smar museum Project FP7-216923 SMARTMUSEUM Cultural Heritage Knowledge Exchange Platform Deliverable d2.2 SMARTMUSEUM Report describing methods for dynamic user profile creation orkpackage WP2- Self adaptive user profile management Task T2.3-Developing theoretical solution for dynamic user profile creation and modification Version 1.00 Date March 11 2009 Classification Public Status Draft Abstract SMARTMUSEUM (Cultural Heritage Knowledge Exchange Platform) is a Research and Development project sponsored under the Europeans Commission,s 7th Framework. The overall objective of the project is to develop a platform for innovative services enhancing on-site personalized access to digital cultural heritage through adaptive and privacy preserving user profiling Using on- site knowledge databases, global digital libraries and visitors'experiential knowledge, the platform makes possible the creation of innovative multilingual services for increasing interaction between visitors and cultural heritage objects in a future smart museum environment, taking full benefit of digitized cultural information The main objective of this deliverable is to describe a theoretical framework for management of dynamic user profiles SmarTmuseuMConsortium(www.smartmuseum Authors Responsible Editors: KTH, ELlKO, TKK apprise H Heritage Malta Grant Agreement Number: FP7-216923 国如 Platform
Grant Agreement Number: FP7-216923 Acronym: SMARTMUSEUM Project title: Cultural Heritage Knowledge Exchange Platform Project FP7-216923 SMARTMUSEUM Cultural Heritage Knowledge Exchange Platform Deliverable D2.2 SMARTMUSEUM Report describing methods for dynamic user profile creation Workpackage WP2 – Self adaptive user profile management Task T2.3 - Developing theoretical solution for dynamic user profile creation and modification Version 1.00 Date March 11, 2009 Classification Public Status Draft Abstract SMARTMUSEUM (Cultural Heritage Knowledge Exchange Platform) is a Research and Development project sponsored under the Europeans Commission’s 7th Framework. The overall objective of the project is to develop a platform for innovative services enhancing on-site personalized access to digital cultural heritage through adaptive and privacy preserving user profiling. Using onsite knowledge databases, global digital libraries and visitors’ experiential knowledge, the platform makes possible the creation of innovative multilingual services for increasing interaction between visitors and cultural heritage objects in a future smart museum environment, taking full benefit of digitized cultural information. The main objective of this deliverable is to describe a theoretical framework for management of dynamic user profiles. Authors SMARTMUSEUM Consortium (www.smartmuseum.eu) Responsible Editors: KTH, ELIKO, TKK
smar museum Executive Summary The purpose of this deliverable D2. 2 is introducing a theoretical framework for dynamic user profile management and its related operations specifically creation and modification of user profiles Existing methods for user behaviour monitoring and formalisation are studied and a state of art in broader picture of contextualization and personalization is given, taken into account the domain of the project, cultural heritage domain Within the proposed framework, we introduce two approaches, mainly contributing mechanisms used for learning and building user profiles as well as mechanisms for enabler personalized recommendation and filtering on behalf of museum users. Our approaches take into account the adaptivity and dynamisms of user profiles, as suggested user profile qualities This enables us to measure how effective profiling framework works The introduced framework allowing us to create and learn dynamic user profiles for SMARTMUSUEM project. We have used the generic user profile structure introduced in D2. 1 for generating generic and dynamic user profile structures, while by utilizin collaborative filtering we learn and use profiled prefernces of users of the platform. As collaborative filtering techniques are the main foundation of the work being done, we introduce recommendation as a substantial part of the proposed framework apprise H Heritage Malta Grant Agreement Number: FP7-216923 国如 Platform
Grant Agreement Number: FP7-216923 Acronym: SMARTMUSEUM Project title: Cultural Heritage Knowledge Exchange Platform Executive Summary The purpose of this deliverable D2.2 is introducing a theoretical framework for dynamic user profile management and its related operations specifically creation and modification of user profiles. Existing methods for user behaviour monitoring and formalisation are studied and a state of art in broader picture of contextualization and personalization is given, taken into account the domain of the project, cultural heritage domain. Within the proposed framework, we introduce two approaches, mainly contributing mechanisms used for learning and building user profiles as well as mechanisms for enabling personalized recommendation and filtering on behalf of museum users. Our approaches take into account the adaptivity and dynamisms of user profiles, as suggested user profile qualities. This enables us to measure how effective profiling framework works. The introduced framework allowing us to create and learn dynamic user profiles for SMARTMUSUEM project. We have used the generic user profile structure introduced in D2.1 for generating generic and dynamic user profile structures, while by utilizing collaborative filtering we learn and use profiled prefernces of users of the platform. As collaborative filtering techniques are the main foundation of the work being done, we introduce recommendation as a substantial part of the proposed framework
