources For the creation of the ontology we adopted a top-down approach; firstly selecting important general concepts, which were later enriched and specialized. The focus of the ontology is the static profile of the user, his/her more or less permanent characteristics and not the dynamic ones, like his her current position. Gruber’s design criteria [32] (clarity, coherence, extensibility, minimal encoding bias, minimal ontological commitment) were taken into account during the creation process. In order to create a simple yet extensible and adaptable model, user profile information models maintained by various applications, like [34] were gathered and examined; general ontologies like the ones presented in [29] were also taken into account. At this point no automatic concept extraction has been used, as the information in the available profile models did not contain high level concepts but rather instances of possible concepts and slot names Consequently, the ontology designers team proceeded by analyzing the semantics of the profile models and suggesting concepts that would adequately model them. Table I exemplifies this procedure by presenting how certain information from the ICQ [34] user profile were mapped to ontology constructs. TABLE I. ICQ USER PROFILE EXCERPT AND ITS TRANSLATION INTO USER PROFILE CLASSES ICQ Profile Category ICQ Profile Property Modeling in the Profile Ontology Home Street Address Zip City State Country Slots of the “Living Conditions” class Place of Birth City State Country Slots of the “Person” class Personal Info Homepage Slot of the “Person” class Gender Slot of the “Person” class Age Not necessary, may be calculated by the date of birth Date of Birth Slot of the “Person” class Zodiac Sign Not necessary, may be calculated by the date of birth Spoken Languages Could be added as Instances of the “Education” class User profile models sourced from bibliography were also considered and concepts from these were appropriately adapted and included in the ontology. Information from bibliographic sources was exploited for selecting the basic set of upper level classes. Tazari et al [18] suggest the following concepts as important for user profiling: User identity, characteristics, capabilities, universal preferences, state of the user, application-specific preferences. Other concepts like current activity, current terminal, location, motion state and orientation are mentioned, but have not been included in this ontology as they refer to a dynamic profile. They also propose a group of parameters concerning personal information (name, birthday, address, bank account, and credit card), general characteristics (physical factors: weight and height, physical disabilities and abilities: reading, speaking and writing), education, occupation, interaction-related information, expertise and user state. Interests ([9], [10], [20], [23]) and preferences [8], [23] are considered of particular importance for most applications that
incorporate profiles. Interests are in some cases organized in Expertise Includes all kinds of expertise, like computer hierarchies of concepts [9][10]. Abilities, both physical and mental also seem to be relevant [24]. For example, the ability Thing Living things or Non Living Things the user may of a user to mentally rotate two or three dimensional objects use,a book or a pet affects the interpretation of a picture [17]. The gender factor Table II presents an overview of the proposed ontology also has been proven to affect the performance of different upper level classes and Figure I the class hierarchy of the users while interacting with the same system [19 ontology as displayed in Protege User expertise, either computer-related or related to another The"Person"class is the central one in the ontology, as it domain is a concept necessary for many profiling applications contains all the user profile characteristics. These may be of [23]. Defining a universal and adequately objective expertise simple type, like the user"name" or"date of birth", or may be measure with clearly defined categories is not an easy task instances of other ontology classes, like"physical [25] and is out of the scope of this work. However, by characteristics","contacts",etc studying the existing literature, properties relevant to user The rest of the classes are used to describe the complex user expertise and competence have been identified and included to characteristics. "Living conditions","Contact,"Education following section and is available in Protege and RDF format describing the reduvity"and"Profession"include a set of slots the user model ontology. The ontology is described in the"Expertise","Ac pective aspects of the user's life as well as a in[23] time period which represents the duration of that particular aspect. For example nay have had a"Contact"of type IV ONTOLOGY DESCRIPTION friend" from 1989 to 2004. The slot"person"of the .. This section presents a brief description of the user profile Contact"class has as type an instance of the class"Person' tology. The ontology may be extended through inheritance This way, relations between different users may be modeled as well and the addition of more classes, as well as concept Interest;,“ Preference,“ Ability","“ Characteristic"and instantiation according to the needs of a specific application. " Thing"contain only three slots: type ,"name"and"score" As a result, it may be used for the representation of both stereotype profiles(i.e. user profiles that represent a specific (or"value"in the case of "Thing")."Thing"has two sub- user category, like" computer expert"or " woman")and classes. "Living Thing "and"Non Living Thing"as modeled in the WORLDNET ontology [30 In the case of This ontology presents information that is mostly static and interests, apart from the"type"slot, which is a String, a slot permanent. More dynamic characteristics like the current amed"interest typeof type"Interest"has been ad ed position of the user when moving are currently not included allow the creation of