Research And Implementation Of Multi-information Fusion Technology For Experts In Chemical Industry

Chemical experts are an indispensable resource in the field of chemical research and chemical production.Chemical experts can provide product consultation,product innovation,and production guidance in the production process of chemical companies.However,in real world,there is a dilemma that companies have no experts and experts have no projects.At present,expert knowledge representation has the problems of single expert information source and incomplete expert information representation.In this paper,we adopt the knowledge representation method of multi-information fusion to study it,in order to improve the comprehensiveness and accuracy of expert information representation,and show the potential cooperation relationship among experts.The main research contents are as follows:The information of chemical experts is obtained from different information sources by web crawlers,such as personal web pages of university teachers,thesis websites,patent websites and fund project websites.Data processing is performed on the acquired data: data cleaning,construction of keyword dictionaries,data word separation,data feature extraction,and the processed data is saved for backup.The description of expert information often adopts logical representation,generative rule method and framework diagram method for expert information presentation.The traditional knowledge representation methods are easily restricted by a single knowledge source and cannot comprehensively display expert knowledge,and at the same time,they cannot explore the cooperation relationship and geographical relationship contained among experts,and cannot display expert information in an all-round way.The LDA topic model is used to extract the topics of expert papers and patent research,and the top five topic words with topic weight value are extracted by TF-IDF algorithm to jointly build the expert topic subnet;the co-occurrence matrix is constructed by extracting author co-occurrence relationship,and the expert cooperation relationship tightness is expressed by the centrality value calculation to build the expert cooperation relationship subnet;the expert personal information is extracted from the expert Web interface by regular expressions The expert Web subnet is constructed by extracting personal information from the expert Web interface by regular expressions.The expert information network is constructed by combining three expert subnetworks and fusion rules with the expert name and organization as constraints to represent the expert information in multiple information fusion.The experimental results show that compared with the traditional expert knowledge representation method,the expert information network based on multiple information fusion can not only display the expert information comprehensively,but also explore the potential cooperation relationship and cooperation potential among experts,which verifies the effectiveness of the chemical expert information representation based on multiple information fusion.