A Comparative analysis of machine learning algorithms applied to multi lingual texts summarization

Abstract

Over the scarce period the World Wide Web (WWW) takes prolonged extremely and huge volumes of information in the form of news articles is available online. Many a times individuals don’t take the spell besides tolerance towards recite whole news divisions or ample long articles. At this time ascends the essential of computerized texts summarization. Uncertainty an instant of the real fillings of the broadcast object is obtainable formerly it will convert calmer for the handler to get a gist of the article as well as it would save a lot of his time. Nearby, numerous methods towards texts summarization which could be off the record on the root of numerous factors such as level of processing, kind of information being processed, etc. The work proposed in this paper tries to integrate these approaches with modern computational linguistics, semantic technologies and machine learning algorithms to devise a novel technique for multi lingual text summarization which could produce summaries aimed at sole too as group of forms. The anticipated method specifically addresses two major languages for the study, one is English being the language used worldwide and second Hindi being the national language  of India. The machine learning techniques used for extraction are neural networks and fuzzy logic systems. Finally, a comparison of these techniques is done to show that fuzzy logic systems give better precision as compared to neural networks for summarization in both the languages. The average difference in precision is around 8-10% for Hindi and around 45-50% for English text documents.