interest hierarchies as the one The temporal aspect of some of the ontology classes has been suggested in [9] and [10]. Table Ill shows an example ININTEREST HIERARCHY MAY BE MODELED taken into account however. The ontology allows the WITH THE USER PROFILE ONTOLOGY existence of multiple instances of classes that represent characteristics that may change with the passage of time, like Interest hierarchy (Type, Name) ving conditions for example. These classes include a period representing the validity period of their instances, for ( Business, Investing) Stocks Bonds (Investing, Stocks bonds example, "Living Conditions: New York, 12/3/2003 (, 8/8/2007 Sports, Basketball) TABLE IL. USER PROFILE ONTOLOGY UPPER LEVEL CLASSES Professional ( Basketball, Professional) College University (Basketball, College Class Name Class Description University) Person Basic User Information like name, date of birth, User expertise according to [26] may be defined as a combination of three dimensions: breadth. the extent or Characteristic General user characteristics, like eye color, height, weight, etc. variety of different tools, skills and knowledge the user may Ability User abilities and disabilities. both mental and posses, depth, the completeness of the user'scurrent knowledge of a particular domain, and finesse, which refers to Condi residence and house type innovativeness and creativity. Breadth and depth are Contas ns, with whom the pe developed over time through a combination of study and hands-on use. whereas finesse is more related to the user's User preferences, for example "loves cats", "likes personality. These properties are included in the user ontology User hobby or work-related interests. Fc slots cample, "interested in sports","interested The notion of"experience atoms" is introduced in [25] They are defined as elementary units of experience as a result User activities, hobby or work related. For of activity in a particular domain. Experience atoms may be 4th crusade" expressed in the user ontology as individual instances of the Education User education including for ex Expertise"class university diplomas and langu Profession The user's profess Experience refering to the use of computers related to duration and frequency of usage [28.A
incorporate profiles. Interests are in some cases organized in hierarchies of concepts [9], [10]. Abilities, both physical and mental also seem to be relevant [24]. For example, the ability of a user to mentally rotate two or three dimensional objects affects the interpretation of a picture [17]. The gender factor also has been proven to affect the performance of different users while interacting with the same system [19]. User expertise, either computer-related or related to another domain is a concept necessary for many profiling applications [23]. Defining a universal and adequately objective expertise measure with clearly defined categories is not an easy task [25] and is out of the scope of this work. However, by studying the existing literature, properties relevant to user expertise and competence have been identified and included to the user model ontology. The ontology is described in the following section and is available in Protégé and RDF format in [23]. IV. ONTOLOGY DESCRIPTION This section presents a brief description of the user profile ontology. The ontology may be extended through inheritance and the addition of more classes, as well as concept instantiation according to the needs of a specific application. As a result, it may be used for the representation of both stereotype profiles (i.e. user profiles that represent a specific user category, like “computer expert” or “woman”) and individual ones. This ontology presents information that is mostly static and permanent. More dynamic characteristics like the current position of the user when moving are currently not included. The temporal aspect of some of the ontology classes has been taken into account however. The ontology allows the existence of multiple instances of classes that represent characteristics that may change with the passage of time, like living conditions for example. These classes include a period representing the validity period of their instances, for example, “Living Conditions: New York, 12/3/2003 – 18/8/2007”. TABLE II. USER PROFILE ONTOLOGY UPPER LEVEL CLASSES Class Name Class Description Person Basic User Information like name, date of birth, e-mail Characteristic General user characteristics, like eye color, height, weight, etc. Ability User abilities and disabilities, both mental and physical Living Conditions Information relevant to the user’s place of residence and house type. Contact Other persons, with whom the person is related, including relatives, friends, co-workers. Preference User preferences, for example “loves cats”, “likes blue color” or “dislikes classical music” Interest User hobby or work-related interests. For example, “interested in sports”, “interested in cooking” Activity User activities, hobby or work related. For example, “collects stamps” or “investigates the 4th Crusade” Education User education issues, including for example university diplomas and languages Profession The user’s profession Expertise Includes all kinds of expertise, like computer expertise Thing Living things or Non Living Things the user may posses or otherwise be related to, like a car, a house, a book or a pet Table II presents an overview of the proposed ontology upper level classes and Figure 1 the class hierarchy of the ontology as displayed in Protégé. The “Person” class is the central one in the ontology, as it contains all the user profile characteristics. These may be of simple type, like the user “name” or “date of birth”, or may be instances of other ontology classes, like “physical characteristics”, “contacts”, etc. The rest of the classes are used to describe the complex user characteristics. “Living conditions”, “Contact”, “Education”, “Expertise”, “Activity” and “Profession” include a set of slots describing the respective aspects