Social Image Semantic Information Refinement And Enrichment Based On Context

Social tagging on online websites provides users interfaces of tagging an image with their own tags, while they upload the social image. Then, people can describe or comment the image anytime. The information associated with the image can be seen as the semantic information of the social image. To better utilize the image and its semantic information, we should refine and enrich the semantic information.Firstly, the article introduces the refinement and enrichment of the semantic information asscociated with the social image. Then, processes is conducted to solve the two problems.Then, the article designs a series of processes to refine the semantic informantion. In order to solve the lexical problem, a process named keyword filtering is designed. Information entropy is used to calculate the keyword initial relevance. Keyword correlation is calculated from two sides. One side is to measure semantic similarity of keyword pairs using the structured information of the free encyclopedia Wikipedia. The other one is to compute the visual similarity of keyword pairs based on the visual representation of the keyword. To re-rank the original keywords, a fast random walk with restart is used and the top ones are reserved as the final keywords.Finally, the article mines association rules on the collected database, and use these rules to complete the semantic information of the social image. A knowledge database is used to add synonyms and hypernym to the semantic information.