Research And Implementation Of Photo Information Recommendation System For Mobile Environment

Along with the explosion of modern information, recommendation systems are playing an important role in our daily life. For mobile device manufactures, how to fast and effectively share and browse photos is a key factor for the promotion of mobile devices. Most current recommendation systems can be divided into three categories:content based recommendation systems, collaborative recommendation systems and hybrid recommendation systems. Content based recommendation systems are wildly used in modern industries, including web recommendation, article recommendation, restaurant recommendation, and TV shows recommendation and so on. And it is a hot research direction of recommendation researches. Typical content based recommendation systems use classifying algorithms like Bayes classifier to classify items into several different groups. They then sort all items in each group and make recommendations according the ranking list in each group.Conventional content based recommendation systems make a ranking list for a group of similar items and generate the same recommendations for items in the same group. One drawback of the approach is that it ignores differences between items in the same group. In order to make use of the rich metadata information of photos, we need a new design of the content based recommendation system. We first map relations between items to an un-directed weighted graph and then make a ranking list for each item in the graph. The proposed content recommendation system makes recommendations based on which photo the user is currently viewing. How to determine the influence of each metadata over the relationship of each pair of items, is the key problem the paper tries to resolve. Meanwhile, some other services are also needed to enrich the functionality of Kaleido Photo software.The paper first introduces some related background technologies, including Web service, Restful, recommendation systems overview and content-based recommendation systems. Then it describes the semantic-relation-graph-based photo recommendation system. After that, we shows our design and implementation of the proposed recommendation system in Kaleido Photo in details, including photo uploading service, user management service and recommendation service. Then we list our testing environment and testing results. At last, we conclude our work, discuss our weak point and point out our future work.