Pathways to Discovery: Supporting Exploration and Information Use in Cultural Heritage Collections
Paula Goodale, UK, Paul Clough, UK, Mark Hall, UK, Mark Stevenson, UK, Kate Fernie, UK, Jill Griffiths, UK, eneko agirre, Basque Country
In this paper we compare the results of the user-centred evaluation of two iterations of the PATHS system, which aims at supporting exploration, navigation and use of information in cultural heritage online collections. We focus on two path creation exercises, and examine the format and content of the paths according to available functionality and different user attributes. We also present findings of the users’ interactions and response to the path concept as a means of discovery, and of meaning-making and information re-use in relation to curatorial, research, education and leisure contexts. Recommendations are made for further refinement and development of the path functionality, and for application within CH online collections, with regard to key use case scenarios.
Keywords: exploration, interaction, curation, online collections, trails, evaluation
Large digital library collections in cultural heritage (CH) and other domains can be difficult to navigate for novice users lacking in specialist knowledge of the content and structure of the collection being accessed. This issue is compounded when collections are aggregated (for example, in the case of Europeana (http://www.europeana.eu), bringing together diverse content with heterogeneous forms of metadata. Further, there is commonly a lack of information seeking support tools in CH collections, meaning that it can be difficult for users to extract information and to use it for their own research purposes and/or creative endeavours. The PATHS Project (http://www.paths-project.eu) addresses both of these problem areas, developing a novel interface for CH collections that is intended to support information exploration, discovery and use through a variety of information visualisation modes. In particular, the interaction device of a ‘path’ or trail is offered, both as a means of navigation (e.g. as a guided tour of a collection or topic area), and for use in meaning-making, interpretation, and information sharing.
At the conceptual level, a path is a device for ordering, connecting and annotating a series of items of interest that have been collected, in this case in a CH digital library collection. It can be created by an expert or novice user, as a means of communicating and sharing interpretations and ideas relating to the content it contains, and may also be contextualised with related and background information to aid further exploration away from the path, or ‘off the beaten track’.
2. Paths and Trails
Trails and guided tours have a strong currency in the physical CH environment, and can be found in the form of audio and interactive guides, lesson plans and activity trails, as well as more general routes through exhibitions. Research into paths and trails in hypertext and search systems has a long history, both as a navigational or learning aid and as a means of tracing where users have been (e.g. Furuta et al, 1997; Reich et al, 199; Shipman et al, 2000; White and Huang, 2010). More recently, paths have been used to connect diverse Web pages, particularly in educational contexts (see for example, Walden’s Paths – Shipman, et al, 2000). In more specialist collections where some level of collection or subject knowledge is required to formulate effective search queries, paths can also provide a useful aid to navigation for novice users (Wilson et al, 2010).
Making connections between items (e.g. as paths) has been suggested as a means of surfacing narratives through the rigid database structures that often underlie online cultural collections, thereby supporting a more cohesive interaction experience. Opportunities to develop multiple narratives through the same information space may arise, allowing for different interpretations to co-exist (Manovich, 1999). Narratives enable meaning to be made when links between people, artefacts, ideas and interpretations are made (Walker et al, 2013), supporting deeper engagement, leaning, creativity and exploration, as well as representing the collecting process (Mulholland and Collins, 2002), and are thus a key element of engagement in cultural heritage. Further, Peterson and Levene (2003) relate implicit trails, created by users as they navigate through physical and virtual museum spaces, to the learning process.
Projects that have incorporated paths as a primary or secondary aid for learning, navigation or creativity in cultural heritage have included StoryBank (Frohlich and Rachovides, 2008), the First World War Poetry Digital Archive (http://www.oucs.ox.ac.uk/ww1lit/), Cultura (Agosti, et al, 2013), and PATHS (Fernie. et al, 2012). In other work, paths have been generated automatically based upon semantic links within the collection (e.g. Makela, et al, 2007; Zdrahal, et al, 2008).
