Document Retrieval Using Proximity-Based Phrase Searching

Exact phrase matching is a powerful tool to quickly retrieve results when a sufficient section of the text is accurately provided as the query. If the section of the text is not completely accurate, phrase searching will fail. A method must be used which will enforce strict enough conditions to achieve high accuracy while allowing for mistakes in the text provided. Here we develop a method using proximity conditions to search for quotes from movies and compare the results against the Vector Space Model. Initial results show a promising accuracy in excess of 78% for documents being successfully ranked within the top ten results.

• Type: Conference
• Year: 2008
• Where: HICSS
• Authors: Kathryn Patterson, Carolyn R. Watters, Michael A. Shepherd
• Keywords: Biometrics, Exact Phrase Matching, HICSS 2008, Sufficient Section, System Sciences, Vector Space Model