An Efficient Methodology for Measuring Sentence Similarity Using Combinational Semantics

摘要

We are living in the days of information explosion. The availability of latest techniques for information management such as clouds, big data analytics etc. promotes the addition of millions of documents in World Wide Web day by day. It’s a tedious job to find the required information from these huge textual volumes without the help of an efficient text processing algorithm. Also such algorithms are the back bones of almost all information management applications such as data extraction, document summarization etc. The calculation of document similarity or specifically sentence similarity is the main component of all such algorithms, which decides the efficiency of the entire text processing applications. Even though a number of approaches are available for measuring textual similarity, neither of the algorithms can predict similarity as well that of a linguistic expert. Analysis shows, the semantic approaches performs better than the traditional syntactic approaches, since they are considering meaning for calculating similarity. In such cases the semantic tool used for calculation and the efficiency of the applied logic decides the accuracy level and the entire performance of the application. In this proposal we are presenting an efficient method for measuring document similarity using a combinational semantic approach which combines multiple semantic calculations and is different from the existing approaches with usage of the semantic tool Themesets.