3. Development of the PATHS Prototypes
The PATHS project aims to address two common information access issues in cultural heritage online collections: first, a lack of subject knowledge in novice users, combined with more exploratory, unfocused information seeking modes that are common in this environment, typified by models of such as information foraging (Pirolli and Card, 1995), berry-picking (Bates, 1989), and information encountering (Erdelez, 1997). Second, PATHS aims to support wider information processes that are often overlooked, incorporating elements of discovery, validation and interpretation, and information use, as exemplified in the information journey model (Adams & Blandford, 2005).
The system design is informed by a high-level interaction model which covers five core activities: developing a Concept; searching, exploring and Collecting items of interest; using the items to Create meaning and to reuse in new information objects (paths); Communicating about items found and paths created; and, accessing or Consuming the information contained within the collection, both individual items and paths. PATHS therefore supports unfocused information seeking via browsing and exploration, and incorporates tools for collecting items and creating paths.
Content within the two prototypes comprises a sub-set of the English language collections held in Europeana, with around 500,000 items available for exploration, mainly from smaller and medium-sized UK-based cultural heritage institutions. The PATHS content comprises a thumbnail image, title, description, selected metadata, and links to the original source (all where available), plus additional data enabled via content enrichment processing, including internal related item links and external links to related content in Wikipedia.
To date, PATHS has been through two cycles of user-centred design, resulting in two iterations of a prototype system. At each stage we have collected comprehensive user requirements, which have informed the interface designs and functional specifications. We have then followed the prototype implementations with comprehensive user studies for evaluation of the system usability, functionality offered, and usefulness with regard to typical user tasks.
3.1 User Requirements for Paths in Cultural Heritage
Research on the typical nature and applications of paths in CH was carried out as part of the initial PATHS User Requirements study (Goodale, et al, 2011). An analysis of existing path-based systems found the minimum characteristics to be: Nodes, generally representing items from an online collection or Web pages, with associated descriptive content and metadata; Connections or links between nodes, enabling navigation across the path; Navigational tools, such as forward and back buttons, and path overviews or layouts; Annotations, added by the path-creator to provide additional context and narrative; and, in some instances, Links to other relevant content, within and external to the current collection being explored. Most systems also offered only a simple linear format for the paths being created, rather than more complex hierarchical or network structures, with the main exception being the Trailmeme system (http://www.trailmeme.com – NB: the website is no longer live).
In addition to the typical characteristics of paths, we also asked 22 cultural domain experts about their current and potential uses of path-based content within a digital environment. Eight interpretations of the path metaphor were found, grouped into 3 main categories of (i) hypertext trails defined by both user and system interactions, (ii) exploratory navigation, and (iii) processes (Goodale, et al, 2012). The main vision for PATHS focuses on the path as simple form of user interaction (path following), with opportunities for exploration (navigating away from the path, discovering content for use in own paths), and as a tool for synthesising information, interpretation, and sharing.
3.2 First PATHS Prototype
In the initial prototype for PATHS (P1) we focused on implementing the essential elements for information discovery and path creation as identified in the user requirements study outlined above. The main discovery interface comprises three main elements: Paths, Explore, and Search, supporting the three main interaction modes of following paths, exploring available content in an unfocused way (in this instance via word and image clouds), and keyword based searching.
An additional path editing interface is also available. This tool set supports the creation of a path from content that has been collected (bookmarked) during the process of searching and exploring the collection. In this prototype paths are created in a linear format, and users may add a variety of additional text content (titles, descriptions, keywords), as well as re-order the items in the path to form a narrative or other structure of their preference.
3.3 Second PATHS Prototype
Following the evaluation of P1, we then designed a second PATHS prototype (P2) with more advanced functionality for exploration and path creation, meeting an extended set of user requirements derived from user feedback. In this iteration we offer three primary exploration modes: a thesaurus, compiled using natural language processing techniques and providing an everyday language topic hierarchy derived from Wikipedia categories; a tag cloud representation of the thesaurus topics; and a semantic 2D map of the same topics. In addition, users can explore via related item links (computed based upon similarity to the current item in view), and background links to external Wikipedia content. Importantly, the exploration modes are intended to support novice users in gaining an overview of topics in the collection, thereby aiding exploratory navigation as well as the derivation of appropriate keywords to aid search effectiveness.