smar museum Introduction SMARTMUSEUM Project Deliverable purpose, scope and context Background Introduction to personalization, user modelling and profiling Profiles presentation 5555667899 Personalization Recommendation and Contextualization in museum domain Overview of existing work on personalization in cultural heritage State-of-art approaches to user profile learning and construction 10 Conceptual Clustering Naive Bayes Bayesian Network 15 Neighbourhood based methods Self-Organizing Map Nearest Neighbour 17 Latent Semantic Indexing/ Analysis State-of-art on recommenders and recommendation techniques Collaborative filtering User-based Approaches Item-based Approaches Model-based App proaches Similarity Measures Content-based filtering Hybrid filtering amic Profile Operations Statistical Learning and Construction of User Profiles Modelling and Measuring User Preferences and Interests 22 Selecting and Combining Initial Profile Learning Techniques Association rule mining Frequent closed item set mining Simple Collaborative Filtering: Utilizing Statistical User Profiles Dataset for the initial modelling Recalculation and incremental learning of user's preference nput data format for algorithms Output data buffering for algorithm Analysis Extended Collaborative Filtering: Utilizing Semantics and Social Trust Extending Recommender Systems with Semantics: A User-Item Ontological Model..28 User ontology Item Ontology Enriching Recommendations using Distributed Social Trust apprise H Heritage Malta Grant Agreement Number: FP7-216923 国如 Platform
Grant Agreement Number: FP7-216923 Acronym: SMARTMUSEUM Project title: Cultural Heritage Knowledge Exchange Platform Introduction ................................................................................................................................ 5 SMARTMUSEUM Project .................................................................................................... 5 Deliverable purpose, scope and context................................................................................. 5 Audience................................................................................................................................. 5 Background ................................................................................................................................ 6 Introduction to personalization, user modelling and profiling............................................... 6 Profiles Presentation............................................................................................................... 7 Profiles Qualities.................................................................................................................... 8 Personalization, Recommendation and Contextualization in Museum Domain........................ 9 Overview of existing work on personalization in cultural heritage ....................................... 9 State-of-art approaches to user profile learning and construction........................................ 10 Data Mining...................................................................................................................... 11 Conceptual Clustering...................................................................................................... 12 Naive Bayes...................................................................................................................... 14 Bayesian Network ................................................................................................................15 Neighbourhood based methods........................................................................................ 16 Self-Organizing Map............................................................................................................16 Nearest Neighbour ...............................................................................................................17 Latent Semantic Indexing / Analysis ...................................................................................17 State-of-art on recommenders and recommendation techniques ......................................... 18 Collaborative filtering ...................................................................................................... 19 User-based Approaches........................................................................................................19 Item-based Approaches........................................................................................................19 Model-based Approaches.....................................................................................................19 Similarity Measures .............................................................................................................20 Content-based filtering..................................................................................................... 