An Analysis of the Applicability of User Trails in Web Applications Web Engineering Workshop August 10, 2004

1. An Analysis of the Applicability of User Trails in Web ApplicationsWeb Engineering WorkshopAugust 10, 2004 Erich Gams, Sigi Reich Salzburg Research Jakob-Haringer Straße 5/III 5020 Salzburg Austria Salzburg Research Forschungsgesellschaft m.b.H. | Jakob-Haringer Str. 5/III, 5020 Salzburg, Austria T +43.662.2288-200 | F +43.662.2288-222 | sreich@salzburgresearch.at | www.salzburgresearch.at

2. Overview • Web Engineering and Web Application Modelling • Navigation Patterns • The Concept of User Trails • The Trail Pattern • Classification of Web Applications • Analysis of the Applicability of the Trail Pattern • Summary and Conclusion © Salzburg Research 2004

3. Web Applications: The Challenge • The World Wide Web has evolved into a global environment delivering applications such as reservation systems, online shopping or auction sites, games, multimedia applications, calendars, maps, chat applications or data entry/display systems, and many more • Web applications are characterized by multiculturalism, continuous change, fast pace and competitiveness, high demands on user adaptivity, ... • Thus, the complexity of designing, developing, maintaining such Web applications has increased significantly © Salzburg Research 2004

4. Web Engineering: The Discipline • Web Engineering, as an emerging new discipline, promotes a process and a systematic approach to development of high quality Web-based systems. • It “is the application of systematic and quantifiable approaches (concepts, methods, techniques, tools) to cost-effective requirements analysis, design, implementation, testing, operation, and maintenance of high-quality Web applications.“ (and “… also the scientificdiscipline concerned with the study of these approaches.“) • And: "A Web application is a software system based on technologies and standards of the World Wide Web Consortium (W3C) that provides Web specific resources such as content and services through a user interface, the Web browser."[Kappel et al. 2004]  Modelling Web applications – identified as core activity – helps to manage the raising complexity of systems and therefore plays an important role in Web engineering. © Salzburg Research 2004

5. Modelling Dimensions for Web Applications Figure in [Kappel, et al. 2000] © Salzburg Research 2004

6. Navigational Patterns – Part of Web Modelling 1/2 • Modelling the structural aspect of the hypertext level results in the hypertext structure model. • This includes classes of the content model that can be visited directly, e.g. by links embedded in a document. • Modelling the behavioural aspect enhances the hypertext structure model with access elements and models: • Dynamic behaviour, e.g. computing end points of links at run time • Access of content through supplementary navigation structures, e.g index pages or guided tours. • Identifying the (navigational) access methods of users results in navigational patterns. © Salzburg Research 2004

7. Navigational Patterns – Part of Web Modelling 2/2 • Navigational Pattern Examples [Rossi, 1999]: • Set-based Navigation: organizes information in sets of related items, e.g. The Browse category functionality in www.ebay.com. • Landmark: Provide access to different unrelated subsystems, e.g. links to your account or help are available throughout the whole website in www.amazon.com • Guided Tour: e.g. Virtual exhibitions • Navigation patterns build complex navigation architectures based on nodes, links and indexes. • Typically navigational patterns only reflect the designers‘ view of the navigational space (and not necessarily the reader‘s view). • Once implemented, the structure and the navigation elements will hardly change (only if the site is redesigned).  We believe that the concept of trails can enrich Web applications with a navigational element that dynamically adapts to users. © Salzburg Research 2004

8. The Trail Concept 1/3 • Trails are an established approach to assist users in navigating hypertexts. • People leave trails by handling information, e.g. by interacting with Web applications. • Trails constitute a specific path through a set of documents. • Interesting paths can be extracted from other people’ s paths, who have “walked” similar ways when searching for similar things or topics. • Trails are a concept, enabling recommendation of different types of categories, namely • similar paths, • related documents and • users with similar interests. © Salzburg Research 2004

9. The Trail Concept 2/3 • Trails also raise the following issues: • “Cold start problem“: for recommendations based on “paths of many people“, a trailbase with a certain amount of data is needed in order to provide useful recomendations. • Privacy, “Big Brother“: there is the potential of recorded data to be given away to others • The fact that many people take certain paths does not necessarily mean that this is useful; but it often is (and in any case, it is a useful "sign-post" for taking your choice). © Salzburg Research 2004

10. The Trail Concept 3/3 • Trails represent the users’ point of how to access and traverse a Web application within the natural navigation borders set by the document space; • e.g. recommended paths or documents may change, (depending on the trails of the major part of people traversing a Web site). • These trails “adapt” dynamically due to users‘ interaction to the user whenever navigation action takes place.  The trail concept defined as a navigational pattern supplies a designer with a navigation structure that can be built into applications during the design phase. © Salzburg Research 2004

11. A Trail Pattern Definition* • Intent • Provide the user with information about other users. e.g. find similar trails of users, documents or similar users. • Problem • Navigation design is always the designers’ point of view and will thus remain restricted in terms of structure and navigational possibilities. • Solution • Rather than author-based, pre-defined tours, automatically established trails of previous users are recorded, stored in a trailbase and provided to end users. * The structure “intent“,“problem“,“solution“ follows [C. Alexander, et al., A Pattern Language: Towns, Buildings, Construction. Oxford University Press, 1977] 1977. © Salzburg Research 2004

