Collaborative Social Annotation And Retrieval Of Multi-dimensional Data In Web2.0 Environment

Information sharing and online collaboration in web2.0 environment allow users not only get various kinds of knowledge conveniently,but also publish any type of data or add annotation on existing data freely.A common application is adding tags or labels on forum articles for the convenience of categorizing and searching.We call this application of adding tags social annotation.These annotations from large amount of users online become an important means for information retrieval.Currently this kind of annotation mainly consists of easy word or phrase tags.To plain text(like blog articles or forum posts),these easy tags are enough to enhance the efficiency of retrieval,but there are still many multi-dimensional data like images or Geographical information,whose content and structural information can not be sufficiently expressed by easy tags.Therefore,providing a new annotation model for multi-dimensional data on the internet,using annotation to reflect people's knowledge of data genuinely,is a necessity to enhance efficiency of these data's retrieval.Based on existing work,according to the features of multi-dimensional data, social annotation and multi-user collaborative annotation,this paper makes a deep research on social annotation's consistency control problems and retrieval problems using structural semantic annotation for multi-dimensional data.The main contents are as follows:The first part is collaborative social annotation and consistency control for multi-dimensional data in Web 2.0 environment.Build a collaborative annotation model - structural semantic annotation model for multi-dimensional data,abstract the annotation information,and suggest algorithms for detecting and solve conflicts.The Second part is retrieval of multi-dimensional data based on social annotation. We build a conceptual space and a probabilistic model for multi-dimensional data according to emergent semantics and latent semantic index.Stat the co-occurrence between data and annotation,and then build emergent semantics between annotation and data,which is used to rank the query result.Users are admitted to search the multi-dimensional data with the highest co-occurrence between the query and itself.The last part is a prototype system of collaborative social annotation for image and retrieval evaluation method.Integrating above methods,we suggest a collaborative annotation framework in Web2.0 environment,use image data that focus on community,to realize a collaborative image annotation and retrieval system which supports large amount of users online.