21 Hybrid filtering................................................................................................................. 21 Dynamic Profile Operations..................................................................................................... 22 Statistical Learning and Construction of User Profiles........................................................ 22 Modelling and Measuring User Preferences and Interests............................................... 22 Selecting and Combining Initial Profile Learning Techniques........................................ 23 Association rule mining .......................................................................................................23 Frequent closed item set mining ..........................................................................................24 Simple Collaborative Filtering: Utilizing Statistical User Profiles.................................. 24 Dataset for the initial modelling...........................................................................................24 Recalculation and incremental learning of user's preference...............................................25 Input data format for algorithms..........................................................................................25 Output data buffering for algorithms ...................................................................................26 Analysis................................................................................................................................26 Extended Collaborative Filtering: Utilizing Semantics and Social Trust ........................ 28 Extending Recommender Systems with Semantics: A User-Item Ontological Model........28 User Ontology .................................................................................................................28 Item Ontology..................................................................................................................29 Enriching Recommendations using Distributed Social Trust ..............................................30
smartmuseum ecommendation prediction process Rating a new item event Generating a New Recommendation Event Extending Recommender Systems with Socio-Semantic Trust: An Evaluation Conclusion General Conclusions Detailed Conclusions: Selection of Algorithms and mechanisms Works Cited apprise H Heritage Malta Grant Agreement Number: FP7-216923 国如 Platform
Grant Agreement Number: FP7-216923 Acronym: SMARTMUSEUM Project title: Cultural Heritage Knowledge Exchange Platform Recommendation prediction process...............................................................................30 Rating a New Item Event ................................................................................................30 Generating a New Recommendation Event ....................................................................31 Extending Recommender Systems with Socio-Semantic Trust: An Evaluation .................32 Conclusion............................................................................................................................ 33 General Conclusions ........................................................................................................ 33 Detailed Conclusions: Selection of Algorithms and Mechanisms................................... 34 Works Cited.......................................................................................................................... 35
smar museum Introduction The purpose of this section is to introduce the SMARTMUSEUM Project Purpose, scope and context of this deliverable Intended audience for the deliverable SMARTMUSEUM Project SMARTMUSEUM(Cultural Heritage Knowledge Exchange Platform) is a Research and Development project sponsored under the Europeans Commissions 7th Framework. The overall objective of the project is to develop a platform for innovative services enhancing on- site personalised access to digital cultural heritage through adaptive and privacy preserving user profiling. Using on-site knowledge databases, global digital libraries and visitors experiential knowledge, the platform makes possible the creation of innovative multilingual services for increasing interaction between visitors and cultural heritage objects in a future smart museum environment taking full benefit of digitized cultural information The SMARTMUSEUM project supports achieving the following general goals Lowering costs of on-site access to digital cultural heritage content Improving structured, user behaviour and preference dependent on-site access to the vast repository of cultural heritage, Improving the individual and shared experiences people receive from cultural and scientific resources Bringing personalised cultural experience closer to non-expert communities Making real reuse of personal experiences related to cultural heritage access for a variety of interest groups Deliverable purpose, scope and context The purpose of this deliverable D2. 2 is to present: 1)A survey of existing approaches to dynamic user profile management and 2)A framework of implemented techniques for user profile management Audience The intended audience includes Primarily SMARTmUSEUM Partners involved in developing the user profile-related operations Project partners involved in SMARTMUSEUM WP2 apprise H Heritage Malta Grant Agreement Number: FP7-216923 国如 Platform
