Personalizing Access to Cultural Heritage Collections using Pathways


Published on

Paul Clough, Nigel Ford and Mark Stevenson.
13 February 2011
Published at PATCH 2011

Published in: Technology
  • Be the first to comment

  • Be the first to like this

Personalizing Access to Cultural Heritage Collections using Pathways

  1. 1. Personalizing Access to Cultural Heritage Collections using Pathways Paul Clough Nigel Ford Mark Stevenson University of Sheffield University of Sheffield University of Sheffield Sheffield, S1 4DP, UK Sheffield, S1 4DP, UK Sheffield, S1 4DP, UK which are not well supported in standard search interfaces.This paper discusses mechanisms for personalizing access Recent trends in information access services haveto cultural heritage collections and suggests that paths or recognized the necessity of providing support for moretrails are a flexible and powerful model for this and could exploratory and serendipitous search behaviours if serviceslink with existing models of cognitive information are to be effective in helping users with discovering andbehaviour. We also describe a European project called assimilating knowledge [11, 12, 13].PATHS (Personalized Access To cultural Heritage Spaces)that aims to support information exploration and discovery This paper suggests the metaphor of “paths” through athrough digital cultural heritage collections. This project collection as a powerful and flexible model for navigationaims to implement the user models discussed in this paper that can enhance the user’s experience of cultural heritageand provide a mechanism for users to create and share collections and support them in their learning andpathways through information spaces for learning and information seeking activities. It also describes a newknowledge discovery. Personalized access to digital cultural project funded by the European Union, PATHS1heritage resources will be provided by adapting suggested (Personalized Access To cultural Heritage Spaces), thatroutes to the requirements of individual users and groups, uses this model to provide users with innovative ways tosuch as students/teachers, professional archivists and access and utilize the contents of digital libraries that enrichhistorians and scholars. their experiences of these resources. The PATHS project aims to create a system that acts as an interactiveAuthor Keywords personalized tour guide through existing digital libraryAdaptive systems, personalization, paths, user models, collections, offering suggestions about items to look at andcultural heritage resources, social navigation. assist in their interpretation by providing relevant contextual information from related items within specificINTRODUCTION collections and items from external sources.Significant amounts of cultural heritage material are now The remainder of this paper is organized as follows. Weavailable through online digital library portals. However, begin with a description of existing approaches that havethis vast amount of cultural heritage material can also be been used to personalize navigation systems, includingoverwhelming for many users who are provided with little details of ways in which cognitive styles of users can beor no guidance on how to find and interpret this modeled. We then introduce the notion of a path through ainformation. Potentially useful and relevant content is collection which we suggest as a useful model ofhidden from the users who are typically offered simple navigation. Finally, we describe the PATHS project whichkeyword-based searching functionality as the entry point will make use of this model and apply it to a large on-lineinto a cultural heritage collection. The situation is very cultural heritage collection.different within traditional mechanisms for viewing culturalheritage (e.g. museums) where items are organized BACKGROUNDthematically and users guided through the collection. Usersof cultural heritage portals have diverse information needs Cultural Heritage and the User Experienceand exhibit highly individualistic information seeking Cultural heritage institutions hold an enormous and richbehaviours (e.g. information encountering and foraging) variety of digital content covering a broad range of subjects: natural history, ethnography, archaeology, historic monuments, fine and applied arts which often cross nationalPermission to make digital or hard copies of all or part of this work for and linguistic boundaries. There is strong motivation topersonal or classroom use is granted without fee provided that copies arenot made or distributed for profit or commercial advantage and that copies bring together content from different cultural institutionsbear this notice and the full citation on the first page. To copy otherwise, into aggregated portals, which have typically offered accessor republish, to post on servers or to redistribute to lists, requires priorspecific permission and/or a fee.PATCH 2011. 1Copyright 2009 ACM 978-1-60558-246-7/09/04...$5.00.
