MULTI-DIMENSIONAL FUZZY SEARCH FOR PERSONAL INFORMATION MANAGEMENT, SYSTEMSPROCESSING, MULTIMEDIA INFORMATION PROCESSING, IN THE CONTEXT OF SOCIAL NETWORKING

Abstract

Explosion of information is driving data users for complex search tools to have access to heterogeneous data in a simple and efficient manner which is the critical need of the hour. As users store and collect large amounts of personal data it is becoming increasingly important to quickly retrieve files in personal information systems. Numerous search tools have been developed to locate personal information stored in file systems.Existing tools often allow IR-style ranking on the textual part of the query, but consider structure (e.g., file directory) and metadata (e.g., date, file type) as filtering conditions. In this paper we propose a novel multi-dimensional approach to semi-structured and heterogeneous data searches in personal information management systems.The paper highlights the suite of techniques that allow search tools to effectively and efficiently evaluate query conditions.The paper identifies how to score fuzzy conditions in each query dimension (content, structure, and metadata) in the search,providing a complex query interface. Furthermore, the paper proposes to integrate the three dimension scores into a meaningful unified score, such that users can specify a single query and that can be evaluated at once across file boundaries. Finally, algorithms and data structures are designed to support efficient processing of the multi-dimensional as well as unified queries. We perform a thorough experimental evaluation of our approach and show that the proposed system has the potential to significantly improve ranking accuracy.In addition, we show that scoring efficiently evaluates fuzzy multi-dimensional queries and also show that query processing strategies perform and scale well, making our fuzzy search approach practical for every day usage.