1. 1 TOWARDS THE ADAPTIVE SEMANTIC WEB (Peter Dolog et al) BASHIR NAGIB STONEY JULY 2 ND 2013 02. July.20131B.N STONEY

2. 2 Outline Background: Adaptive Hypermedia Systems Concept When are Adaptations Useful?? Application Areas Adaptation Techniques Adaptive Vs Adaptable Systems What is Personalization? Does one size fit all? How does Personalization Work? Recommender systems Logic-based Definition of Adaptive Educational Hypermedia Systems Simple: A Simple Adaptive Educational Hypermedia System Semantic Web  Triples 02. July.20132B.N STONEY

3. 3 Definition: Adaptive Hypermedia Systems ( P. Brusilovsky: Methods and Techniques of Adaptive Hypermedia. User Modeling and User- Adapted Interaction 6 (2-3), 1996) Different voices on adaptive hypermedia The ambition of adaptivity is that not only `everyone should be computer literate', but also that 'computers should be user literate' (Browne, 1990) “By adaptive hypermedia systems we mean all hypertext and hypermedia systems which reflect some features of the user in the user model and apply this model to adapt various visible aspects of the system to the user." 02. July.20133B.N STONEY

4. 02. July.2013B.N STONEY4 HYPERMEDIA SYSTEM ADAPTATION COMPONENT USER MODEL Data about user Request of user Adaptation results update Get current knowledge Process flow of and of Adaptive Hypermedia System

5. Adaptive Hyperwhat?? 02. July.2013 In 1945, Vanevar Bush envisioned a machine, the memex. By consulting several sources consecutively, a user builds an associative trail of related documents, which can be labeled and annotated with notes and comments. consulting several sources consecutively, a user builds an associative trail of related documents, which can be labeled and annotated with notes and comments. Similar to this idea, hypertext is a collection of documents that is connected by (associative) links. The World Wide Web is the most common form of hypertext. 5B.N STONEY

6. When are Adaptations Useful? when the hypertext is sufficiently large  need for guidance, recommendations and information filtering and is used by people with different goals and backgrounds  need to adapt to resulting differences in needs and user context Adaptive hypermedia are an answer to the problem of disorientation, commonly known as “Lost in Hyperspace“ The root of disorientation is the additional cognitive overhead needed for users to: identify their current position in a hyperdocument; reconstruct the way that led to this position; distinguish among different options for moving on from this position. 02. July.20136B.N STONEY

7. Lost in Hyperspace When users fail to keep track of their navigation through a document, they might arrive at a particular page and forget what was to be done there, they might neglect to return from interesting side-tracks or they might fail to find some pages that contain relevant information. Compare this with books: in a novel you can easily track how much you have already read and how much there is still to read books provide easy-to-remember cues for remembering: page number, position within the page Everyday examples on the Web: Where can I find whether my train is on-time? What was the name of that book I found on Amazon? Which query did I use to find this list of hotels that I did not save? 02. July.20137B.N STONEY

8. Application Areas Educational hypermedia. Traditionally, the most popular application area. User needs are well-identified: acquiring knowledge on a certain topics. User characteristics (e.g. knowledge level, interest) are well-researched. Online information systems. Nowadays we would simply refer to them as Web sites. A popular adaptation technique is collaborative filtering and automated recommendations of products or other items. E-commerce. Product recommendations and personalized offers are particularly important in the field of e-commerce Web information retrieval. The filtering, sorting, categorizing or annotating of search results based on the users' search history. 02. July.20138B.N STONEY

9. Adaptation Techniques There are two main categories of adaptation techniques: adaptive presentation and adaptive navigation support. Nodes:- Content-level adaptation, adaptive presentation techniques Which means to select\modify\rearrange the content of the documents Edges:- Navigational-level adaption, adaptive navigation support Which means to select appropriate hypertext links, delete links,or generate new links. 02. July.20139B.N STONEY

10. 02. July.201310B.N STONEY Hypertext

11. Adaptive Vs Adaptable Systems 02. July.2013 ttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttt Adaptable means that users can adapt system behavior themselves e.g iGoogle personalized start page 11B.N STONEY