of the user’s life as well as a time period which represents the duration of that particular aspect. For example, a user may have had a “Contact” of type “friend” from 1989 to 2004. The slot “person” of the “Contact” class has as type an instance of the class “Person”. This way, relations between different users may be modeled as well. “Interest”, “Preference”, “Ability”, “Characteristic” and “Thing” contain only three slots: “type”, “name” and “score” (or “value” in the case of “Thing”). “Thing” has two subclasses, “Living Thing” and “Non Living Thing” as modeled in the WORLDNET ontology [30] [31]. In the case of interests, apart from the “type” slot, which is a String, a slot named “interest type” of type “Interest” has been added to allow the creation of interest hierarchies, as the ones suggested in [9] and [10]. Table III shows an example. TABLE III. AN EXAMPLE OF HOW AN INTEREST HIERARCHY MAY BE MODELED WITH THE USER PROFILE ONTOLOGY Interest hierarchy “Interest” Instances (Type, Name) Business Investing Stocks & Bonds Sports Basketball Professional College & University (, Business) (Business, Investing) (Investing, Stocks & Bonds) (, Sports) (Sports, Basketball) (Basketball, Professional) (Basketball, College & University) User expertise according to [26] may be defined as a combination of three dimensions: breadth, the extent or variety of different tools, skills and knowledge the user may posses, depth, the completeness of the user’s current knowledge of a particular domain, and finesse, which refers to innovativeness and creativity. Breadth and depth are developed over time through a combination of study and hands-on use, whereas finesse is more related to the user’s personality. These properties are included in the user ontology as slots. The notion of “experience atoms” is introduced in [25]. They are defined as elementary units of experience as a result of activity in a particular domain. Experience atoms may be expressed in the user ontology as individual instances of the “Expertise” class. Experience refering to the use of computers is very often related to duration and frequency of usage [28]. A
questionnaire of perceived user expertise in a series of end documents) and consists of documents issued in the user computing related sub-domains is used in [27] in order to University since its foundation(1837) calculate the expertise level of the user by the combination of Currently, the Historical Archive can be visited by anybody the scores supplied by the user in each question. The idea of -members of the University or not- who is interested in assigning a score or level to expertise is expressed through the searching for information relative to the contents of such an score" slot in the "Expertise" class Archive. Visitors submit requests to the Historical Archive It should be mentioned here that in the case of the staff, which subsequently undertakes the task of locating the Expertise" class, the aim was to collect from the existing relevant documents and presenting them to the visitors literature user characteristics that may serve as indications or In the above-mentioned framework, a novel information factors during the assessment of the user expertise level. The retrieval system is being developed in order to render the definition of the expertise levels themselves and the expertise corpus available directly to its users measures are application-specific and out of the scope of the Users who come to the Historical Archive to retrieve current work information vary in multiple ways. For example, they have To sum up, the"Expertise" class has been created as a different educational levels, ranging from users who only container for both expertise measures and expertise scores in completed elementary or secondary school to users who order to accommodate the particular needs of individual possess postgraduate degrees (MScs, PhDs)in various pplications that make use of profiling. The following section scientific subjects. They also have differences in their provides two examples of such applications and how the user experience using the computer, ranging from those who are ontology may be used in each case. beginners to the ones very experienced and computer-skilled CLASS BROWSER Regardless of their computer expertise, users have been For Project: o profile-onto-v3a found to have different ways for foraging the information they are interested in, and these differences depend on individual Class Hierarchy preferences and existing knowledge. Individual differences THING a major factor that influences the user profile O: SYSTEM-CLASS from personal preferences and existing knowledge, cognitive e Person (4) abilities, specific aims and tasks to be solved, the gender, the age, the profession and the living environment of the user ● Contact(2 constitute properties of individuality, which is a fundamental v● Expertise part of the user profile. Moreover, the steps a user performs O Computer Expertise(1) while trying to reach the information needed, the -so called ● Preference(3) history of the user, plays an important role in sketching out ● Activity(1) his/her profile and deciding which visualization system suits him/her best, so as to employ this system the next time s/he ● Education(1) returns to search for information. Important properties related ● Living Conditions(1) to the user profile are listed in Table Iv ● Profession(1) ● Characteristics User Context Property Valu ● Ability ● Cognitive Abiliy(4) Elementary ● Physical Abilty University relation/role/title v●Thng ● Non Living Thing Student ● Living Thing Information Retrieval Knowledge Fig. I The User Profile Ontology as displayed in Protege V CASE STUDIES/EXAMPLES In order to demonstrate the adaptability of the proposed Gender user profile ontology, two case studies will be presented, one Profession Instance of"profession related to personalized, adaptive visualization and the other to Living environment Instance of"Living environment abilities personal information management. A. Personalized visualization otation A context-related research is being developed in the Motor skills framework of digitizing the Historical Archives of the Apart from the user profile, the visualization environment University of Athens, Greece. The corpus of the above- proposed for the information search in the Historical Archive mentioned archive is very large(more than 4.000,000 takes into consideration both the system context and the
