期刊名称：International Journal of Computer Science and Network Security
印刷版ISSN：1738-7906
出版年度：2019
卷号：19
期号：2
页码：113-119
出版社：International Journal of Computer Science and Network Security
摘要：Machine Translation, Information Retrieval and Knowledge Acquisition are the three main applications of sentiment analysis based on aspect word. But in some cases, there may be two aspect words having different spellings, yet belonging to the same aspect category, e.g. the words ‘costly’ and ‘expensive’ belong to the same category ‘price.’ Grouping such aspects into the same category is critical for sentiment analysis. Some have done this by using synonym, LDA (Latent Dirichlet Analysis) groups and Constrained-LDA. The WordNet dictionary is mostly used for such analysis. Although it contains a lot of synonyms, yet sometimes there are such aspects which are not synonymous, but still refer to the same broad category. For example, in the following two sentences: “this camera will not easily fit in a coat pocket” and “this camera is of large size ” aspects ‘fit’ and ‘size’ both are not synonyms yet belong to the same category. In this study, the proposed framework consists of four steps: DefString, Supporting-Words, SetWords and Intersection for Decision. The fourth stage will decide that the two aspects belong to the same category or not. After experimenting on a dataset comprising 252 pairs of same-meaning aspects and 62 pairs of different-meaning aspects, 91% accuracy and 86% F score were detected.
关键词：Machine Translation; Information Retrieval; Sentiment Analysis; Aspect Extraction.