Grant Agreement Number: FP7-216923 Acronym: SMARTMUSEUM Project title: Cultural Heritage Knowledge Exchange Platform Introduction The purpose of this section is to introduce the: SMARTMUSEUM Project Purpose, scope and context of this deliverable Intended audience for the deliverable SMARTMUSEUM Project SMARTMUSEUM (Cultural Heritage Knowledge Exchange Platform) is a Research and Development project sponsored under the Europeans Commission‘s 7th Framework. The overall objective of the project is to develop a platform for innovative services enhancing onsite personalised access to digital cultural heritage through adaptive and privacy preserving user profiling. Using on-site knowledge databases, global digital libraries and visitors‘ experiential knowledge, the platform makes possible the creation of innovative multilingual services for increasing interaction between visitors and cultural heritage objects in a future smart museum environment, taking full benefit of digitized cultural information. The SMARTMUSEUM project supports achieving the following general goals: • Lowering costs of on-site access to digital cultural heritage content, • Improving structured, user behaviour and preference dependent on-site access to the vast repository of cultural heritage, • Improving the individual and shared experiences people receive from cultural and scientific resources, • Bringing personalised cultural experience closer to non-expert communities, • Making real reuse of personal experiences related to cultural heritage access for a variety of interest groups. Deliverable purpose, scope and context The purpose of this deliverable D2.2 is to present: 1) A survey of existing approaches to dynamic user profile management and 2) A framework of implemented techniques for user profile management. Audience The intended audience includes: Primarily SMARTMUSEUM Partners involved in developing the user profile-related operations Project partners involved in SMARTMUSEUM WP2
smar museum Background In this section a brief information about the main elements of this deliverable is given. We define personalization and usage profiling, and then the presentation aspects of profiles well as qualities considered for profiling the usage in museum domain are discussed Introduction to personalization, user modelling and profiling User profiling takes its roots in human studies. A user profile is defined as gathering of raw personal material about the user, according to (Koch, May &th, 2005). User profiles gather and present cognitive skills, abilities, preferences and interaction histories with the system ( Gauch, et al., 2007). According to(Middleton, et al., 2004), User profiling is either knowledge-based or behavior-based. Knowledge-based approaches construct static models of users and match users to the closest model. Questionnaires and special forms are used to gather this user knowledge. Behavior-based methods consider the behavior as a modeling base, commonly by utilizing machine-learning techniques(Middleton, et al., 2004)( Bloedorn, et al., 1996)to discover useful patterns in the behavior. Behavioral gathering and logging is used in order to obtain the data necessary to detect and extract patterns, according to( Kobsa, 2001 Personalization systems are based on user profiles. 45 personalization systems are listed by Pretschner(Pretschner, 1999), according to Gauch(Gauch, et al., 2007). Personalization (Sieg, et al., 2007) can be provided to user by customizing the content or the visualization of the system based on user's profile( Weibelzahl, 2003) Personalization techniques fall into two main categories(Olog, et al., 2003) First category are based mostly on adapting user interfaces and content selection and rendering based on the user's performance and behavior in a certain domain. These techniques are collectively referred to as Adaptive hypermedia techniques Brusilovsky, 2001) Another category of techniques is based on cognitive patterns(such as interests, preferences, likes, dislikes, and goals)a user have. This information is mostly stored ofiles and stored at some kind of profiling or modeling server(Kobsa, 2007). These methods are known as filtering and recommendation techniques. They filter resources based on features(mostly metadata)extracted and gathered from a resource or according to ratings(generally weights) of a user of similar profile, according to Dolog and Nejdl (Dolog, et al., 2003) apprise H Heritage Malta Grant Agreement Number: FP7-216923 国如 Platform
Grant Agreement Number: FP7-216923 Acronym: SMARTMUSEUM Project title: Cultural Heritage Knowledge Exchange Platform Background In this section a brief information about the main elements of this deliverable is given. We define personalization and usage profiling, and then the presentation aspects of profiles as well as qualities considered for profiling the usage in museum domain are discussed. Introduction to personalization, user modelling and profiling User profiling takes its roots in human studies. A user profile is defined as gathering of raw personal material about the user, according to (Koch., May 8th, 2005). User profiles gather and present cognitive skills, abilities, preferences and interaction histories with the system (Gauch, et al., 2007). According to (Middleton, et al., 2004), User profiling is either knowledge-based or behavior-based. Knowledge-based approaches construct static models of users and match users to the closest model. Questionnaires and special forms are used to gather this user knowledge. Behavior-based methods consider the behavior as a modeling base, commonly by utilizing machine-learning techniques (Middleton, et al., 2004) (Bloedorn, et al., 1996) to discover useful patterns in the behavior. Behavioral gathering and logging is used in order to obtain the data necessary to detect and extract patterns, according to (Kobsa, 2001). Personalization systems are based on user profiles. 