Towards the Exploitation of Statistical Language Models for Sentiment Analysis of Twitter Posts

Abstract : In this paper, we investigate the utility of linguistic features for detecting the sentiment of twitter messages. The sentiment is defined to be a personal positive or negative feelings. We built n-gram language models over zoos of positive and negative tweets. We assert the polarity of a given tweet by observing the perplexity with the positive or negative language model. The given tweet is considered to be close to the language model that assigns lower perplexity.