ABSTRACT:

With the explosion in the number of semi-structured information users access and store in personal info management systems, there is a essential would like for powerful search tools to retrieve often terribly heterogeneous knowledge in a easy and efficient manner. Existing tools usually support some IR-vogue ranking on the textual part of the question, but solely think about structure (e.g., file directory) and metadata (e.g., date, file type) as filtering conditions. We tend to propose a completely unique multi-dimensional search approach that allows users to perform fuzzy searches for structure and metadata conditions additionally to keyword conditions. Our techniques individually score every dimension and integrate the three dimension scores into a meaningful unified score. We tend to additionally design indexes and algorithms to efficiently establish the foremost relevant files that match multi-dimensional queries. We perform a thorough experimental evaluation of our approach and show that our relaxation and scoring framework for fuzzy question conditions in non-content dimensions will significantly improve ranking accuracy. We tend to conjointly show that our question processing methods perform and scale well, making our fuzzy search approach sensible for each day usage.