The path editing interface was also updated to provide more complete and flexible tools for path creation and sharing. New features included a graphical drag-and-drop editing space to allow for the construction of more complex hierarchical, branching path structures; the addition of text nodes for adding path sub-categories and other descriptive content; and new metadata and sharing options.
4. User Studies
The two iterations of the PATHS system (P1 and P2) have each been comprehensively evaluated via controlled task-based user studies with both novice and expert participants. For P1, 31 users were recruited from the UK and Ireland, and for P2, 34 users were recruited from the UK. In both instances, participants were sought who identified with one of three scenarios: regular museum visitors using CH information for leisure purposes; students using CH information in their degree course; and professionals using CH information for work-related tasks (curatorial, education, and research). These evaluation studies comprised a number of simulated work tasks, along with user profiling, observations, and feedback via questionnaires and interviews. Tasks included several simple 5-minute information-seeking activities, varied by prototype and functionality offered, plus a 30-minute unstructured path creation exercise, which users could each interpret in their own way according to an outline scenario. For example:
“Imagine you are a student who has been asked to create a path as part of a university assignment. You have been asked to use primary source materials to create a mini online exhibition suitable for a target group within the general public and/or school visitor categories. Your goal is to introduce a historical or art-focussed topic in a popular, accessible way, and to encourage further use and exploration of cultural heritage resources.”
The path creation task is the main focus of this paper, and results have been analysed using screen-recording observations, attributes of the paths created, and user feedback on their experience of this activity.
4.1 Comparison of Paths by Prototype
Both PATHS prototypes offered functionality for creating and following paths, but with some variation in the interaction features and editing functionality. In the first prototype, simple, linear paths were enabled, and in the second, more sophisticated hierarchical paths with branching could be created, allowing for more complex narratives or multiple themes to be applied to a topic of interest. In this section we analyse the characteristics of the paths created and user experiences according to the two prototype iterations.
First, we consider the amount of time taken by users to create a path. A limit of 30 minutes was allowed, and in each evaluation, a number of participants used the full amount of time, whilst others used considerably less. The time taken covers the entire task, from exploring content, collecting items and then creating the path using the editing tools. It is interesting to note that despite the more complex functionality available in P2, the mean average time taken is some 49 seconds less than for paths created with P1. As the majority of participants for both evaluations were new to the system, this result perhaps indicates that there were some aspects of the task that were found to be easier, or quicker to complete in P2.
|Time taken (minutes)||
Table 1: Time taken to complete the path creation task
User ratings for ease of task completion and enjoyability are broadly similar for both prototypes, suggesting that this is more a factor of user preference than of system complexity. On a 7-point scale from +3 to 0 to -3, path creation was given a positive rating for being both easy to complete (58%) and enjoyable (74%) for P1, compared with a positive rating of 53% and 68% consecutively for P2.
Next we consider the main characteristics of the paths created by users. The number of items each path contained varied considerably. Users were given no guidelines on the number of items to collect for their path, but were able to view a variety of existing paths created by other users. In both prototypes, the mean number of items added is 10-11, although paths created via P2 are a little shorter overall. In both instances, the most popular range is for 6-10 items, although for P2 there are more paths created with a higher number of items. This is interesting in that fewer expert users participated in the evaluation of P2, and so it appears to indicate that this more advanced system is more conducive to novice users creating longer, more complex paths.
|Number of items||
Table 2: Number of items in paths created by users
For comparison, we compare the PATHS evaluation results with an analysis of trails from the Trailmeme API undertaken as part of the PATHS User Requirements study (Goodale, et al, 2011). Trailmeme offered one of the most sophisticated path creation tools at the outset of the PATHS project, targeted at the education community, and allowing for network-style path structures. This more complex functionality resulted in trails created in the range of 0-125 items, but even so, the mean is similar to the PATHS results at 12 items, and the median (9 items) and most common range (6-10 items) are identical to the PATHS results. Overall, this suggests a relatively low average number of nodes for paths created in an experimental environment, but there is an indication that over time, with increased expertise, or in specific use cases, there is the potential for longer paths to be created.
