Nowadays,with the development of communication networks,unstructured “soft data” (such as audio,text,microblog,blog,BBS,et al) with the characteristics of fuzziness,randomness and subjectivity emerges in large numbers.The effective integration of unstructured “soft data” with structured “hard data” from traditional physical sensors is of great practical and urgent significance for the comprehensive understanding of complex situations.On the basis of explaining the concept and connotation of fusion of soft and hard information,this paper focuses on the knowledge extraction of unstructured “soft data” and uncertainty information representation and reasoning technology.Finally,to address the urgent need of improving the performance and robustness of the soft and hard information fusion system,this paper makes a thorough discussion on the following four aspects:performance analysis and evaluation of the fusion algorithm,correlation analysis of information and its fusion,task and resource scheduling,and research and development of the application system.