I. Introduction
With increasing on-line stores of opinion based documents (movie reviews, product feedback, etc.), a large area of research and analysis has grown up around them [1]. This had created a growth area for tasks such as polarity classification, agreement detection and analysis of the degree of positivity related to the polarity of documents such as those mentioned above [2]. This level of analysis is useful as it allows for a high level summary of the tone of a document to be identified, through the use of different machine learning and natural language processing techniques. However, this disregards the fact that a document may be made up of aspects linking it to multiple sentiments or opinions, and this highlights a limitation of how many classifiers are being applied to the sentiment analysis field. To address this issue, we propose a fuzzy rule based system for classification of sentiment.