Table 3: Features used in the path creation task, % of users
Analysis of the use of path augmentation tools tells us how users enhanced their paths through the addition of titles, descriptions, item annotations, user-defined text nodes, and thumbnail images. Addition of a path title, description and keywords was more common in P1 than in P2. We attribute this largely to the presentation of the path editing interface; in P1 a single form was used for all elements of path editing, but in P2 features were accessed via button/dialog box combinations. This difference also impacted the order of users’ interactions; in P1 most users began by describing the path and adding metadata, whilst in P2 users first assembled the path using the new drag and drop feature, and attended to metadata and descriptions at a later stage.
Contrary to this finding, there was an increase in P2 in the addition of annotations, to provide additional contextual information for the path items. These again required the use of a dialog box, but their use may have been encouraged by the use of the new path and node preview features, which clearly illustrate the lack of text when annotations have been overlooked. Interestingly, one of the most common comments on what users would do to improve their path given more time would be addition of more annotations.
New path editing tools in P2 included the selection of a thumbnail image to represent the path (added by 53% of users), and the use of user-defined text nodes for inclusion of additional descriptive content and thematic organisation (adopted by 41%). These features were requested by users of P1, and show promise for creating richer, more fully-featured paths.
Finally we consider the path content, looking at both the topics of the paths created and the ordering of content within the paths. Figure 6 illustrates the diversity of topics selected by users participating in the evaluation of P2. Topics were defined by the users themselves, and there was generally some familiarity with the subject chosen, based upon leisure, study or work interests. Some users did report, however, that they were constrained by the coverage of the collection they were using, and there was some evidence of users adapting topics from paths that were already in the system.
Topics can be grouped in several main categories. In P1, popular categories for paths were history subjects (26%), places (21%), art subjects (18%), and objects (15%). Users of P2 created paths on history subjects (27%), places (21%), culture (21%), and art and architecture (18%). Whilst there is some variation, the core categories are similar, with the main difference being the emergence of paths related to aspects of culture in P2, which may have been stimulated via the visibility of this topic in the thesaurus.
Both prototypes allowed items in a path to be arranged into a sequence according to the user’s preference. In P1 this meant re-ordering a linear list of path items, whilst in P2 hierarchical structures could be created. This change in functionality resulted in 76% of users introducing some degree of branching within their paths, and only 24% sticking to the simple linear format. On analysing the type of order in which the items were arranged, we found that for P1 the dominant mode of ordering was either chronological (32%) or narrative (23%), whilst in P2 preferences were for themes (50%) and narratives (24%). Narrative is clearly evident in the use of both systems, but the varying functionality offered for linear and branching paths would appear to affect the use of chronological and thematic structures.
4.2 Comparison of Paths by User Type
Differences in the type of paths created in P2 were also analysed by type of user, novice or expert, categorised according to their degree of specialist subject and domain knowledge. The novice/expert type has been shown to affect information task performance, especially with regard to finding content, but may also impact the type and complexity of paths created; we may expect that CH professionals, teachers and researchers in the expert category are more used to assembling content in an organised way.
Results indicate that experts are more likely to construct shorter, more focused paths than their novice counterparts, with 80% adding no more than 10 nodes, compared with only 50% of novices, although the longest path was created by an expert user. Additionally, novices are more likely to construct simple linear paths, whilst experts are more likely to construct complex branching paths, with even proportions of each user category creating an intermediate type of simple branching path.
Experts are somewhat more likely to add a path title, description and keywords than novice users. All of these are added via a dialog box, along with an option for a thumbnail image to illustrate the path, and yet for thumbnails, novices are marginally more likely than experts to add this feature.
