Collective Intelligence-based Web Page Search: Combining Folksonomy and Link- based Ranking Strategy * * This study was supported by a research grant (2008) from the University of Seoul and supported by Seoul R&BD Program (NT080624). 1 Corresponding author Tao Zhang, Byungjeong Lee 1 School of Computer Science The University of Seoul Seoul, Korea kerryking@ieee.org, bjlee@uos.ac.kr Hanjoon Kim Department of Electrical and Computer Engineering The University of Seoul Seoul, Korea khj@uos.ac.kr Sooyong Kang Division of Information and Communications Hanyang University Seoul, Korea sykang@hanyang.ac.kr Jinseog Kim Department of Statistics and Information Science Dongguk University Gyeongju, Gyeongbuk, Korea jinseog.kim@gmail.com Abstract — With the exponentially growing amount of information available on the Internet, retrieving web pages of interest has become increasingly difficult. While several web page recommender systems have been developed, it is still difficult to search related information which reflects users’ preference. In this paper, we propose a new type of web page search which is based on the collective intelligence. It combines folksonomy and link-based ranking evaluation scheme so as to accommodate users’ preferences. We implemented the prototype system and demonstrate the feasibility of the proposed web page search scheme. Keywords - Collective Intelligence; Folksonomy; Ranking Strategy; Link-Based Web Search I. INTRODUCTION With the exponentially growing amount of information available on the Internet, we often need to spend much time to search necessary web pages from a large number of Internet documents created every day. Many web page search techniques have been developed and search engines became more complicated to process a large number of web pages. However, in terms of search quality, about one half of all retrieved web pages have been reported to be irrelevant [1]. Even though these search systems were developed so as to search appropriate web pages reflecting user’s favor [2] [3], there are still some problems unresolved. The major problem is that even though the search systems can acquire large amount of web pages reflecting users’ preference from Internet, it is still unsatisfactory to analyze and cluster them because of the huge number of web pages [4]. To obtain better search results from massive web pages on the Internet, we propose a prototype web page search system based on collective intelligence. Collective intelligence has become increasingly popular and more important with the advent of new communications technologies [5]. Collective intelligence [6] means the combining of behaviors, preferences, or ideas of a group of people to create novel insights. Folksonomy [7] [8] supports a new way for classification of the information provided by the users. It reflects users’ behaviors on web searching. Link-based page ranking [9] is a kind of intelligent web page search strategy and is used for optimizing the search results. It tries to reflect users’ preferences on the process of web search. In this paper, we propose a collective intelligence- based search system which combines folksonomy and link- based page ranking scheme. In what follows, we introduce folksonomy and ranking based on links in Section Ⅱ. In Section Ⅲ, some related works are presented. In Section Ⅳ , we describe the architecture and algorithms of the proposed web page search system. In Section Ⅴ, some experimental results of the prototype system are presented. In the end, we summarize our work and introduce future work in Section Ⅵ. II. FOLKSONOMY AND LINK-BASED RANKING STRATEGY Recently, folksonomy and ranking strategy based on links are getting popular and spreading widely. Folksonomy is one of the components of Web 2.0 which is famous for Semantic Web. Link-based page ranking is also an important approach in web search strategy. A. Folksonomy Folksonomy is a new classification technique which collaboratively creates and manages tags to annotate and categorize contents. It means the attached tags or labels to each web page to suffice the practice and method of annotating and categorizing contents. Users put keywords called “tags” to each page freely and subjectively, based on their meaning. Anyone can choose any word as tag and can put multiple tags to one page. Fig. 1 shows an example of tagging web pages by different users. In this example, we IEEE Ninth International Conference on Computer and Information Technology 978-0-7695-3836-5/09 $26.00 © 2009 IEEE DOI 10.1109/CIT.2009.118 166 Authorized licensed use limited to: Dongguk University. Downloaded on November 20, 2009 at 06:16 from IEEE Xplore. Restrictions apply.