questionnaire of perceived user expertise in a series of end user computing related sub-domains is used in [27] in order to calculate the expertise level of the user by the combination of the scores supplied by the user in each question. The idea of assigning a score or level to expertise is expressed through the “score” slot in the “Expertise” class. It should be mentioned here that in the case of the “Expertise” class, the aim was to collect from the existing literature user characteristics that may serve as indications or factors during the assessment of the user expertise level. The definition of the expertise levels themselves and the expertise measures are application-specific and out of the scope of the current work. To sum up, the “Expertise” class has been created as a container for both expertise measures and expertise scores in order to accommodate the particular needs of individual applications that make use of profiling. The following section provides two examples of such applications and how the user ontology may be used in each case. Fig. 1 The User Profile Ontology as displayed in Protégé V. CASE STUDIES/EXAMPLES In order to demonstrate the adaptability of the proposed user profile ontology, two case studies will be presented, one related to personalized, adaptive visualization and the other to personal information management. A. Personalized Visualization A context-related research is being developed in the framework of digitizing the Historical Archives of the University of Athens, Greece. The corpus of the abovementioned archive is very large (more than 4,000,000 documents) and consists of documents issued in the University since its foundation (1837). Currently, the Historical Archive can be visited by anybody -members of the University or not- who is interested in searching for information relative to the contents of such an Archive. Visitors submit requests to the Historical Archive staff, which subsequently undertakes the task of locating the relevant documents and presenting them to the visitors. In the above-mentioned framework, a novel information retrieval system is being developed in order to render the corpus available directly to its users. Users who come to the Historical Archive to retrieve information vary in multiple ways. For example, they have different educational levels, ranging from users who only completed elementary or secondary school to users who possess postgraduate degrees (MScs, PhDs) in various scientific subjects. They also have differences in their experience using the computer, ranging from those who are beginners to the ones very experienced and computer-skilled. Regardless of their computer expertise, users have been found to have different ways for foraging the information they are interested in, and these differences depend on individual preferences and existing knowledge. Individual differences constitute a major factor that influences the user profile. Apart from personal preferences and existing knowledge, cognitive abilities, specific aims and tasks to be solved, the gender, the age, the profession and the living environment of the user constitute properties of individuality, which is a fundamental part of the user profile. Moreover, the steps a user performs while trying to reach the information needed, the -so called- “history” of the user, plays an important role in sketching out his/her profile and deciding which visualization system suits him/her best, so as to employ this system the next time s/he returns to search for information. Important properties related to the user profile are listed in Table IV. TABLE IV PROPERTIES OF THE USER CONTEXT User Context Property Values Education Primary Elementary Higher University relation/role/title Faculty members Administrative personnel Student None of the above or no relation Information Retrieval Knowledge Perfect Medium Novice Aim Research Publication Personal Information Age Integer Gender Male/Female Profession Instance of “Profession” Living environment Instance of “Living environment” Abilities Visual memory Arithmetic memory Color recognition Mental rotation Motor skills Apart from the user profile, the visualization environment proposed for the information search in the Historical Archive takes into consideration both the system context and the
document collection context. Tables v and vi list matches it against the features of all available visualization representative properties of the system and document methods. The method found most appropriate for the contexts collection contexts at hand is selected to perform the visualization In order to select the most prominent visualization method B. Personal Information Management for each case, the values for all properties of the user, system As part of the EU DELOS Network of Excellence, the TIM and document collection context are computed and, subsequently, the computed property list is matched against project(Task-centered Information Management) is studying the feature profile of each available visualization method. the potential for users to store files, email, etc, indexed by Matching is performed through a set of rules, which each rule personal ontologies. Design and implementation of the method is considered to be helpful, impeding or neutral for a ke totype tool OntoPIM [7] is still in early stages but several indicating whether a particular feature of a visualization prc issues are already apparent. specific context characteristic. For example, the rule (user context, spatial memory, yes)=> Document Collection Context Property Values (metaphor, landscape, 