12. Method of Analysis • [Kappel et al.,2004] classify Web applications by complexity and development history into 8 categories each containing several typical application types. • We investigated how the concept of trails — according to the criteria • similar documents • similar paths • similar interests • can be applied to each of these categories, or each of the types, respectively. © Salzburg Research 2004

13. Document Centric Web Applications • Stored ready-made on a Web server, without holding any dynamic component. • E.g. Web presentations of companies, communities, or a Web radio.  Generally, the applicability of trails depends on the overall size and range, the number of pages and links, of the Web site. © Salzburg Research 2004

14. Interactive Web Applications • Allow users to interact with radio buttons, selection menus, etc. • News sites (e.g. www.cnn.com), virtual exhibitions, etc.  Virtual exhibitions are most suitable for path recommendations, news sites could be enhanced with documents, or news messages that users have been interested in. © Salzburg Research 2004

15. Transactional Web Applications • Focus on business activities. • E.g. Online banking systems, online shopping (www.amazon.com), online booking systems (www.expedia.com)  These applications offer a wide range of possibilities for the applicability of trails to assist users. (Gather and analyse consumers’ data from shopping applications).  Other application types (e.g. online banking System) have limited applicability of recommendation functionality. © Salzburg Research 2004

16. Collaborative Web Applications • Communication between cooperating users is high. • support generating, editing and managing information in a shared workspace. • Simple chat systems, collaborative work or e-learning platforms.  There exist many possibilities to apply the trail concept to various collaborative Web applications, but depending on the number of previously existing trails e.g. the number of users participating. © Salzburg Research 2004

17. Portal-Oriented Web applications • Combine various resources of information and services in single point of access. • Netscape, Microsoft, newspapers or search engines, or portals targeting specific communities, shopping, auction or marketplaces.  There exist many possibilities to apply the trail concept to various collaborative Web applications, in general this depends on the number of previously existing trails e.g. the number of users participating (“cold start problem of trails“). © Salzburg Research 2004

18. Ubiquitous Web Applications • Provide access to services anytime, from anywhere • Enabling multi-platform delivery, customization or location-dependency. • e.g. up-to-date information on specific location, buildings featuring services for tourists..  Enhance the notion of trails from where they originate — the physical world. Documents can be represented by physical objects (Linked to the virtual world by e.g. via radio frequency tags or 2-D barcodes) e.g. buildings, places. © Salzburg Research 2004

19. Workflow based Applications and Semantic Web Applications • Workflow based Applications • Examples: B2B solutions in an e-commerce or e-government domain. • E-government applications offer a predefined workflow of forms, which can only be accessed in a determined order. The applicability of trails is rather limited. • B2B systems are mainly based on Web services and do not fulfil the requirements of human-computer interaction via a user interface. • Semantic Web • One objective is to provide information on the Web not only in human-readable form but also machine readable form. • However, we assume that trails are only constituted by humans. • Machines could provide information about their goal thus agents would be able to find agents doing similar work and exchange search results. • This domain and its applications are still evolving. © Salzburg Research 2004

20. Summary • Navigational patterns play an important part in modelling the hypertext layer of Web applications. • The concept of trails — enabling recommendations of paths, documents and persons — has been used to defined a trail navigation pattern that is able to express the users’ view on how to navigate a document space. Furthermore, the navigational access to the document space changes dynamically. • However, they only represent a static view and also an author‘s point of view. • We presented a first analysis on the applicability of the trail pattern to Web applications. © Salzburg Research 2004

21. Conclusions and Outlook • Web applications can benefit from the fact that the trail pattern allows to recommend users as well as documents. • Authors and designers are supplied with a navigation structure that is able to model the users‘ navigational point of view. • The purpose of Web applications strongly influences the degree of usage of the trail concept (e.g., booking vs. shopping applications). • Trails are generally well suited for those Web applications with a high number of interacting users, and mostly static, i.e. non personalized content (e.g. virtual exhibitions, community portals). © Salzburg Research 2004

22. References • [Kappel, et al. 2000] Gerti Kappel, Werner Retschitzegger, and Wieland Schwinger. Modelling customizable web applications — a requirement’s perspective, In Proceedings of the Kyoto International Conference on Digital Libraries: Research and Practice, November 13 - 16, 2000, Kyoto University, British Library (U.K.), pages 168–179, 2000. • [Kappel, et al. 2004] Gerti Kappel, Birgit Pröll, Siegfried Reich, and Werner Retschitzegger, editors. Web Engineering — Systematische Entwicklung von Web-Anwendungen. Dpunkt Verlag, Heidelberg, 2004. In German. • [Rossi,1999] G. Rossi, D. Schwabe, and F. Lyardet. Patterns for Designing Navigable Information Spaces. Addison-Wesley, Reading, Massachusetts, 1999. © Salzburg Research 2004