Approximate Keyword Search Using Content Modelling

Abstract

We present Approximate Keyword Search using content modelling, a novel approach to
approximate querying. We focus on data where the search results in a meaningful response instead of exact
matching of keywords. The eigenvectors for all terms together form a (vector space) index and can be used
for document retrieval techniques. The use of the Eigenvector reduces the size of the index. It also helps in
identifying the documents with more relevance even though there is no match of keywords represented by the
user query. Great results show that the Usage of Eigenvalue reduces the index size.