Context-aware search

• Main Author: Lou, Chii Hian.
• Other Authors: Sun Aixin
• Format: Final Year Project
• Language: English
• Published: 2012
• Subjects: DRNTU::Engineering::Computer science and engineering
• Online Access: http://hdl.handle.net/10356/50858
• Institution: Nanyang Technological University

At present, to deal with ambiguous queries, search engines return diverse search results in the hope of securing a user’s needs with at least one result from the first page of returned results. Due to this, users are often bombarded with results covering a whole range of topics, with only a handful relevant. Previous studies on context-aware search focus mainly on building user profile based on the browsing behavior and query formulation. However, many have neglected customized search from explicit information. This research aims to create a customized search experience based on the context knowledge of the query. The author proposed a novel way to deal with customizing search result given the explicit information. The author suggested using page structure summary to represent the page content and extract a list of keywords from the summary which best describe the query. Thereafter, the query is entered into a search engine together with the keywords to remove any ambiguity on the original query. The author examined seven text extraction techniques, paired with five feature selection. Out of the 35 combination of techniques, extracting paragraphs containing the query term paired with feature select noun phrases (T5F5) has the highest relevance precision score. The proposed system is also benchmarked against Google Suggested Searches (GoogleSS). On average, the proposed system obtained 68.8% higher relevance keywords compared to GoogleSS. In particular, T5F5 performed nearly four times higher relevance precision than GoogleSS.