12. 02. July.2013 Adaptive means that the system adapts its own behavior on the users behalf e.g. Amazon recommendations 12B.N STONEY

13. What’s Personalization?? 02. July.2013 Personalization is a designers' approach to achieve harmony between users, tasks, environments and the system (Benyon, 1993) Personalization is an overrated concept. Rather than investing time and energy on trying to predict individual users' needs it would be better to enhance the overall site design. (Nielsen, 1998) 13B.N STONEY

14. Does one Size fit all?? In a library, a person looks for some books on China. What will the librarian recommend? Is the person a small child who saw a TV show about China and wants to learn about this exotic country? Or a high school student working on a paper? Perhaps a prospective tourist? A scholar interested in Eastern philosophy? Someone who can read Chinese? Elaine Rich: User Modeling via Stereotypes. Cognitive Science 3, 329-354 (1979) 02. July.201314B.N STONEY

15. Most likely the librarian will make an educated guess, based on the person's appearance :- age, style of clothing, accent, choice of words,... 02. July.201315B.N STONEY

16. This initial guess might be confirmed or refuted by observations. It is assumed that a European cannot read Chinese, unless said otherwise Children are generally not (yet) interested in Eastern philosophy, but there are exceptions.. …………. The educated guess, a stereotype can be refined with follow-up questions. Persons expect a personalized advice, even though the librarian does not know them. 02. July.201316B.N STONEY

17. How does Personalization Work? In a nutshell, a personalized system tries to understand the user using information that he or she provides activities that a user carries out And tries to guess what you currently want to do, by comparing you with people that are similar to you things that you own or like with similar things 02. July.201317B.N STONEY

18. Recommender Systems Items may be: movies, tv programs, music, books, news, images, web pages, scientific literature,... 02. July.2013 Recommender systems work from a specific type of information filtering system technique that attempts to recommend items that are likely to be of interest to the user. 18B.N STONEY

19. Content-based systems:- examine properties of the items recommended. For instance, if a Netflix user has watched many cowboy movies, then recommend a movie classified in the database as having the “cowboy” genre. Collaborative filtering systems recommend items based on similarity measures between users and/or items. The items recommended to a user are those preferred by similar users. 02. July.201319B.N STONEY

20. A logic-based definition of Adaptive Educational Hypermedia Systems(AEHS) E-learning has been an important application area of adaptive hypermedia research. clearly defined user goals (learning outcomes) established ways to test whether the goals have been met (e.g. tests) known individual differences between learners (e.g. knowledge, learning style, learning goals) 02. July.201320B.N STONEY

21. Components of a AHS An adaptive educational hypermedia system (AEHS) is a Quadruple ( DOCS, UM, OBS, AC ) DOCS: Document Space: A finite set of first order logic (FOL) sentences with constant symbols for describing documents (and knowledge concepts), and predicates for defining relations between these (and other) constant symbols. UM: User Model: A finite set of FOL sentences with constant symbols for describing individual users (user groups), and user characteristics, as well as predicates and formulas for expressing whether a characteristic applies to a user. OBS: Observations: A finite set of FOL sentences with constant symbols for describing observations, and predicates for relating users, documents / con- cepts, and observations. AC: Adaptation Component: A finite set of FOL sentences with formulas for describing adaptive functionality. 02. July.201321B.N STONEY

22. “Simple”: A Simple Adaptive Educational Hypermedia System This adaptive educational hypermedia system can annotate the links of the hypertext according to the actual knowledge state of the user. Simple: Document Space A set of n constants (n corresponds to the number of documents in the document space) which represent the documents: D1, D2, : : :, Dn. A finite set of predicates stating the documents that need to be studied before a document can be learned, e.g. Dj is a prerequisite for Di: preq(Di, Dj) for certain Di ≠ Dj. Simple: User Model A set of m axioms, one for each individual user: U1, U2, : : :, Um. 02. July.201322B.N STONEY