70) Criterion of categorization states that if a visualization is to be performed for a user having spatial memory, then methods employing the Categories of documents Number of elements landscape metaphor are considered as"strong candidates"(as indicated by the score value 70), since the particular users ability allows him/her to exploit the visualized items'spatial Text documents Manuscript only placement so as to perceive the visualization more effectively 21]. Score values are drawn from the range [-100, 100] with positive values being used for"helpful features and negative values being used for"impediments". For more information Metadata on the visualization method selection algorithm, the interested parten reader is referred to[22] TABLE V PROPERTIES OF SYSTEM CONTEXT Static ystem Context Property Values Collection origin OntoPIM relies on the use of a Personal Ontology that Input devices describes the users domain of interest. The ontology is ed input devices ( 3D ersonal in the sense that it reflects the user's view of the domain(s). It is used to assign semantics to the information contained in the user document repository in order to be able devices 3D monitors to retrieve this information more easily. With the use of the Head mounted displays Semantic Save, it provides the user the possibility to store any object of interest according to its semantics, i.e. to relate it to Other hardware equipment C the concepts of the Personal Ontology, where an object may be an e-mail, a document, a picture, or any other type of data Bearing in mind that every user has his/her own domain of interest, personalization issues are very relevant here The user profile is represented in the system via the user OntoPIM proposes the creation of an initial library of profile ontology. An example of such a user profile, also available as an Instance in the profile ontology available in ontologies suitable for various user groups and domains [23, is the following These ontologies should be created beforehand after an A female(gender) 20 years old(age) student (profession) elaborate user study There are cases, of course that some final tuning will be Maria Papadopoulou(name), wants to retrieve information necessary to adjust the ontology template in order to about the Department she is studying in. She wants to write an accurately reflect user characteristics and interests. This can about the evolution of the Department of Informatics and done both manually(by the user)and automatically(by the Telecommunications as far as teaching in it is concerned. She system), employing a user profiling mechanism. The ontology is very experienced in using the computer ( computer presented in this work may be used to model this user profile expertise: high) and in searching for information, mainly The profile ontology may, to some extent, be populated automatically with user information that is available within the though the Internet(web search expertise high). She uses a file system. This may include PC with traditional Io devices (system context: mouse, keyboard, 15, 2D monitor). She has already visited the Chosen language and time-zone. These could give Historical Archive in the past and according to her previous interaction she likes exploring 3D environments (registered conditions. Dialing codes and ip addresses can be also used for determining the users preferences-history Current file structure. If the has created a more The system collects the above contextual information and elaborate file structure than that already provided by the
document collection context. Tables V and VI list representative properties of the system and document collection contexts. In order to select the most prominent visualization method for each case, the values for all properties of the user, system and document collection context are computed and, subsequently, the computed property list is matched against the feature profile of each available visualization method. Matching is performed through a set of rules, which each rule indicating whether a particular feature of a visualization method is considered to be helpful, impeding or neutral for a specific context characteristic. For example, the rule (user_context, spatial_memory, yes) => (metaphor, landscape, 70) states that if a visualization is to be performed for a user having spatial memory, then methods employing the landscape metaphor are considered as “strong candidates” (as indicated by the score value 70), since the particular user’s ability allows him/her to exploit the visualized items’ spatial placement so as to perceive the visualization more effectively [21]. Score values are drawn from the range [-100, 100] with positive values being used for “helpful” features and negative values being used for “impediments”. For more information on the visualization method selection algorithm, the interested reader is referred to [22]. TABLE V. PROPERTIES OF SYSTEM CONTEXT System Context Property Values Input devices Mouse Keyboard Joystick Specialized input devices (3D mouse, glove, etc.) Output devices 2D monitors 3D monitors Head mounted displays Other hardware equipment Processor Memory Graphics The user profile is represented in the system via the user profile ontology. An example of such a user profile, also available as an Instance in the profile ontology available in [23], is the following. A female (gender) 20 years old (age) student (profession), Maria Papadopoulou (name), wants to retrieve information about the Department she is studying in. She wants to write an article to publish in the Department's newspaper (activity) about the evolution of the Department of Informatics and Telecommunications as far as teaching in it is concerned. She is very experienced in using the computer (computer expertise: high) and in searching for information, mainly though the Internet (web search expertise: high). She uses a PC with traditional I/O devices (system context: mouse, keyboard, 15΄΄, 2D monitor). She has already visited the Historical Archive