Overview and Preliminary Results for a Opinions Summarizarion Method

Web resources contain many reviews that can be used by both individuals and companies. The growing volume of these reviews becomes a problem for consumers who should read a number of reviews of other consumers to identify the strengths and weaknesses of the product. Opinion summarization is necessary to help consumers understand, in a simplified way, the large amount of reviews or present them summaries of opinion for the competing entities. In this article we present a graph-based summarization algorithm that simplifies the opinions messages graph taking into account three criteria: the graph filtering in order to remove the messages with opinions that do not refer to the target entity and the aggregation of the nodes that include opinions of the same polarity.