Variations were also found in the topics selected and the ordering of path items. For the most popular topics noted above in 4.1, novices are more likely to choose topics relating to historical subjects, art and architecture, or places, whilst experts are more likely to choose topics relating to culture. Specialist topics such as archaeology, education and sociology are unique to expert users, whilst activity topics and topics classed as ‘unclear’ are unique to novice users. In ordering the items in a path, 90% of expert users chose a thematic or narrative structure, and whilst novices also favoured thematic ordering (54%), the remainder were spread thinly across a diverse mix of other types.
We also asked for feedback on preferences for finding and sharing information, where there are further differences between novice and expert users. A striking example of this is in the perception of which tools are most useful for finding items to include in a path. Whilst both types of users tried all features, novices are more likely to find the thesaurus, tag cloud and map exploration tools to be useful, whilst expert users are more likely to find the search box and facets to be useful. Novices are also more likely to find the related item links and links to Wikipedia content to be more useful than experts.
After creating a path, 71% of novices and 70% of experts would be happy to share it publicly for other users to view and follow. However there is a marked difference in the permissions that would be given for editing and re-using the path, with experts (60%) much more in favour of re-use permissions than novices (29%).
4.3 Applications for Paths in Digital Cultural Heritage
As part of the user requirements gathering and both prototype evaluations we interviewed potential users about likely applications for PATHS in real world environments. From these qualitative results, several core use case scenarios are emerging. These can be classified into cultural heritage, teaching and learning, research, community and leisure, and examples for each are given below:
- Cultural heritage: collection overviews, online exhibitions, guided tours, creative works derived from CH objects
- Teaching and learning (teacher): lesson plans and learning objects, demonstrating good practice in information tasks
- Teaching and learning (student): group work, inquiry-based learning activities
- Research: keeping a history of items found in a collection, keeping notes, sense-making, discussing with colleagues
- Community: co-creation, sharing and discussing items and paths
- Leisure: local history research, keeping track of favourite items, discussing with other enthusiasts, sharing with friends and family
Many of these use cases are already feasible with the latest PATHS prototype, with the main exceptions being tasks requiring collaborative work, and more advanced discussion and communication functionality. Applications outside of the cultural heritage domain have also been identified, including those more closely allied to professional information work in areas such as law and health. Interestingly though, very few users explicitly identified using paths as information objects in physical cultural heritage environments, although the potential is clearly there, especially if the system is adapted for use on mobile devices.
Following user evaluation studies with two iterations of the PATHS system, we have found that the task of creating a path from items found in a cultural heritage collection can be completed to some degree of satisfaction and competency by a wide range of users, both novice and expert. Given time constraints in the user studies, and users’ lack of previous system and collection knowledge, their outputs from the path creation task are quite remarkable and bode well for use in real world contexts.
Paths created in controlled evaluation sessions may typically be expected to be shorter than, say, exhibitions created for the physical space, and inevitably they are somewhat sparse in content due to the constraints noted above. However, when asked what they would do to improve their paths given more time and resources, common responses included adding more and/or better items, adding detailed and informative contextual information, and working on a more satisfying order and structure. ‘Better’ items might have a more attractive or interesting image, more descriptive content, or be more useful or representative in terms of the narrative and themes being conveyed. Contextual information might be added by the user, but could also be provided via links to relevant external content, located within authoritative and/or popular Web-based resources. All of these responses indicate a high level of engagement with the path creation task, and a desire by users to create something that is interesting, useful and well-structured.
Other key findings are that system functionality and interface design have significant effects on the structure and content of paths created. The linear path format that is more commonly found in online CH environments is more likely to result in chronological paths, whilst thematic paths are better suited to hierarchical structures. Narrative can be supported in both linear and hierarchical formats, and as Manovich (1999) suggests, there are opportunities for multiple narratives to co-exist within a collection. To this end, we have begun to consider how to visualise inter-related paths on the same topic by surfacing paths via thesaurus topics and by indicating ‘crossing paths’ in item records.