  2. 2. services based on traditional catalogues used in libraries, system in iClass, which is a framework of services designedmuseums and archives. Search services have been geared to support teachers and learners in providing personalizedtowards subject specialists and experienced users. Yet the learning experiences. In terms of standards, OWL is usedenvironment in which users and digital library services are for domain ontologies; IMS Learning Design (IMS LD) foroperating has changed. People come to digital libraries with structuring learning activities; IMS LIP for learner profiles;experience of using the web and with new expectations [8]. and SCORM for learning object manifests. Like APeLS, iClass separates the generation of personalized conceptualCultural institutions wish to be able to offer users of their learning paths from specific resources that can instantiateportals with an experience that is continuous with the way such paths. However, an important way in which iClasspeople experience the web. They are seeking to enable differs from APeLS is that pedagogy is explicitlyricher user experiences that support connectivity between represented in a separate model distinct from domainpeople, content and applications, to support writers as well knowledge. The representation of pedagogies in a separateand readers, and to enable collaborations with and between model means that they can be reused.users. A new generation of cultural portals is encouraginguser participation by offering people with opportunities to Castillo, Gama and Breda [7] report GIAS, a system alsointeract with content (for example encouraging them to tag designed to select learning resources from a repositoryresources) and to make recommendations to other users [6, which are appropriate to individual learners’ current states19]. of knowledge and their preferred learning styles. However, the system differs from others in that from a baseline initialInstitutions are looking for ways to recreate in the digital psychometric assessment of learning style, it fine-tunes itsinformation space the opportunities that visitors to libraries, model of each learner’s style in response to feedback frommuseums and archives have of sharing books, objects and interactions between the learner and learning resources.ideas with each other. They would like to be able to This on-going adaptation is necessary because of inherentpersonalize the experience for their digital library users, for uncertainty in any psychometric assessment of learningexample suggesting content that is more likely to be of style, and also possible drift in learners’ preferences overinterest based on their profile and highlighting associations time and as a result of interactions with resources.between related items [1]. But cultural heritage portals lackthe massive volumes of users and interactions that are A number of systems are designed to enable resourcerequired to enable the effective analysis of user behaviours, discovery not only of materials stored in specialisedusage patterns and interests. Issues related to user- repositories, but also open corpus material available on theadaptivity such as controllability, trust, privacy, Web. Dolog et al. [10], for example, report development ofobtrusiveness, predictability, transparency and breath of the “Personal Reader”, which is also based on metadata-user’s experience are also important [14]. based reasoning mechanisms. From the starting point of a learning resource being studied by a learner, the “PersonalAdaptive Personalized Systems Reader” discovers resources that are related to the currentlyA number of adaptive personalized systems have been viewed resource. It can recommend resources whichdeveloped, particularly to support learning. The Adaptive provide a different perspective on the topic – for example, aPersonalized eLearning Service (APeLS) reported by summary, more general or more specific material, orConlan et al. [9] can take metadata describing an examples illustrating the concepts involved. The authors’individual’s learning needs, prior competences and personal aim is to work towards the vision of an adaptive webcharacteristics, and construct a personalized course by capable of leveraging open corpus material [3, 4].discovering and sequencing appropriate learning objectsdiscovered within one or more distributed specialist Social Navigation Systemsrepositories. Input to a rule-based adaptive engine includes A number of social navigation systems have beenmetadata relating to the learner, content in the form of developed that attempt to harness the collective knowledgelearning resources, and narratives. A narrative is a of their user communities. As noted by Brusilovsky et al.description of the required conceptual pathway for a [5], such systems use a variety of techniques, which theyparticular learner, and is built prior to searching for categorize in terms of:particular learning resources, taking account for example ofprerequisite concepts and their sequencing. Thus, the • Types of past user behaviour recorded;system’s narrative model refers to concepts rather than • How these data are used to generate collectivespecific resources and can be made to model the learner’s knowledge;prior knowledge and preferences. • How this collective knowledge is used in order to benefit users in terms of their information accessing.O’Keeffe et al. [16] note that although early adaptivehypermedia systems personalized learning according to Brusilovsky et al. describe the use of social navigation inlearners’ prior knowledge, goals and preferences, they did digital libraries as implemented in the Ensemble system.not explicitly address pedagogy. They report the Users can follow “trails” of library items which are createdincorporation and further development of the APeLS on specific topics in the manner being adopted in our work.
