Unsupervised Document-Level Sentiment Analysis of Reviews Using Macaronic Parser

Abstract

t Exponential rise in the multilingual web content affects the present-day decision support system to a great extent. To normalize such web content is the need of an hour. Reliability of decision support system broadly depends on the flawless processing of the language data present over the web. Macaronic text is one of the text usually found over the web. It is basically the text that contains number of languages in a single document instead of uniform language for whole document. To cope up with such a text, in this paper we propose a macaronic parser. This parser is language-independent and task-independent. The output of the proposed system is the normalized uniform base language text. This output can further be used in many other language processing tasks.