in the past and according to her previous interaction she likes exploring 3D environments (registered preferences-history). The system collects the above contextual information and matches it against the features of all available visualization methods. The method found most appropriate for the contexts at hand is selected to perform the visualization. B. Personal Information Management As part of the EU DELOS Network of Excellence, the TIM project (Task-centered Information Management) is studying the potential for users to store files, email, etc, indexed by personal ontologies. Design and implementation of the prototype tool OntoPIM [7] is still in early stages but several key issues are already apparent. TABLE VI. PROPERTIES OF THE DOCUMENT COLLECTION CONTEXT Document Collection Context Property Values Categories of documents Criterion of categorization Number of elements Relation between categories Text documents Full text Image Manuscript only Meetings’ minutes Metadata Author Title Type Date of issue Department of issue Keywords Categories Collection origin Static Dynamic OntoPIM relies on the use of a Personal Ontology that describes the user's domain of interest. The ontology is personal in the sense that it reflects the user’s view of the domain(s). It is used to assign semantics to the information contained in the user document repository in order to be able to retrieve this information more easily. With the use of the Semantic Save, it provides the user the possibility to store any object of interest according to its semantics, i.e. to relate it to the concepts of the Personal Ontology, where an object may be an e-mail, a document, a picture, or any other type of data. Bearing in mind that every user has his/her own domain of interest, personalization issues are very relevant here. OntoPIM proposes the creation of an initial library of ontologies suitable for various user groups and domains. These ontologies should be created beforehand after an elaborate user study. There are cases, of course, that some final tuning will be necessary to adjust the ontology template in order to accurately reflect user characteristics and interests. This can be done both manually (by the user) and automatically (by the system), employing a user profiling mechanism. The ontology presented in this work may be used to model this user profile. The profile ontology may, to some extent, be populated automatically with user information that is available within the file system. This may include: • Chosen language and time-zone. These could give information about the user nationality and living conditions. Dialing codes and IP addresses can be also used for determining the users’ location. • Current file structure. If the user has created a more elaborate file structure than that already provided by the
operating system to store his/her files, then it could also B R Cuel, "A Survey on Ontology Creation Methodologies be a source of new concepts. a dictionary of synonyms 24 69,2005 ould be used here to make the matching more effective. (4) Survey and Comparative Review", Al Magazine, 18(3), Fall 1997, 53. The user could also be prompted to indicate folder C. Hafner, "The State of the Art in Ontology Design, structures that contain documents relevant to his/her nterests or activities. For example, if the user has a folder [ 5] C. Fluit, M. Sabou, F. van Harmelen,"Ontology-based Information named“ Articles” with sub- folders like“ basketball”or ualisation", In Visualising the Semantic Web, Springer Verlag, 2002 gardening"usedtofurthercategorizethedocuments[6]deLosNoe,http://http://delos-noe.iei these concepts may be used to populate the"Interest [7] V. Katifori, A. Poggi, M y Class of the profile. The file content could be used as well (2005). OntoPIM: how to rely on a personal ontology for Personal to support this concept extraction Information Management. In Proc. of the Ist Workshop on The Semantic The system can also scan address books in order to [8]A.Kobsa, User Modelling: Recent work, prospects and hazards, retrieve contact information and populate the"Contac class with instances Hufschmidt, T Khme, U. Malinowski, eds. 1993) calendars and to-do lists may be used to identify user 9] J. Trajkova, S Gauch, Improving Ontology-based user Profiles, Proc. of activities RIAO 2004, University of Avignon (Vaucluse), France, April 26-28 2004,pp.380-389 The web cache and bookmark/favorites structure could [10S. Gauch, J Chaffee, A. Pretschner, Ontology-Based User Profiles f also be a possible source for deriving interests and Search and Browsing, User Modeling and User-Adapted Interaction: preferences. The user can also be prompted, through an appropriate questionnaire to provide information Modeling for Web and Hypermedia Information Retrieval, voL, (2003) concerning his/her personal data, interests, preferences, [I L Razmerita, AAngehrn, A Maedche, Ontology based user modeling ontacts etc onference, Pittsburgh, USA, Springer Verlag, pp. 213-217, 2003 As an example available in [231, Elias Daradimos(name), 29 [121 S. Lawrence, (2000). Context in web search. IEEE Data Engineering years old (age/date of birth) is a network administrator Bulletin,23(3):25-32 Profession). The OntoPIM user profile extraction mechanism [13] D. Raz, A. Juhola, J. Serrat Fernandes, A. Galis, Fast and Efficient identifies him as a resident of Athens, Greece (Living Context-Aware Services, 2006 John Wiley Sons, Ltd Conditions) and creates a list of his contacts names and e- [14]R. Dinoff, R Hull, B Kumar, D Lieuwen, P. Santos, Learning and mails(Contacts). By examining application preferences and International Workshop on Context in Advanced interfaces May 2006 system settings, asking him to fill a questionnaire, and by scanning indicated parts of the file system, the mechanism [15] E Rich, Users are individuals: individualizing user models, Intermational concludes that the user is interested in electronics, airplanes Journal of Man-machine Studies 18 (3), 199--214 otorbike mechanics and