  3. 3. An early system for the development of educational Insofar as decisions are made as to which user behaviourssystems based on an “ecological” paradigm is reported by and inputs are recorded and used as input to an adaptivityMcCalla [2]. This is an approach to providing adaptive engine, such selection is based on at least an implicit modelpersonalized learning which not only makes use of standard or theory of which factors are likely to be most influentialpedagogical metadata, but also dynamically collects and in the calculation of appropriate adaptive system behaviour.cumulates other metadata relating to resources, learners, A criticism of a number of the explicit or implicit userand the interactions between the two, at the time of use. models employed by a number of existing adaptive systemsThe approach also entails the mining of this data in order to is that the models selected are not those which are arguablygenerate new pedagogic knowledge as required for different best suited to drive adaptivity relating to the navigation ofpurposes. Metadata is gathered at the time of use of the information spaces.resource as opposed to being pre-assigned by human Although there are many models of both informationexperts. As a learner accesses a particular resource, seeking and learning via pedagogical mediation, few mapmetadata is attached to that resource, which relates to the directly onto the type of choices most relevant to anlearner, the resource and the interaction between the two adaptive system designed to advise on paths through a(e.g. dwell time on a resource). Over time, each learning complex set of information sources. Of particular relevanceresource will accumulate many models. This data can then here is the work of Pask and Witkin [21, 22], the models ofbe subjected to data mining in order to discover patterns whom have been extensively studied for almost 40 years.that are useful in achieving particular tasks. This (and related) work suggests that cognitive processingTang and McCalla, [18] report a system based on this takes place across two major orthogonal dimensions (Figureapproach, designed to search Citeseer and recommend 1).relevant papers to research students. Each time a paper isread by a learner, an instance of the learner’s model is Autonomousattached to the paper’s metadata. The system also enablesthe learner to annotate the resource, this informationfeeding in to model (recording the learner’s interactionswith resources). Work is reported in further developingapproaches to the learner modelling that is a key feature of Local Globalecological systems [15, 20]. (analytic)USER MODELSSome existing systems employ explicit user models to driveadaptivity. Such models may entail, for example, levels of Dependentexpertise or cognitive styles. Others make use of recorded Figure 1. Key cognitive dimensions.user behaviour without any explicit underlying cognitivemodels. The work suggests that:User models may be utilized in both inductive and • Different individuals may have different predominantdeductive approaches to system design. An inductive navigational styles. These appear to be linked to moresystem is defined here as one in which the model driving fundamental cognitive styles. High academic achieversadaptivity does not derive from any pre-existing user as well as less academically achieving people may stillmodel, and is created by analyzing user behaviours and have a predominant style. These navigation stylesinputs. A deductive system is one in which an existing user translate into different paths.model is used to generate adaptive behaviour. Deduction • Adopting a navigation path that matches one’sentails using pre-existing user models to driver adaptive predominant style can influence the effectiveness of thebehaviour (e.g. recommended paths and next moves). resultant learning. Where an individual navigates usingInduction entails gathering and using data generated by a path that mismatches their predominant style theirusers (the navigational behaviour they exhibit, and the learning may be disrupted; matching a navigationalcomments/reactions/annotations/tags they provide relating path may enhance learning (for a given learning task).to nodes and paths). However, these results have been found in exper-These categories are not mutually exclusive: the two may imental rather than more natural learning used interactively – to begin with parameters derived • Navigation paths adopted by individuals may also varyfrom known user models are likely to be effective, and according to their level of subject expertise in the areaallow the refinement and modification of these models via being from user behavior and inputs during system use, for • Different paths may also be more, and less appropriateexample user-system interactions recorded in transaction for achieving different types of goal/task (e.g. relativelylogs. convergent fact-finding versus more divergent creative exploration).
