We explore the generation of starred ratings of films from reviews available only in text form. This is motivated by the typical review reader behavior that looks only at the film‟s star rating (out of 4 or 5 stars) as an initial filter – to decide whether to see a film or not - and defers a full reading of the film to after seeing it so that the viewing is fresh and unaffected by potential spoilers in the review. We propose a mechanism to generate finegrained ratings (on a scale of a specified number of stars) with the help of tools that generate a ternary (positive, neutral, negative) rating of a given body of text using techniques of sentiment analysis. We also point out possible directions for further investigations.
Volume 12 | 04-Special Issue
Pages: 812-816
DOI: 10.5373/JARDCS/V12SP4/20201549