45 personalization systems are listed by Pretschner (Pretschner, 1999), according to Gauch (Gauch, et al., 2007). Personalization (Sieg, et al., 2007) can be provided to user by customizing the content or the visualization of the system based on user‘s profile (Weibelzahl, 2003). Personalization techniques fall into two main categories (Dolog, et al., 2003). First category are based mostly on adapting user interfaces and content selection and rendering based on the user's performance and behavior in a certain domain. These techniques are collectively referred to as Adaptive hypermedia techniques (Brusilovsky, 2001). Another category of techniques is based on cognitive patterns (such as interests, preferences, likes, dislikes, and goals) a user have. This information is mostly stored as user profiles and stored at some kind of profiling or modeling server (Kobsa, 2007). These methods are known as filtering and recommendation techniques. They filter resources based on features (mostly metadata) extracted and gathered from a resource or according to ratings (generally weights) of a user of similar profile, according to Dolog and Nejdl (Dolog, et al., 2003)
smar museum Profiles presentation Emergence of Semantic Web, created new possibilities for profile-driven personalization Ontologies, at the heart of Semantic Web technologies, are used for two major purposes (Olog, et al., 2003). Mainly, they are utilized to formalize domain concepts which allow describing constraints for generation or selection of resource contents belonging the domain the user is keen towards, as well as being used to formalize the user model or profile ontology (okoohaki, et al., 2008) that helps making decision which resources to be adapted(for instance, shown or not shown) to the user. Ontologies along with reasoning create formalization that boost personalization decision making mechanisms, according to dole and Nejdl (Olog, et al., 2007) Ontological user profiles are becoming widely adopted. EU-project Spice constructs a multi ontological approach, across the domain of mobile communications, from which one of the most important axis is user profile ontology(Sutterer, et al., 2008). Within the domain of digital cultural heritage, CHIP project is definitely a significant stake holder. Considerable amount of research attention has been payed on semantically formalizing the user domain (Wang, et al., 2007)(Aroyo, et al., 2007), as well as personalization of information retrieval Hybrid ontological user models are consumed to learn, gather, store and use personal user data, according to which semantically-enriched art works are recommended to, during both on-line and on-site visit to exhibition We have considered utilizing hybrid user models(Dokoohaki, et al., 2008), which incorporate a semantic presentation of personal information about users as well as incorporating notions of trust, privacy and ranking for items the user has interest towards. For a complete description of the formalization of SMARTMUSEUM user profile, reader is advised to read D2.1 1.theChip(culTuRalHeritageInformationPresentation)project.http://www.chip-proiect.orgl apprise H Heritage Malta Grant Agreement Number: FP7-216923 国如 Platform
Grant Agreement Number: FP7-216923 Acronym: SMARTMUSEUM Project title: Cultural Heritage Knowledge Exchange Platform Profiles Presentation Emergence of Semantic Web, created new possibilities for profile-driven personalization. Ontologies, at the heart of Semantic Web technologies, are used for two major purposes (Dolog, et al., 2003). Mainly, they are utilized to formalize domain concepts which allow describing constraints for generation or selection of resource contents belonging the domain the user is keen towards, as well as being used to formalize the user model or profile ontology (Dokoohaki, et al., 2008) that helps making decision which resources to be adapted (for instance, shown or not shown) to the user. Ontologies along with reasoning create formalization that boost personalization decision making mechanisms, according to Dolog and Nejdl (Dolog, et al., 2007). Ontological user profiles are becoming widely adopted. EU-project Spice constructs a multiontological approach, across the domain of mobile communications, from which one of the most important axis is user profile ontology (Sutterer, et al., 2008). Within the domain of digital cultural heritage, CHIP project1 is definitely a significant stake holder. Considerable amount of research attention has been payed on semantically formalizing the user domain (Wang, et al., 2007) (Aroyo, et al., 2007), as well as personalization of information retrieval. Hybrid ontological user models are consumed to learn, gather, store and use personal user data, according to which semantically-enriched art works are recommended to, during both on-line and on-site visit to exhibition. We have considered utilizing hybrid user models (Dokoohaki, et al., 2008), which incorporate a semantic presentation of personal information about users as well as incorporating notions of trust, privacy and ranking for items the user has interest towards. For a complete drescription of the formalization of SMARTMUSEUM user profile, reader is advised to read D2.1 . 1. 1 The CHIP (Cultural Heritage Information Presentation) project. http://www.chip-project.org/
