An Adaptation Algorithm for Adaptive Web Based Systems based on Link Structure and Document Similarity BUJAR RAUFI South East European University Dpt. of Computer Science Ilindenska nn, 1200 Tetovo MACEDONIA b.raufi@seeu.edu.mk JULIANA GEORGIEVA Technical University of Sofia Fac. of Computers and Control Systems Kl. Ohridski 8, 1000 Sofia BULGARIA july@tu-sofia.bg ARTAN LUMA South East European University Dpt. of Computer Science Ilindenska nn, 1200 Tetovo MACEDONIA a.luma@seeu.edu.mk FLORIJE ISMAILI South East European University Dpt. of Computer Science Ilindenska nn, 1200 Tetovo MACEDONIA f.ismaili@seeu.edu.mk XHEMAL ZENUNI South East European University Dpt. of Computer Science Ilindenska nn, 1200 Tetovo MACEDONIA xh.zenuni@seeu.edu.mk Abstract: Many traditional adaptive web based systems today are complex in its design and implementation which makes them hard to implement in general purpose web sites. The lack of efficient approach and adaptation al- gorithm makes their usage in isolated cases like: e-commerce sites or only education systems. In this paper we introduce a new adaptation algorithm based on web sites link structure and document similarities both represented through matrices. Initial results show good performance of the algorithm for both per-user based adaptation and adaptation for groups of users. Key–Words: Adaptive web based systems, adaptation algorithm, document similarity, data mining 1 Introduction Many of the traditional web pages used today suffer from a ”one-size-fits-all” [1] problem. This means that when a user accesses a web page of a library, it will be presented with the same set of information ev- ery time it visits it. A tourist that tries to find informa- tion about travel and hotel arrangements will be pre- sented with the same order of information regardless of its destination preferences [1]. The same issue can be generalized for many, but not all, web applications in various domains like: e-commerce, corporate web sites and virtual education. The solution for overcom- ing this situation is the development of user-adaptive software systems [1] or Adaptive Web Sites. Adaptive Web Sites represent Web-Based Systems that tend to arrange their content, structure or both based on user access preferences. Starting as a pioneering work of a few research communities in the middle of 90’s [2] today it repre- sents a research milestone for many communities in- volving various research disciplines like: user model- ing, Web Usage Mining, natural language processing, intelligent tutoring systems, cognitive sciences and se- mantic web. The adaptation in such systems is done by building a model of goals, knowledge and prefer- ences as well as using this for interaction with the user during its browsing. In traditional adaptive web based systems such an interaction is done by building a user model and applying the same for adapting the content for the purpose of the user itself [3]. This relationship between system, user and adaptation is presented by Brusilovsky in the form of user modeling-adaptation loop [3]. In this approach, the system collects data about user, processes the same and finally results are presented in an adapted way. The adaptation pro- cess, as stated by Brusilovsky, is performed through the adaptation methods and adaptation techniques[3]. By adaptation methods we consider a more gen- eralized and abstracted approach concerning adapta- tion. This usually represents a description of what the adaptation should do rather than how to do it. In adaptation methods, the implementation per se can be done in various ways. Methods usually include a more abstracted representation of adaptation rules such as: ” insert a certain text fragment in a page if a certain concept is been visited by the user etc”. In adaptation techniques on the other hand, a more detailed and well specified approach for delivering adaptation is used Proceedings of the 9th WSEAS International Conference on TELECOMMUNICATIONS and INFORMATICS ISSN: 1790-5117 29 ISBN: 978-954-92600-2-1