  4. 4. • Individuals also vary in the extent to which they thrive approach. S/he will have a high tolerance of uncertainty in navigational conditions characterised by external and ambiguity, and will – relative to his or her local mediation (guidance) versus autonomy. This difference counterparts – be happy to spend quite a long time in appears to be linked to fundamental cognitive style. tentative exploratory uncertain mode. The more extreme global person is likely to be a divergent (creative) thinkerTable 1 shows characteristics, as identified in this work, of and will want to be stimulated by new connections. S/hethe horizontal (local/global) dimension in terms of a will want to explore widely spaced nodes – not keep tonumber of aspects of cognitive processing relevant to users’ close and strictly logical connections. S/he may be lookingnavigation of a semantic space. for new ideas/directions “off the beaten track”. Local (analytic) Local (analytic) Global The stereotypical local (analytic) user will be concerned to Learning/problem-solving goals learn in a step by step way, mastering one aspect of the Convergent goals. Divergent goals. topic (node) in depth before going on to the next. The “big “Find an answer”. Creatively explore. picture” will tend to emerge relatively late in the learning Learn pre-defined content. Come up with new ideas. Process goals process. S/he will find distracting information which the Concerned with procedures Concerned with conceptual global information processor might consider valuable and vertical deep detail overview and horizontal enrichment material that would enhance learning. S/he is (procedure building). broad inter-relationships likely to be a relatively convergent thinker who knows what (description building). s/he wants and does not want to be distracted by anything Navigation styles “irrelevant”. S/he will want to explore step by step in a Serialist navigation style Holist navigation style “one thing at a time” mode, with a relatively narrow focus. Narrow focus. Broad global focus. At any given point in time, the optimal next node to be One thing at a time. Many things on the go at explored will be local (the next logical step). S/he will have the same time. a low tolerance of uncertainty and ambiguity, and will not Short logical links between Rich links between nodes. nodes. like to spend long in tentative exploratory uncertain mode. Intolerance of strictly Welcoming of enrichment Dependent irrelevant material. (but strictly irrelevant) The dependent learner may be new to the subject area, and material. may want to follow a path (whether global or local) that has Finish with one topic before Layered approach returning been approved and has been pre-trodden by experts and going on to the next. to nodes at different level of detail. other authoritative figures. Positive learning outcomes Autonomous Good grasp of detailed Well developed conceptual The more autonomous learner may want to beat his/her own evidence. overview. exploration path, and may want to (make and possibly be Deep understanding of Broad inter-relationship of stimulated to make) more original less conventional individual topics. ideas. connections between nodes. In-depth understanding of the Good grasp of the “big parts. picture”. PATHS AS A METAPHOR FOR NAVIGATION Characteristic learning pathologies The notion of a “path” or “trail” through digital library Poor appreciation of topic Poor grasp of detail. collections provides a flexible model of navigation that can inter-relationships. provide a powerful model onto which various levels of Failing to see the “big Over-generalisation. personalisation can be added. In hypertext systems, the picture”. notion of guided tours and trails refer to scenarios when users are introduced to an unfamiliar subject and activities Table 1. Characteristics of local and global information carried out with individual steps [17, 22]. processing styles. A path is defined here as a route through a semantic space.Scenarios of use In this case, the semantic spaces are defined digital collections of heritage resources. A “route through” aGlobal semantic space only makes sense in terms of a sequence ofThe stereotypical global user will want to explore a new processing of connected components of that semantic space.topic broadly at first, first establishing a relatively tentative/ The components of semantic space that enable sequentialprovisional conceptual overview before going back to processing are defined here as nodes. The semantic space isnodes and filling in details. S/he will take a broad focus, navigable by the sequenced selection of connected nodes.sampling nodes at different levels taking a “layered”
