Improving Object Based Ranking of User Comments from Social Web using Hodge Decomposition

摘要

The user shares their thoughts on social web sites often through posts and comments. Users register to communities using their personal information. The social web sites like Yahoo, YouTube, Facebook and Twitter provides a large volume of general information of users interest. The popularity of social websites is increasing very fast because of the large scale of user participation, through contributor tags, ratings and comments. The user comments provide a large and rich source of contextual information. The user contributed comments may be in a mixed form, comments may be relevant as well as not relevant to relevant to the particular post. The proposed system ranks the user contributed comments. Weights are assigned to comments as per different criteria and we used Hodge Decomposition algorithm for ranking. The LambdaMART algorithm is used for comparing the relevance ranking performance with Hodge Decomposition.