smar museum Profiles Qualities The quality of user profiles is a key to success of profile-based systems. From the user's point of view, there are two potential problems, according to(Cetintemel, et al., 2000). The most important one is precision problem: If a large proportion of the items that the system sends to a user are irrelevant, then the precision of the system and its performance becomes a question to user Qualities can be characterized and studied to measure the precision of the system. We have considered two main qualities for user profiles in SMARTMUSEUM scenario First is self-adaptivity and second is dynamicity We have proposed a self-adaptive profiling framework which represents user interests as a dynamic set of profile records. As pointed out previously, we presented the structure of these profiles in D2. 1. The structure of the profiles were chosen to be generic, in order to adapt themselves to dynamic changes in environment, which in our case, these changes will come from chronical changes in users' interests as well as the number of records, due to CRUd (create, read, update and delete)operations that take place on user profiled material This creates a notion of flexibility. This flexibility enables our approach to trade off effectiveness and efficiency, which in turn, enables it to be tuned based on the requirements/characteristics of our target environment. As a matter of fact, effective profile management requires techniques for representing data items and profiles, assessing the relevance of the profiles to data items, and updating the profiles based on user feedback. As stated previously, generic format of the records in which we store user data creates a felxible structure which allows dynamic information about users to be stored and retrieved Dynamic aspect of the profiling framework allows preferences to be updated and changed regularly to create a more precise model of the user's cognitive patterns and interests The framework of dynamic operation, discussed later on point outs the dynamicity of the user profiles and it shows how effective they maintain the dynamicness of the content stored and retrieved from the user profiles apprise H Heritage Malta Grant Agreement Number: FP7-216923 国如 Platform
Grant Agreement Number: FP7-216923 Acronym: SMARTMUSEUM Project title: Cultural Heritage Knowledge Exchange Platform Profiles Qualities The quality of user profiles is a key to success of profile-based systems. From the user‘s point of view, there are two potential problems, according to (Cetintemel, et al., 2000). The most important one is precision problem: If a large proportion of the items that the system sends to a user are irrelevant, then the precision of the system and its performance becomes a question to user. Qualities can be characterized and studied to measure the precision of the system. We have considered two main qualities for user profiles in SMARTMUSEUM scenario. First is self-adaptivity and second is dynamicity. We have proposed a self-adaptive profiling framework which represents user interests as a dynamic set of profile records. As pointed out previously, we presented the structure of these profiles in D2.1. The structure of the profiles were chosen to be generic, in order to adapt themselves to dynamic changes in environment, which in our case, these changes will come from chronical changes in users‘ interests as well as the number of records, due to CRUD (create, read, update and delete) operations that take place on user profiled material. This creates a notion of flexibility. This flexibility enables our approach to trade off effectiveness and efficiency, which in turn, enables it to be tuned based on the requirements/characteristics of our target environment. As a matter of fact, effective profile management requires techniques for representing data items and profiles, assessing the relevance of the profiles to data items, and updating the profiles based on user feedback. As stated previously, generic format of the records in which we store user data creates a felxible structure which allows dynamic information about users to be stored and retrieved. Dynamic aspect of the profiling framework allows preferences to be updated and changed regularly to create a more precise model of the user‘s cognitive patterns and interests. The framework of dynamic operation, discussed later on point outs the dynamicity of the user profiles and it shows how effective they maintain the dynamicness of the content stored and retrieved from the user profiles