  5. 5. 1 2 10 3 11 4 7 14 5 6 9 12 13 15 16 8Figure 2. Stereotypical “local/analytic” path. Characteristics: small steps between nodes in the sequence, mainly depth-first;infrequent re-visiting of nodes. Logical “one step at a time” brick-by-brick approach. 1 2 6 3 7 10 11 18 12 14 9 13 4 16 19 15 5 8 17 20Figure 3. Stereotypical “global” path. Characteristics: large steps between nodes; significantly breadth-first; frequent re-visiting ofnodes attending to different aspects in a layered “parallel processing” approach. Frequent accessing of “enrichment” material ifoffered at any node.Nodes for a resource or information item might contain the • Information on available paths in which the currentfollowing information: resource forms a node.• System-provided subject metadata (subject keywords) Nodes may be connected in a number of different ways. about this resource; They may be pre-defined or computed on the fly. They• User-provided subject metadata (tags) for this resource; may be defined by the system/designers and/or by users –• Broader, narrower and related terms according to the individual or aggregated. A node consists of items of default (“authority”) subject ontology for the topics information aggregated and represented by means of one or that this resource is about; shared tags (metadata). A route is a sequence in which one• Recommended “next” nodes(s) according to the or more nodes are attended to. Attending to a node may pedagogical ontology for the topic(s) the resource is entail intellectual processing of different types and at about; different levels. Thus a user may return to previously• Information about the resource at abstract level (e.g. “attended to” nodes and attend to it in a different way overview, summary, image thumbnail etc.); and/or at a different level in order to increase (clarify/• Inter-relationships showing how the information stored modify/ deepen/ modify/ refute/ etc.) his or her at this node inter-relates with other nodes to form the understanding. bigger picture; Nodes for a resource may link to a variety of items:• Link(s) to the detailed contents of this resource (e.g. full text, detailed images etc.); • Subject ontologies, e.g. a generally agreed default• Information on which ontologies are available for the “authority” ontology; topic(s) this resource is “about” – both subject and • Pedagogical ontologies that contain information on, pedagogical; e.g. which topics are pre-requisite for the• User annotations abut this resource (this user and understanding of others; others); • User metadata relating to goals, preferred navigation• Information on paths that are available relating to the styles, levels of topic knowledge etc.; topic(s) this resources is “about” – both subject and • Information on interactions in which this node has been pedagogical; involved including user goals in using it, its part in any paths, user annotations (including user-perceived
  6. 6. usefulness of this node and ultimate success of the path Different users will have differing needs from pathways and of which is formed a part); an important aspect of the research will be to build up• Information on the user’s current path; knowledge and understanding of user profiles, their needs,• Information on other available paths, with their own interests and preferences and feed this into the system metadata for describing for whom/which learning development. It is expected that both explicit information goals/at which level they may be appropriate.  input from the user will be utilized (e.g. in the form of a profile to be completed) and implicit information fromThere are potentially alternative routes to the same users’ interaction (e.g. from transaction log files) will alsodestinations. A destination is defined here in terms of a shape their information seeking experience.user’s desired state of knowledge or understanding. This THE PATHS PROJECTmay be defined by the user and/or someone else (for The PATHS project aims to make use of current knowledgeexample, a teacher and/or subject expert). Different users of personalisation to develop a system for navigatingmay have different starting points for similar destinations. cultural heritage collections that is based around theBy way of example, Figures 2 and 3 illustrate stereotypical metaphor of paths and trails through them. The PATHSlocal and global pathways through a set of nodes that are system will provide an