Pointwise Mutual Information for Opinion Mining Feature Level Product Reviews

The determination of whether a product holds a positive or negative sentiment can be inferred from reviews provided by previous customers. In recent years, various websites have emerged that offer product reviews, in which the value of a product is evaluated through user-generated ratings and textual comments. However, the abundance of reviews often makes it challenging for prospective customers to interpret the overall sentiment accurately. To address this issue, a classification approach can be employed to determine the polarity of product reviews. Opinion Mining, also known as sentiment analysis, is a field of study that focuses on analyzing individuals’ opinions toward entities, individuals, issues, events, topics, and their associated attributes. The implementation of feature extraction prior to the classification process has been shown to significantly enhance the accuracy of sentiment assessment. One effective method for feature extraction is Pointwise Mutual Information (PMI), which leverages search engine statistics to identify meaningful term associations in real time. PMI enables the system to capture semantic relationships between words, thereby improving the reliability of sentiment classification.