smartmuseum Personalization recommendation and Contextualization in museum domain This section gives a thorough survey of state-of-art research in contextualization and personalization. The focus has been given to works within similar domain Overview of existing work on personalization in cultural heritage Within the domain of cultural heritage different approaches can be distinguished, some of them are more content orientated and some context oriented depending on the task to be solved Several projects are initiated to provide the best content in a certain context for users, like for providing personalized information in museums(Hyvonen, et al., 2005)(Hyvonen, 2007) (Sparacino, 2004), location-aware tourist guides(Abowd, et al. )(Cheverst, et al. )(Fink, et al., 2002)(Lam, et al., 2007) for handling personalized customer relationships(Kobsa, 2001) for providing a news program(Billsus, et al. )(Singh, et al. for managing context information in mobile devices(Korpipaa et al., 2003),etc As we could generalize the main directions of the research are the context and the content When the term ' context-aware' was first introduced (Schilit, et al., 1994)(Schilit, et al 1994)then it defined the context as location, emphasizing nearby people and objects and changes related to those objects. (Dey, et al, 2001)define context more broadly and context refers to any information that characterizes a situation related to the interaction between humans, applications, and the surrounding environment. Context is typically the location, identity, and state of people, groups, and computational and physical objects. A survey of context-aware applications is given by( Chen, et al., 2000)and a solution for supporting context aware application prototyping by the Context Toolkit is presented Day et al., 2001) To provide context specific services on the one hand context information can be used, but on the other hand users previous activities or a profile must be taken into account. To provide information in a personalized manner, personalized systems observe a user'behaviour and, based thereon, make generalizations and predictions about them (fink Kobsa, 2002). Some methods and tools must be used to generalize and predict users behaviour and also to offer personalized content Content on the web is available in different forms as text documents audio and video files mages, etc. Therefore the content can be defined content broadly as the stuff in your Web site (Rosenfeld, et al, 1998). on the web all kind of cultural content is available. Question is how to find out most relevant information sources and to provide them for the user in appropriate manner. apprise H Heritage Malta Grant Agreement Number: FP7-216923 国如 Platform
Grant Agreement Number: FP7-216923 Acronym: SMARTMUSEUM Project title: Cultural Heritage Knowledge Exchange Platform Personalization, Recommendation and Contextualization in Museum Domain This section gives a thorough survey of state-of-art research in contextualization and personalization. The focus has been given to works within similar domain. Overview of existing work on personalization in cultural heritage Within the domain of cultural heritage different approaches can be distinguished, some of them are more content orientated and some context oriented depending on the task to be solved. Several projects are initiated to provide the best content in a certain context for users, like for providing personalized information in museums (Hyvönen, et al., 2005) (Hyvönen, 2007) (Sparacino, 2004), location-aware tourist guides (Abowd, et al.) (Cheverst, et al.) (Fink, et al., 2002) (Lam, et al., 2007) for handling personalized customer relationships (Kobsa, 2001), for providing a news program (Billsus, et al.) (Singh, et al.) for managing context information in mobile devices (Korpipää et al., 2003), etc. As we could generalize the main directions of the research are the context and the content. When the term ‗context-aware‘ was first introduced (Schilit, et al., 1994) (Schilit, et al., 1994) then it defined the context as location, emphasizing nearby people and objects and changes related to those objects. (Dey, et al., 2001) define context more broadly and context refers to any information that characterizes a situation related to the interaction between humans, applications, and the surrounding environment. Context is typically the location, identity, and state of people, groups, and computational and physical objects. A survey of context-aware applications is given by (Chen, et al., 2000) and a solution for supporting context aware application prototyping by the Context Toolkit is presented (Day et al., 2001). To provide context specific services on the one hand context information can be used, but on the other hand user‘s previous activities or a profile must be taken into account. To provide information in a personalized manner, personalized systems observe a user‘ behaviour and, based thereon, make generalizations and predictions about them (Fink & Kobsa, 2002). Some methods and tools must be used to generalize and predict user‘s behaviour and also to offer personalized content. Content on the web is available in different forms, as text documents, audio and video files, images, etc. Therefore the content can be defined content broadly as the stuff in your Web site (Rosenfeld, et al., 1998). on the web all kind of cultural content is available. Question is how to find out most relevant information sources and to provide them for the user in an appropriate manner