movies(Interests) [16] B. Cornelis Personalizing search in digital libraries, Master,s thesis Cs 03-01. University of Maastricht [17 B. Gutkauf, S. Thies, G. Domi User Ad This work is an attempt to create an ontology that http://wwwcs.uni-paderborn.de/csag- incorporates concepts and properties used to model the user domik/arbeitsschwerpunkte/ucmmidias/root. html profile. ExISting literature, applications and ontologies related [18] M.R. Tazari, mm, M. Finke, (2003), Modeling User Context, to the domain of user context and profiling have been taken Interaction(HCIl2003), Crete(Greece), June 2003 nto account in order to create a general, comprehensive and [9]G. S. Hubona and G. W. Shirah, The Gender Factor Performing extensible user model. The model, available in [23, is also Visualization Tasks on Computer Media, Proceedings of the 37th presented through two examples in two different areas, Annual Hawaii International Conference on System Sciences (HICSS04)-Track 4, Big Island, HI, P 40097c, 2004 As this model focuses more on static user characteristics it [20] J. Teevan, S. T. Dumais, and E. Horvitz, Personalizing Search personal information management and adaptive visualization Automated Analysis of Interests and Activities, Proceedings of SIGir is our future aim to study the incorporation of dynamic and 2005, ACM Press, August 2005 temporal characteristics in order to cater for a wider range of van Dantzich, Data Mountain: Using Spatial Memory for Document applications that include profiling. Furthermore, the Management. Proceedings of ACM UIST 98 Symposium on User acquisition of the profile properties for individual users Interface Software&Technology, San Fransisco, CA, 153-162. through questionnaires is investigated, in order to compliment November 1998 the user profile ontology with a means to populate it 22] M. Golemati, C. Halatsis, C. Vassilakis, V. Katifori, G. Lepouras, A Context-Based Adaptive Visualization Environment, Proceedings of the Information Visualization IV'06 Conference, London, Great Britain REFERENCES 23M. Golemati, A. Katifori, C. Vassilakis, G. Lepouras, C. Halatsis, User [ N. F Noy, D L McGuiness, "Ontology Development 101: A Guide to eating Your First Ontology", Stanford Knowledge Systems Laboratory Technical Report KSL-01-05, March 2001, Available at 24]B. Kules, User Modeling for Adaptive and Adaprable Software Systems, 2000,Availableathttp://www.otal.umd.edw/uuguide/Wmk/ v-mcguinness html edche, S. Staab, ""Mining Ontologies from Text", EKAW 2000 25]J. Fink, A. Kobsa, and A. Nill, Adaptable and Adaptive Infromation Access for All Users, Including the Disabled and the Elderly, In User 189-202 Modeling: Proceedings of the Sixth International Conference, UM97 Viena, New York, Springer Wien New York, 1997, pp 171-173
operating system to store his/her files, then it could also be a source of new concepts. A dictionary of synonyms could be used here to make the matching more effective. The user could also be prompted to indicate folder structures that contain documents relevant to his/her interests or activities. For example, if the user has a folder named “Articles” with sub-folders like “basketball” or “gardening” used to further categorize the documents, these concepts may be used to populate the “Interest” Class of the profile. The file content could be used as well to support this concept extraction. • The system can also scan address books in order to retrieve contact information and populate the “Contact” class with instances. • Calendars and to-do lists may be used to identify user activities. • The web cache and bookmark/favorites structure could also be a possible source for deriving interests and preferences. The user can also be prompted, through an appropriate questionnaire to provide information concerning his/her personal data, interests, preferences, contacts, etc. As an example available in [23], Elias Daradimos (name), 29 years old (age/date of birth) is a network administrator (Profession). The OntoPIM user profile extraction mechanism identifies him as a resident of Athens, Greece (Living Conditions) and creates a list of his contacts names and emails (Contacts). By examining application preferences and system settings, asking him to fill a questionnaire, and by scanning indicated parts of the file system, the mechanism concludes that the user is interested in electronics, airplanes, motorbike mechanics and movies (Interests). VI. CONCLUSIONS AND FUTURE WORK This work is an attempt to create an ontology that incorporates concepts and properties used to model the user profile. Existing literature, applications and ontologies related to the domain of user context and profiling have been taken into account in order to create a general, comprehensive and extensible user model. The model, available in [23], is also presented through two examples in two different areas, personal information management and adaptive visualization. As this model focuses more on static user characteristics, it is our future aim to study the incorporation of dynamic and temporal characteristics in order to cater for a wider range of applications that include profiling. Furthermore, the acquisition of the profile properties for individual users through questionnaires is investigated, in order to compliment the user profile ontology with a means to populate it. REFERENCES [1] N. F. Noy, D. L. McGuiness, “Ontology Development 101: A Guide to Creating Your First Ontology”, Stanford Knowledge Systems Laboratory Technical Report KSL-01-05, March 2001, Available at http://protege.stanford.edu/publications/ontology_development/ontology 101-noy-mcguinness.html [2] A. Maedche, S. Staab, “Mining Ontologies from Text”, EKAW 2000, 189-202 [3] M. Cristani, R. Cuel, “A Survey on Ontology Creation Methodologies”, International Journal on Semantic Web and Information Systems, Vol. 1, No. 2, 49 – 69, 2005 [4] N. F. Noy, C. Hafner, “The State of the Art in Ontology Design, A Survey and Comparative Review”, AI Magazine, 18 (3), Fall 1997, 53- 74. [5] C. Fluit, M. Sabou, F. van