adaptive and rich information-arranged in a hierarchical order, which represent a typical seeking encounter for the user. It system will make user-hierarchical subject representation, with top level topics specific recommendations about items of potential interestsubdivided into lower order subordinate topics. The as individuals navigate through the collection. The user willnumbers represent the order in which the imagined “local” be offered links to information both within and outside theand “global” users might navigate the nodes. It is proposed collection that provide contextual and backgroundthat: information, individually tailored to each user and their• Users can construct their own paths (“independent context. paths”) which can be saved for future reference, edited The PATHS project will take a user-centred approach to or shared with other users. These paths will be more development by bringing users into the research cycle from than a simple list of items in a collection that the user the beginning of the project, gathering their input at all has visited; they will also contain information about the stages in the development on how it can help to meet their links between the items, details of others items needs and feedback on the functionality as prototypes are connected to them and connections to information, field-tested. The PATHS project consists of several both within and outside the collection that provides separate, but connected, packages of work, including the context. following:• Groups of users can work collectively to create paths (“collaborative paths”), adding new routes of discovery • Gathering user requirements and creating functional and annotations that can build upon the contributions specifications from a broad range of users including made by others. those belonging to different groups, e.g. students,• User-generated paths will exist as information objects family historians and photographers and of different in their own right: they can be indexed, organised and types, e.g. learning styles and needs from Cultural shared with others, and will be potential learning Heritage collections. These requirements will be used objects that can be offered to users alongside the to develop a functional specification for the systems cultural heritage content. developed during the project. These requirements will build upon those identified in previous work for• Users can also follow pre-defined paths (“guided cultural heritage information access systems [24, 25]. paths”) created by domain experts, such as scholars or teachers. Guided paths provide an easily accessible • Processing cultural heritage content and enriching it entry point to the collection that can be either followed through identifying connections between items within a in their entirety or left at any point to create an collection and complementing connections with independent path. Guided paths can be based around existing relations and providing links to material both any theme, for example artist and media (“paintings by within and external to the collection that provides Picasso”), historic periods (“the Cold War”), places background information (e.g. to Wikipedia). (“Venice”), famous people (“Muhammed Ali”), • Designing effective user interfaces through which emotions (“happiness”), events (“the World Cup”) or users will interact with the PATHS system. These any other topic (e.g. “Europe”, “food”). interfaces will provide users with personalised• Users’ paths can be private and only accessible to them navigation through Cultural Heritage collections that is or made public and stored in a common (searchable) enriched with the additional information added through repository of organised paths on different themes and processing the digital content. The user interface will subjects. allow users to follow pathways created by other users and to share their own. This will build on previous work on personalisation in museums and digital libraries [26, 27].