23. Simple: Observations One constant for the observation whether a document has been visited: Visited. And a set of predicates obs(Dj, Ui, Visited) for certain Dj, Ui. Simple: Adaptation Component One constant for describing the values of the adaptive functionality “learning state": Recommended for reading, and two constants representing values of the adaptive functionality: Green_Icon, Red_Icon. Rules for describing the learning state of a document 02. July.201323B.N STONEY

24. TRIPLES 02. July.2013 The Semantic Web is an extension of the current web in which information is given a well-defined meaning, better enabling computers and people to work in cooperation. Tim Berners Lee 24B.N STONEY

25. TRIPLE Overview Triple is rule language for the Semantic Web. A core building block of RDF is triples: subject - predicate - object. The RDF Data Model “Peter is interested in Sweden” subject: http://www.peter.de/foaf.rdf#me; predicate: foaf:interest; object: http://en.wikipedia.org/wiki/Sweden; 02. July.201325B.N STONEY

26. 02. July.201326B.N STONEY

27. Conclusions An Approach to Personalization on the Semantic Web based on reasoning was described. Adaptive Hypermedia is an Alternative to the traditional “one-size- fits-all” static approach of hypermedia systems Adaptive Web vision is to provide users with optimized access to distributed electronic information on the Web according to particular needs of individual users or group of users. 02. July.201327B.N STONEY

28. SEMANTIC WEB TECHNOLOGIES FOR THE ADAPTIVE WEB Peter Dolog and Wolfgang Nejdl Outline Hypertext Adaptive Web Systems Ontogolies Reasoning Techniques eLearning Domain 02. July.2013B.N STONEY28

29. Definition  The Links are provided either explicitly, encoded by authors of the pages, or they are generated automatically, for example based on the results of the query.  The Adaptive Web Systems extend the adaptive navigation and presentation techniques from closed corpus adaptive hypermedia to the open corpus information resources available on the Web. 02. July.2013B.N STONEY29 Hypertext interlinks related pieces of information(pages) and allows the user to browse through the information space

30. Ontologies: represent shared and agreed upon conceptual models in a domain, which describe the main concepts of the domain and their relationships.  Ontologies can thus serve as reference models for generating links in this domain, and represent hypertext, content and user information.  Reasoning techniques can then work on metadata based on these ontologies, and generate links based on content, user context and user background. 02. July.2013B.N STONEY30

31. Hypertext and Links 02. July.2013B.N STONEY31 Links in Conceptual open hypermedia are usually described as associations between the source and target information fragments Figure showing the adaptive context of a learning resource in a course.

32.  The Link is annotated by a traffic light metaphor to inform the user which of the resources are ready for him to use according to his background The Green Symbol:- means that a link is recommended Red:- a link is not recommended Yellow:- User has to acquire some prerequisite background needed to access the resource. 02. July.2013B.N STONEY32

33. 02. July.2013B.N STONEY33 A prototype for search user interface.

34. Reasoning on the Semantic Web Several query and reasoning languages have been introduced to query for, and reasonon, metadata on the semantic web such as QEL or SPARQL. The semanticsof the languages are often based on Datalog, as used in the Edutella Query Language(QEL) and extended rule and logic programming languages. QEL offers a full range of predicates in addition to equality, general Datalog rules,and outer join. An example for a simple QEL query over resources is the following: s(X,, Y), s(X,, S), qel:equals(S, ). 02. July.2013B.N STONEY34

35. The query tries to find resources where dc:subject equals java:OO Class. The prefixes qel:, dc:, and java: are abbreviations for URIs of the schemas used. Variable X will be bound to URIs of resources, variable Y will be bound to titles of the resources, and variable S will be bound to subjects of the resources. A rule language especially designed for querying and transforming RDF models isTRIPLE. Rules defined in TRIPLE can reason about RDF-annotated information resources, translation tools from RDF to TRIPLE and vice versa are provided. 02. July.2013B.N STONEY35

36. Sergej Zerr36 THANK YOU!!!!!!! 02. July.201336B.N STONEY