Harmelen, “Ontology-based Information Visualisation”, In Visualising the Semantic Web, Springer Verlag, 2002 [6] DELOS NoE, http://http://delos-noe.iei.pi.cnr.it/. [7] V. Katifori,, A. Poggi,. M. Scannapieco, T. Catarci, & Y. Ioannidis (2005). OntoPIM: how to rely on a personal ontology for Personal Information Management. In Proc. of the 1st Workshop on The Semantic Desktop. [8] A. Kobsa, User Modelling: Recent work, prospects and hazards, Adaptive User Interfaces: Principles and Practices (SchneiderHufschmidt, T. Khme, U. Malinowski, eds. 1993) [9] J. Trajkova, S. Gauch, Improving Ontology-based User Profiles, Proc. of RIAO 2004, University of Avignon (Vaucluse), France, April 26-28, 2004, pp. 380-389 [10] S. Gauch, J. Chaffee, A. Pretschner, Ontology-Based User Profiles for Search and Browsing, User Modeling and User-Adapted Interaction: The Journal of Personalization Research, Special Issue on User Modeling for Web and Hypermedia Information Retrieval, vol. , (2003) [11] L Razmerita,, A.Angehrn, A. Maedche, Ontology based user modeling for Knowledge Management Systems, Proceedings of the User Modeling Conference, Pittsburgh, USA, Springer Verlag, pp. 213-217, 2003 [12] S. Lawrence, (2000). Context in web search. IEEE Data Engineering Bulletin, 23(3):25-32 [13] D. Raz, A. Juhola, J. Serrat Fernandes, A. Galis, Fast and Efficient Context-Aware Services, 2006 John Wiley & Sons, Ltd [14] R. Dinoff, R. Hull, B. Kumar, D. Lieuwen, P. Santos, Learning and managing user context in personalized communications services, International Workshop on Context in Advanced interfaces May 2006, Venice Italy [15] E. Rich, Users are individuals: individualizing user models, International Journal of Man-machine Studies 18(3), 199--214 [16] B. Cornelis Personalizing search in digital libraries, Master's thesis CS 03-01, University of Maastricht [17] B. Gutkauf, S. Thies, G. Domik, User Adaptive Chart Editing and Presentation - Applied Through User Modeling and Critiquing Available at http://wwwcs.uni-paderborn.de/cs/agdomik/arbeitsschwerpunkte/ucmm/idias/root.html [18] M.R. Tazari, M. Grimm, M. Finke, (2003), Modeling User Context, Proceedings of the 10th International Conference on Human-Computer Interaction (HCII2003), Crete (Greece), June 2003 [19] G. S. Hubona and G. W. Shirah, The Gender Factor Performing Visualization Tasks on Computer Media, Proceedings of the 37th Annual Hawaii International Conference on System Sciences (HICSS'04) - Track 4, Big Island, HI, p 40097c, 2004 [20] J. Teevan, S. T. Dumais, and E. Horvitz, Personalizing Search via Automated Analysis of Interests and Activities, Proceedings of SIGIR 2005, ACM Press, August 2005 [21] G. G. Robertson, M. Czerwinski, K. Larson, D. Robbins, D. Thiel, M. van Dantzich, Data Mountain: Using Spatial Memory for Document Management. Proceedings of ACM UIST ’98 Symposium on User Interface Software&Technology, San Fransisco, CA, 153-162, November 1998. [22] M. Golemati, C. Halatsis, C. Vassilakis, V. Katifori, G. Lepouras, A Context-Based Adaptive Visualization Environment, Proceedings of the Information Visualization IV'06 Conference, London, Great Britain [23] M. Golemati, A. Katifori, C. Vassilakis, G. Lepouras, C. Halatsis, User Profile Ontology version 1, available at http://oceanis.mm.di.uoa.gr/pened/?category=publications [24] B. Kules, User Modeling for Adaptive and Adaprable Software Systems, 2000, Available at http://www.otal.umd.edu/UUGuide/wmk/ [25] J. Fink, A. Kobsa, and A. Nill, Adaptable and Adaptive Infromation Access for All Users, Including the Disabled and the Elderly, In User Modeling: Proceedings of the Sixth International Conference, UM97, Viena, New York, Springer Wien New York, 1997, pp 171 – 173
[26]A. Mockus, and J. D. Herbsled, Expertise Browser: A Quantitive Intemational shop on Forma pproach to Identifying Expertise. 2002, ICSE, May 19-25, Orlando, ugust 1993. Published in interna Florida, USA Studies. Volume 43. Issue 5-6 Nov. 四微 [27M. C. Munro, S.L. Huff, B L Marcolin, and D. R Compeau, 1997, 31] G. A. Miller, 1990. WORDNET: An On-Line Lexical anding and Measuring User Competence. Information Intemational Journal of Lexicography 3-4: 235-312 Management 33: 45-57 32]WordnEt,http://wordnet.princeton.edu/ [28]G. Torkjadeh, J. Lee, Measures of Perceived End-user Computing Skill B3]N. F Noy, C D. Hafner, The State of the Art in Ontology Design, A Information Management 40, Elsevier, 2003, Pp. 607-615 rvey and comparative Review, Al Magazine, Fall 1997, 53-7 [29A.Aula,k.nordHaussen,ModelingSuccessfulPerformanceinWeb[34)icq,http://www.icq.com/ arch, University of Tampere, Series of D- Net Publications, D-2005 4,June2005 30] T.R. Gruber, (1993)Toward principles for the design of ontologies used for knowledge sharing. Originally in N. Guarino and R. Poli,(Ed
[26] A. Mockus, and J. D. Herbsled, Expertise Browser: A Quantitive Approach to Identifying Expertise. 2002, ICSE, May 19-25, Orlando, Florida, USA [27] M. C. Munro, S. L. Huff, B. L. Marcolin, and D. R. Compeau, 1997, Understanding and Measuring User Competence. Information & Management 33:45–57 [28] G. Torkjadeh, J. Lee, Measures of Perceived End-user Computing Skills, Information & Management 40, Elsevier, 2003, pp. 607-615 [29] A. Aula, K. Nordhaussen, Modeling Successful Performance in Web Search, University of Tampere, Series of D – Net Publications, D-2005- 4, June 2005 [30] T. R. Gruber, (1993) Toward principles for the design of ontologies used for knowledge sharing. Originally in N. Guarino and R. Poli, (Eds.), International Workshop on Formal Ontology, Padova, Italy. Revised August 1993. Published in International Journal of Human-Computer Studies, Volume 43 , Issue 5-6 Nov./Dec. 1995, Pages: 907-928 [31] G. A. Miller, 1990. WORDNET: An On-Line Lexical Database. International Journal of Lexicography 3–4: 235–312. [32] WORDNET, http://wordnet.princeton.edu/ [33] N. F. Noy, C. D. Hafner, The State of the Art in Ontology Design, A Survey and comparative Review, AI Magazine, Fall 1997, 53-74 [34] ICQ, http://www.icq.com/