Connotative features based affective movie recommendation system

There is difficulty in assessing the emotions by movies with subject to the emotional responses to the content of the film by exploring the film connotative properties. Connotation is used to represent the emotions described by the audiovisual descriptors so that it predicts the emotional reaction of user. The connotative features can be used for the recommendation of movies. There are various methodologies for the recommendation of movies. This paper gives comparative analysis of some of these methods. This paper introduces some of the audio features that can be useful in the analysis of the affections represented in the movie scenes. A hybrid approach using machine learning and cluster analysis can also be used for recommending the movies. The video features can be mapped with emotions. Interest, boredom, frustration, and puzzlement and some emotional states such as neutral, happiness, sadness, anger, disgust, fear, and surprise can be detected by using multi-stream fused Hidden Markov Model. Movies music features can also be utilized for emotion recognition. This paper compares all these methods that can be utilized for the recommendation of movies based on user's emotions.