  7. 7. • Designing evaluation methodologies and conducting seeking tasks. We suggest that the notion of paths through of field trials to assess the performance (effectiveness, collections forms a powerful metaphor onto which efficiency and satisfaction) of the systems implemented personalisation can be added that will map onto existing in PATHS in realistic scenarios. Evaluation will models of user’s information behaviour. Pathways can be culminate in field trials in end-user scenarios. used to support various styles of cognitive information Particular focus will be on evaluating users’ search processing which will surface as different routes users may sessions2 and the value of paths generated by the user. take through an information space. Offering users suggested Also, focusing on the evaluation of browsing routes or paths through an information space will locate and techniques will form part of this research. associate information that will ultimately help them to fulfill broader activities, such as constructing knowledgeThe vision is to build a system that: around a given subject or theme.• Exploits existing knowledge of users to optimise the effectiveness of interacting with digital heritage The PATHS project aims to investigate and implement resources. pathways in a naturalistic setting for a range of users and• Enables the testing and refinement of such knowledge. groups that regularly make use of cultural heritage• Enables new knowledge to be discovered. information. A large-scale operational system will developed for navigating on-line cultural heritageThis system will provide personalized access to resources collections in a more effective manner than currentby adapting suggested routes to the personalized searching functionalities. Pathways will be used to guiderequirements of individual users and groups. It will seek to: and assist individuals and user communities with• Respond to users in a cognitively ergonomic way – i.e. information discovery and exploration within cultural by matching navigation to a learner’s preferred style heritage information spaces for learning and information and minimising any mismatch and consequent seeking activities. This will support multiple information additional cognitive processing load. In this way, the seeking behaviours and enhance the user’s information learner will find exactly what s/he wants with the least access experience of digital library resources. effort. Navigation entails travelling the shortest path ACKNOWLEDGMENTS between starting point and desired end point. The research leading to these results has received funding• Challenge and “stretch” the user by via controlled and from the European Communitys Seventh Framework constructive mismatching. In this way, learners may Programme (FP7/2007-2013) under grant agreement n° develop increasing autonomy and versatility – i.e. the 270082. We acknowledge the contribution of all project ability autonomously to thrive in information partners involved in PATHS (see: http://www.paths- environments not necessarily matched to their own preferred style. PATHS will also explore the extent to which users may be encouraged and helped to engage REFERENCES in cognitive processing in which they are less strong. 1. Bowen, J P and Filippini-Fantoni, S (2004) For example, the extreme divergent thinker may “Personalization and the Web from a Museum usefully be encouraged (in certain learning Perspective”, In D Bearman and J Trant (eds.), circumstances) not to underplay complementary Museums and the Web 2004: Selected Papers from an convergent processing. Cognitive research suggests International Conference, Arlington, Virginia, USA, 31 that s/he may, without such complementary processing, March – 3 April. Archives & Museum Informatics, 63– exhibit over-generalisation and lack of grasp of detail. 78. Conversely, the strong convergent thinker may be 2. Brooks, C. and McCalla, G. (2006). “Towards flexible encouraged to explore and think more divergently learning object metadata”, International Journal of (creatively) to avoid fragmented learning and failing to Continuing Engineering Education and Life Long see the wood for the trees. PATHS will explore the Learning, 16(1-2), 50-63. potential of suggesting sub-optimal, but constructive paths to users. 3. Brusilovsky, P. (2001). “Adaptive hypermedia” User Modeling and User-Adapted Interaction, 11 (1-2), 87-CONCLUSIONS 100.With growing amounts of digital cultural heritage 4. Brusilovsky, P., & Maybury, M. (2002). “From adaptiveinformation becoming available online users will require hypermedia to the adaptive web” Communications of theassistance, particularly for more exploratory information ACM, 45 (5), 30-33. 5. Peter Brusilovsky, Lillian N. Cassel, Lois M. L. Delcambre, Edward A. Fox, Richard Furuta, Daniel D.2 We began work on the evaluation of search sessions through organising a Garcia, Frank M. Shipman III, Paul Logasa Bogen II,session at the US Text REtrieval Conference (TREC) in 2010 called the Michael Yudelson: Enhancing Digital Libraries withSession Track:
