A FUZZY LOGIC BASED TEXT SUMMARIZATION

Abstract

Today’s world is all about information, with most of it online which enables anytime, anywhere, easy and unlimited access; participation & publishing of information has consequently escalated the suffering of ‘Information Glut’. Assisting users’ informational searches with reduced reading or surfing time by extracting and evaluating accurate, authentic & relevant information are the primary concerns in the present milieu. Automatic text summarization is the process of condensing an original document into shorter form to create smaller, compact version from the abundant information that is available, preserving the content & meaning such that it meets the needs of the user. Though many summarization techniques have been proposed but there are no ‘silver bullets’ to achieve the superlative results as of human generated summaries. Thus, the domain of text summarization is an active and dynamic field of study, practice & research with the continuous need to expound novel techniques for achieving comparable & effectual results. Fuzzy logic has appeared as a powerful theoretical framework for studying human reasoning and its application has been explored within the domain of text summarization in the past few years. One key aspect of text summarization is accurate identification of keywords from the given textual content. In this project, a new technique based on fuzzy logic has been proposed using two graph based techniques named as TextRank and LexRank and one semantic based technique named as Latent semantic analysis(LSA). In our work, we have also investigated their relative performance with the proposed method. All these methods used in developing hybrid model are of extractive summarization type. The techniques are evaluated on Opinosis data set using ‘ROUGE-1’ and ‘time to extract the keywords’. The proposed technique has outperformed the existing techniques, when compared with the results given by the original studies.