smartmuseum The use of standardized approach makes the access to the cultural context easier and therefore some content models for semantic cultural portals have been developed(Hyvonen, 2007) Several projects have been initiated using the semantic web that are oriented to provide personalized and context-aware information for the museum visitors. Like the project MUSEUMFINLAND(Hyvonen, 2007; Hyvonen et al., 2005 ). In this project by sharing a set of ontologies, it is possible to make collections semantically interoperable, and provide the museum visitors with intelligent contentbased search and browsing services to the global collection base Another example of a personalized cultural semantic portal is the ChiP project(Aroyo, et al 2007)In the project an interactive approach is used to collect data about museum visitors in terms of their interests and preferences about artefacts from the rijksmuseum collection. This data is stored in user profiles used further to recommend routes through the museum and to guide the users towards artefacts related to their interests and preferences. The overall goal of the project is to explore different users' characteristics and personalize users museum experiences within the Rijksmuseum virtual and physical collections(Aroyo et al., 2007) And also a tourist context-aware guiding system, JJADE Free Walker, which uses Semantic Web technologies, integrates GPS, ontology and agent technologies to support location awareness for providing the precise navigation and classify the tourist information for the users ( Lam Lee, 2007) State-of-art approaches to user profile learning and construction In this section(Kirt, 2008)the methods used in different content and context based applications are introduced. Depending on the problem to be solved there are used a number of method We have picked out five most referred methods that can also be used in our project The five groups of methods and their applications are as follows Data Mining-mining association rules between sets of items in a large database of customer transactions(Agrawal, et al., 1993)(Agrawal, et al., 1994), data mining for web personalization(Mobasher, et al., 2000): Conceptual Clustering-a web document clustering algorithm-WebDCC( Godoy, et al., 2006), adaptive Web sites(Perkowitz, et al. ) and user'interests estimator(Kim,et al,2008) Naive Bayes-identifying interesting Web sites to a user(Billsus, et al., 1999), a model for news story classification, named News Dude(Billsus, et al., 1999), developing context-aware mobile application( Korpipaa, et al., 2003)(Korpipaa, et al., 2003); apprise H Heritage Malta Grant Agreement Number: FP7-216923 国如 Platform
Grant Agreement Number: FP7-216923 Acronym: SMARTMUSEUM Project title: Cultural Heritage Knowledge Exchange Platform The use of standardized approach makes the access to the cultural context easier and therefore some content models for semantic cultural portals have been developed (Hyvönen, 2007). Several projects have been initiated using the semantic web that are oriented to provide personalized and context-aware information for the museum visitors. Like the project MUSEUMFINLAND (Hyvönen, 2007; Hyvönen et al., 2005). In this project by sharing a set of ontologies, it is possible to make collections semantically interoperable, and provide the museum visitors with intelligent contentbased search and browsing services to the global collection base. Another example of a personalized cultural semantic portal is the CHIP project (Aroyo, et al., 2007) In the project an interactive approach is used to collect data about museum visitors in terms of their interests and preferences about artefacts from the Rijksmuseum collection. This data is stored in user profiles used further to recommend routes through the museum and to guide the users towards artefacts related to their interests and preferences. The overall goal of the project is to explore different users' characteristics and personalize users' museum experiences within the Rijksmuseum virtual and physical collections (Aroyo et al., 2007). And also a tourist context-aware guiding system, iJADE FreeWalker, which uses Semantic Web technologies, integrates GPS, ontology and agent technologies to support location awareness for providing the precise navigation and classify the tourist information for the users (Lam & Lee, 2007). State-of-art approaches to user profile learning and construction In this section (Kirt, 2008) the methods used in different content and context based applications are introduced. Depending on the problem to be solved there are used a number of methods. We have picked out five most referred methods that can also be used in our project. The five groups of methods and their applications are as follows: Data Mining–mining association rules between sets of items in a large database of customer transactions (Agrawal, et al., 1993) (Agrawal, et al., 1994), data mining for web personalization (Mobasher, et al., 2000); Conceptual Clustering–a web document clustering algorithm-WebDCC (Godoy, et al., 2006), adaptive Web sites (Perkowitz, et al.), and user' interests estimator (Kim, et al., 2008). Naive Bayes–identifying interesting Web sites to a user (Billsus, et al., 1999), a model for news story classification, named News Dude (Billsus, et al., 1999), developing context-aware mobile application (Korpipää, et al., 2003) (Korpipää, et al., 2003);