A Novel Approach to Enhance Personalized Recommendation

Recommendation techniques are important in the fields of E-commerce and other facilities like online shopping. One of the main problems is dynamically providing high-quality recommendation on less data. In this paper, a new dynamic personalized recommendation approach is introduced, in which information contained in both ratings and profile contents are utilized by inventing internal relations between ratings, a set of dynamic features are developed to describe user preferences in different phases, and then a recommendation is made by adaptively weighting the features. This approach performed well on public data sets.