  8. 8. Social Navigation: The Case of Ensemble. ECDL 2010: experiences”, International Journal on E-Learning, 116-123. 5(1), 113-127.6. Carmagnola, F., Cena, F., Cortassa, O., Gena, C., and 17. Reich, S. F., Carr, L., De Roure, D. and Hall, W. (1999). Torre, I. (2007) “Towards a Tag-Based User Model: “Where have you been from here? Trails in hypertext How Can User Model Benefit from Tags?” In systems”, ACM Computing Surveys, 31(4). Proceedings of the 11th International Conference on 18. Tang, T. and McCalla, G. (2003) “Smart User Modeling (Corfu, Greece, July 25 - 29, 2007), 445- recommendation for an evolving elearning system” In 449. Proceedings of the Workshop on Technologies for7. Castillo, G., Gama, J. & Breda, A.M. (2007). “An Electronic Documents for Supporting Learning ,at the Adaptive Predictive Model for Student Modeling, in International Conference on Artificial Intelligence in Online and Distance Learning: Concepts, Education (AIED 2003), Sydney, Australia. Methodologies, Tools and Applications”, Volume III, 19. Trant, J. (2009). “Tagging, Folksonomies and Art Chapter 2.43, Information Science Reference Pub. Museums: Early Experiments and Ongoing Research”8. Connaway, L. S. (2007) “Mountains, valleys, and Journal of Digital Information, 10(1). pathways: Serials users’ needs and steps to meet them. 20. Vassileva, J., McCalla, G., and Greer, J. (2003). “Multi- Part I: Identifying serials users’ needs: Preliminary agent multi-user modeling in I-Help” Journal of User analysis of focus group and semi-structured interviews Modeling and User-Adapted Interaction, 13, 179-210. at colleges and universities,” Serials Librarian, 52(1/2), 223-236. 21. Pask, G. (1976) “Styles and Strategies of Learning”, British Journal of Educational Psychology (46:II), 128–9. Conlan, O., Hockemeyer, C., Wade, V., Albert, D. 148. (2003). “Metadata driven approaches to facilitate adaptivity in personalized elearning systems”, Journal 22. Witkin, H.A., Moore, C.A., Goodenough, D.R., and of the Japanese Society for Information and Systems in Cox, P.W. (1977) “Field Dependent and Field- Education Independent Cognitive Styles and Their Education Implications”, Review of Educational Research, 47(1),10. Dolog, P., Henze, N., Nejdl, W. and Sintek, M. (2004). 1- 64. “The Personal Reader: Personalizing and Enriching Learning Resources using Semantic Web Technologies” 23. Wexelblat, A. and Maes, P. (1999) “Footprints: History- In Proceedings of AH2004 - International Conference Rich Tools for Information Foraging”, CHI 1999: 270- on Adaptive Hypermedia and Adaptive Web-Based 277. Systems, Eindhoven, The Netherlands. 24. Minelli, S. H., Marlow, J., Clough, P., Cigarran11. Eaglestone, B., Ford, N., Brown, G. and Moore, A. Recuero, J.M., Gonzalo, J., Oomen, J. and Loschiavo, (2007) “Information systems and creativity: an D. (2007) “Gathering requirements for multilingual empirical study” Journal of Documentation, 63(4), 443- search of audiovisual material in cultural heritage”, In 464. Proceedings of Workshop on User Centricity - state of the art (16th IST Mobile and Wireless Communications12. Foster, A. and Ford, N. (2003) “Serendipity and Summit), Budapest, Hungary, 1-5 July, 2007. information seeking: an empirical study” Journal of Documentation, 59(3), 321 – 340. 25. Clough, P., Ireson, N. and Marlow, J. (2009) “Extending Domain-Specific Resources to Enable Semantic Access13. Ford, N. (1999) “Information retrieval and creativity: to Cultural Heritage Data”, Journal of Digital towards support for the original thinker” Journal of Information, Volume 10(6). Available online: Documentation, 55(5), 528-542. Jameson, A. (2003). “Adaptive interfaces and agents”, 26. Bonis, B., Stamos, J., Vosinakis, S., Andreou, I. and In: Jacko, J., Sears, A. (eds.) The Human-Computer Panayiotopoulod, T. (2009) “A platform for virtual Interaction Handbook: Fundamentals, Evolving museums with personalized content”, Multimedia tools Technologies and Emerging Applications, pp. 305–330. and applications, 42(2), 139-159. Erlbaum, Mahwah, NJ. 27. Schmitz, P.L. and Black, M.T. (2008) “The Delphi15. Kumar, V., Greer, J. and McCalla, G. (2005). toolkit: enabling semantic search for museum “Assisting online helpers”, International Journal of collections”, In: J. Trant and D. Bearman (eds): Learning Technology, 1, (3), 293-321. Museums and the Web 2008: Proceedings of the16. O’Keeffe, I., Brady, A., Conlan, O. and Wade, V. International Conference for Culture and Heritage (2006). “Just-in-time generation of pedagogically Online 2008. Montreal, Quebec, Canada. Available sound, context sensitive personalized learning online: schmitz/schmitz.html.