At a more fundamental level, the layout of the path editing interface clearly affects the use of path augmentation features. There is further work to do here to find an optimal interface design that encourages the addition of contextual information and metadata without making the editing space cluttered and difficult to use. The structure of paths also appears to be closely aligned to the system functionality, with branching paths created by a majority of users when the functionality is available.
User experience has also been seen to affect the type of paths created, as well as the interaction modes preferred for completing this task. Experts are more likely to construct concise, well-structured paths based upon topics relevant to their domain and specialist interest, using a thematic or narrative sequence, whilst novice users create more diverse paths in terms of length, topic and order, with simpler structures, and more general subject matter. Since CH online collections are intended for use by a wide range of user types in different scenarios, these differences need careful consideration both in terms of the functionality offered for path creation, and the types of paths created and shared. As some degree of imitation was evident in the paths created by novice users, it may be possible to exemplify good practice through the provision of professionally curated paths.
Experience from the PATHS project indicates a strong case for including path-creation tools in online CH collections. These can be utilised to support a variety of user tasks, and may be created by both experts and novices. Paths created by professionals such as curators, educators and researchers have the potential for use in providing collection and topic overviews as a starting-point for content exploration, in supporting informal learning, and as a means of sharing best practice and discussing ideas. User-created paths provide a new dynamic, with opportunities for inquiry-based and informal learning activities, co-creation, community and collaborative endeavours, and wider sharing and communication with other users.
The linear path format is shown to be a baseline for path creation activities that is easily supported and used without detailed instruction, enabling the creation of chronologies and simple narratives. However, the hierarchical path format has been demonstrated to offer much greater opportunities to generate the thematic and narrative content interpretations that are more commonly associated with cultural heritage content presentations. As yet, more advanced network structures are relatively untested, and provide significantly greater challenges for interface design and use; we would therefore propose that further research is needed in this area before they are recommended for use, especially with novice users. Hierarchical paths would therefore seem to be the optimal solution, with a balance of flexibility, and relative ease of use.
From the PATHS user study evaluations several areas have been identified that could be improved upon within the hierarchical path format. Feedback from users has indicated that elements of path editing, path presentation and following could be further simplified, and made more visual than in the present implementation. Work on content enrichment to support wider exploration via links to related and recommended items, within paths created by users and within the collection in general, is also a work in progress. We have achieved some success in this area, and although participants made little use of these features during the evaluations, they indicate that this type of contextualisation would be useful. In a similar way, links to external content are seen as useful additions to paths and to the collection as a whole, but are little used as yet, and there is a desire for links to more authoritative sources than Wikipedia.
Additional needs for tools to support enhanced communication and sharing have also been highlighted by several users. These are fairly commonly offered within existing social media environments and as such would not represent a new development. They are therefore not an immediate concern for the PATHS research, although we seek to better understand which existing tools would be of most use in specific user scenarios.
Most of these additional requirements are unlikely to be addressed before the project ends at the close of 2013, but demonstrate considerable scope for extending the PATHS work in future projects, especially in those areas related to collaborative working and creation of more complex paths. Support for co-creation and group work has significant implications for extending the use of paths as tools for meaning-making and creativity, and as information objects for sharing within curatorial, community and educational initiatives.
Further work is also required to examine the content of paths that is most useful and desirable within different user contexts. These needs will be explored to some degree through forthcoming qualitative field trials in a variety of naturalistic settings, with use of PATHS by individuals over an extended time period. From this work, initial guidelines on the length, structure (branching), topics, and ordering of paths may be developed, which will be useful for CH professionals and educators as a basis for designing content for novice audiences. Additional consideration for the presentation of multiple complementary or conflicting narratives within a collection should also be explored, along with discussion of the processes for evolving path content over time, and reusing paths in open content situations.
The research leading to these results was carried out as part of the PATHS project (http://www.paths-project.eu) funded by the European Community’s Seventh Framework Programme (FP7/2007-2013) under grant agreement no. 270082. The authors gratefully acknowledge the contributions of all partners in the PATHS Project, and all participants in the user requirements and evaluation activities.
