[Truncated abstract] The rise of blogs, forums, social networks and review websites in recent years has provided very accessible and convenient platforms for people to express thoughts, views or attitudes about topics of interest. This huge volume of opinionated information is valuable, for example when we want to forecast the sales of a product to improve the business or predict the election outcome when the competition is so tight and the time is limited. In other words, we are partial to know what people think about miscellaneous subjects, more speci cally, we are interested in opinions, the subjective information. In order to collect and analyse such content on the Internet, a combination of techniques such as Information Retrieval, Computational Linguistics and Natural Language Processing is used which is often referred to as Opinion Mining and Sentiment Detection. Various sentiment detection techniques have been developed based on an integration of part-of-speech tagging, negation handling, lexicons and classi ers but there is always a question whether computers will ever be capable of understanding the natural language or they could only be used to look up for signs of positivity and negativity in the discourse? Could we implement more complex systems to handle the complexity of written and spoken language? Aside from the commonly used approaches in sentiment detection which are based on bag of words and classi ers, techniques like SO-LSA (Semantic Orientation from Latent Semantic Analysis) are proposed...
|Publication status||Unpublished - 2012|