Abstract:Although Web search techniques have greatly facilitate users’ information seeking, there are still quite a lot of search sessions that cannot provide satisfactory results, which are more serious in Web image search scenarios. How to understand user intent from observed data is a fundamental issue and of paramount significance in improving image search performance. Previous research efforts mostly focus on discovering user intent either from clickthrough behavior in user search logs (e.g., Google), or from social data to facilitate vertical image search in a few limited social media platforms (e.g., Flickr). This article aims to combine the virtues of these two information sources to complement each other, that is, sensing and understanding users’ interests from social media platforms and transferring this knowledge to rerank the image search results in general image search engines. Toward this goal, we first propose a novel social-sensed image search framework, where both social media and search engine are jointly considered. To effectively and efficiently leverage these two kinds of platforms, we propose an example-based user interest representation and modeling method, where we construct a hybrid graph from social media and propose a hybrid random-walk algorithm to derive the user-image interest graph. Moreover, we propose a social-sensed image reranking method to integrate the user-image interest graph from social media and search results from general image search engines to rerank the images by fusing their social relevance and visual relevance. We conducted extensive experiments on real-world data from Flickr and Google image search, and the results demonstrated that the proposed methods can significantly improve the social relevance of image search results while maintaining visual relevance well.

An Information Retrieval Method Based on Social Annotations

Query Subtopic Mining Via Subtractive Initialization of Non-negative Sparse Latent Semantic Analysis

A LDA Topic Model Based Collection Selection Method for Distributed Information Retrieval

Joint Latent Dirichlet Allocation for Social Tags.

Improving Web Image Retrieval Using Image Annotations and Inference Network

Modeling Social Annotation: a Bayesian Approach

A Search-Based Web Image Annotation Method

Research on Tag Semantic Retrieval in Social Tagging System Based on LSA

Retrieval Modification Based on Citation Tagging

ON SENTENCE-BASED INFORMATION RETRIEVAL MODEL

On Annotation Methodologies for Image Search Evaluation

Improving social book search using structure semantics, bibliographic descriptions and social metadata

Semi-supervised document retrieval

A Social Search Model for Large Scale Social Networks

Annotating Images by Mining Image Search Results

Social-Sensed Image Search.

Efficient Tag Mining Via Mixture Modeling for Real-Time Search-Based Image Annotation.

Bringing the Associative Ability to Social Tag Recommendation.

The Use of Categorization Information in Language Models for Question Retrieval

Extracting Interest Tags for Non-famous Users in Social Network