Amharic aspect based opinion summarization using bootstrap on hotel domain

Abstract

Over the last few years, this special task of summarizing opinions has stirred tremendous interest amongst the NLP and Text Mining communities. The simplest form of an opinion summary is the result of sentiment prediction. Aspect-based summarization divides input texts into aspects, which are also called as features and subtopics, and generates summaries of each aspect. Today millions of web-users express their opinions about many topics through blogs, wikis, web, chats and social networks. Especially sectors such as e-commerce and e-tourism, it is very useful to automatically analyze the huge amount of social information available on the Web, but the extremely unstructured nature of these contents makes it a difficult task. One of the main reasons for the lack of study on opinions is the fact that there were little opinionated texts available before Web. Today millions web-users express their opinions about many topics through blogs, wikis, web, chats and social networks. The objective of this research is to design and develop aspect opinion summarization for opinionated Amharic documents. The major components of this work are Amharic opinions/reviews, preprocessing phase, aspect and opinion learner, opinion word Seed Lexicon, polarity defining and Aspect based Amharic opinion summarization by graph. The Average performance of user center evaluation for our aspect based Amharic opinion summarization system for bootstrapping is 91.38% or 4.569 out of 5 weigh. In system centered evaluation the semi supervised bootstrap method achieved the averages of positive and negative effectiveness 92% precision, 72.8% recall and 81.40% Fmeasure. Also in the naïve Bayes hotel review classification approach average performance test result is that 75.68% precision, 87.05% recall and 77.86% F-measure less than bootstrapping. Up to now there is no systems that digesting those huge amount of customer opinions given in Ethiopic (Amharic language) for understanding opinion holder (customers) need on a given domain particular entity. Therefore in this work Ethiopic (Amharic language) customer opinions on a hotel domain was summarize with respect to their aspects /features by graph visualization with the performance of above. This work will help for organization (such as hotels), individual (such as hotel users), government intelligence, and business intelligence