A Social Curiosity Inspired Recommendation Model to Improve Precision, Coverage and Density and by Combining Collaborative Filtering
Problem The aim is to design a recommendation system that is not only dependent on the user’s historical content, but also on his/ her friend’s activity that might be a surprise element to the user which will elevate his/her curiosity and will compel the user to explore more about that recommendation. In doing so, I attempt to combine this with collaborative filtering in order to further improve accuracy of the recommendation. Importance Traditional recommendation systems consider user-item information for recommendation systems. However, they overlook the social connections and the effect of social recommendations on the particular user’s curiosity. For example, a user who does not like horror movies will have less suggestions about horror movies. Her friend might also not like horror movies, but her friend likes or watches a horror movie. This piques the user’s interest and curiosity in the horror movie as well, even if she does not like the movie. Recommending such out-of-context, yet curiosity arousing items can be appreciated by the user and can increase the usability of the system. Furthermore, we can also perform collaborative filtering to search for users whose interests are similar to the subject user. Any surprise action or item accessed/ seen by the similar users may also interest our subject user.