Efficient Similarity Measures for Texts Matching

Abstract

Calculation of similarity measures of exact matching texts is a critical task in the area of pattern matching that needs a great attention. There are many existing similarity measures in literature but the best methods do not exist for closeness measurement of two strings. The objective of this paper is to explore the grammatical properties and features of generalized n-gram matching technique of similarity measures to find exact text in electronic computer applications. Three new similarity measures have been proposed to improve the performance of generalized n-gram method. The new methods assigned high values of similarity measures and performance to price with low values of running time. The experiment with the new methods demonstrated that they are universal and very useful in words that could be derived from the word list as a group and retrieve relevant medical terms from database . One of the methods achieved best correlation of values for the evaluation of subjective examination.