Adams, A. & Blandford, A. (2005). Digital Libraries’ Support for the User’s ‘Information Journey’. In: Proceedings of the 5th ACM/IEEE-CS Joint Conference on Digital Libraries, (Denver, Colorado, June 07-11, 2005). ACM, New York, NY, 160-169. DOI= http://dx.doi.org/10.1145/1065385.1065424
Agosti, M., et al. (2013). Interacting with digital cultural heritage collections via annotations: the CULTURA approach. In Proceedings of the 2013 ACM symposium on Document Engineering, pp. 13-22. ACM.
Bates, M.J. (1989). The Design of Browsing and Berrypicking Techniques for the Online Search Interface. Online Information Review, 13(5), 407-424.
Fernie, K., et al. (2012). PATHS: Personalising access to cultural heritage spaces. In: Proceedings of the 18th International Conference on Virtual Systems and Multimedia (VSMM 2012), 469-474. IEEE.
Frohlich, D., M. & Rachovides, D. (2008). Using Digital Stories for Local and Global Information Sharing. In: Proceedings of CHI’08, Florence, Italy, 5-6th April 2008. ACM.
Goodale, P, et al. (2012). User-Centred Design to Support Exploration and Path Creation in Cultural Heritage Collections. In: Proceedings of EuroHCIR2012, (Nijmegen, Netherlands, August 25, 2012), 75-78.
Goodale, P., et al. (2011). PATHS Project D1.1 User Requirements Analysis. PATHS Project. http://www.paths-project.eu/eng/Resources/User-Requirements-Analysis.
Mäkelä, E., Suominen, O. and Hyvönen, E. (2007). Automatic Exhibition Generation Based on Semantic Cultural Content. In: Proceedings of Cultural Heritage on the Semantic Web Workshop at ISWC+ ASWC 2007.
Manovich, L. (1999). Database as Symbolic Form. Convergence: The International Journal of Research into New Media Technologies 5(2), 80-99.
Mulholland, P. & Collins, T. (2002). Using Digital Narratives to Support the Collaborative Learning and Exploration of Cultural Heritage. In: Proceedings of the 13th International Workshop on Database and Expert Systems Applications, 2002. IEEE.
Peterson, D. & Levene, M. (2003). Trail Records and Navigational Learning. London Review of Education, 1(3), 207-216.
Pirolli, P. & Card, S. (1995). Information Foraging in Information Access Environments. In: Proceedings of the SIGCHI conference on Human Factors in Computing Systems. ACM, New York, 51-58. DOI= http://dx.doi.org/10.1145/223904.223911.
Shipman, F.M., et al. (2000). Guided Paths through Web‐Based Collections: Design, Experiences, and Adaptations. Journal of the American Society for Information Science 51(3), 260-272.
Walker, K., Main, A. & Fass, J. (2013). User-Generated Trails in Third Places. HCI-3P Workshop on Human Computer Interaction for Third Places at CHI ‘ 13, Paris, France, April 27 – May 2.
White, R. W. & Huang, J. (2010). Assessing the Scenic Route: Measuring the Value of Search Trails in Web Logs. In Proceedings of the 33rd international ACM SIGIR conference on Research and development in information retrieval, 587-594.
Wilson, M., et al. (2010). From Keyword Search to Exploration: Designing future search interfaces for the web. Foundations and Trends in Web Science, 2(1),1–97.
Zdrahal, Z., Mulholland, P. & Collins, T. (2008). Exploring Pathways across Stories. In: Proceedings of International Conference on Distributed Human-Machine Systems. 2008.
P. Goodale, P. Clough, M. Hall, M. Stevenson, K. Fernie, J. Griffiths and e. agirre, Pathways to Discovery: Supporting Exploration and Information Use in Cultural Heritage Collections. In , N. Proctor & R. Cherry (eds). Silver Spring, MD: Museums and the Web. Published October 2, 2013. Consulted .