An Optimization Technique for Image Search in Social Sharing Websites
Saurabh Trikande · 2013
sharing websites allow users to personalize media. The user can generate, share, tag and comment on media. The large-scale user-generated meta-data not only facilitates users in sharing and organizing multimedia content, but provides useful information to improve media retrieval and management of the websites. Personalized search serves as one of such examples where the web search experience is improved by generating the returned list according to the modified user search intents. In this paper, we exploit the social annotations and propose a novel framework considering both the user and query relevance to learn to personalized image search. The basic premise is to embed the user preference and query-related search intent into user-specific topic spaces. Since the users' original annotation is too sparse for topic modelling, we need to enrich users' annotation pool before user-specific topic spaces construction. The proposed framework contains two components. Firstly a Ranking based Multi-correlation Tensor Factorization model that is proposed to perform annotation prediction, which is considered as users' potential annotations for the images. Secondly we introduce User-specific Topic Modelling to map the query relevance and user preference into the same user-specific topic space. For performance evaluation, two resources involved with users' social activities are employed. Experiments on a large-scale dataset are used to demonstrate the effectiveness of the proposed method.