Short text computing based on lexical similarity model

Abstract

Short text similarity deals with determining the closeness of two text mean the same thing by lexical or semantic. Various short text similarity approaches have been proposed which are based on lexical matching, semantic knowledge background or combining models. Lexical based model does not capture the actual meaning behind the words. However, semantic approach are relying on knowledge background or corpus which cannot be assumed to be available in handling such huge new word of data sparseness and noise in short text. This work are focusing on lexical-based similarity models for analysing the unstructured short text. The term-based and edit distance model are used in comparing the applicability of these model to compute the similarity value of short text. The experimental results shows that each model have their key strengths and limitations in computing similarity value of short text.