Web Semantics for Personalized Information Retrieval

Web Semantics for Personalized Information Retrieval

Aarti Singh (Guru Nanak Girls College, Yamuna Nagar, India) and Anu Sharma (ICAR-IASRI New Delhi, India)
Copyright: © 2017 |Pages: 21
DOI: 10.4018/978-1-5225-2483-0.ch008

Abstract

This chapter explores the synergy between Semantic Web (SW) technologies and Web Personalization (WP) for demonstrating an intelligent interface for Personalized Information Retrieval (PIR) on web. Benefits of adding semantics to WP through ontologies and Software Agents (SA) has already been realized. These approaches are expected to prove useful in handling the information overload problem encountered in web search. A brief introduction to PIR process is given, followed by description of SW, ontologies and SA. A comprehensive review of existing web technologies for PIR has been presented. Although, a huge contribution by various researchers has been seen and analyzed but still there exist some gap areas where the benefits of these technologies are still to be realized in future personalized web search.
Chapter Preview
Top

Anatomy Of Personalized Information Retrieval System On Web

The main components of the system are personalized user query expansion, personalized techniques for user profiling and ranking of web search results followed by personalized filtering of search results based on long and short term interest. The anatomy of a web PIR system is shown in Figure 1.

Figure 1.

Anatomy of a Web PIR system

These components are described in brief in next sub-sections.

Query Expansion Techniques

The most challenging issue to consider in web information retrieval is to resolve the ambiguity arising out of poorly defined queries. Information requirements may vary with the different search sessions. Query expansion techniques aims at reformulating the query to meet the user requirements. Mainly two types of techniques are available- global and local. Global methods utilizes the existing thesaurus /WorldNet, create thesaurus automatically or perform spell check. Local methods consist of recording the relevance, pseudo relevance and indirect relevance feedback. Relevance feedback is recorded either implicitly or explicitly. Implicit user feedback is calculated by observing their behavior on web. These methods are further divided into word co-occurrence, probabilistic methods, context and location based methods. Figure 2 represents schematically the classification of query expansion techniques.

Figure 2.

Various Types of Query Expansion Techniques

Various methods for creating user profiles are given in next sub-section.

Purchase this chapter to continue reading all 21 pages >

Complete Chapter List

Search this Book:
Reset