Study On Experts Recommendation System For Industry-University-Research Cooperation

At present,industry-university-research cooperation is primarily carried out under the auspices of a related institution,which first collects supply and demand information manually and then organizes the personnel of enterprises,colleges and research institutes to work together on site.This mode has defects such as non-transparent information,poor communication,limited scale and scope and so on,so it cannot easily improve the low connection efficiency and achievement transformation rate of university-industry-research cooperation.However,there are just technical requirements and achievements on the numerous online technology transaction platforms established by the local government or intermediary agencies,which can neither solve the problem of smart match between technology needs and talents,as well as the problem of timeliness and convenience.So,an intelligent expert recommendation system that can meet technical demand urgently needs to be developed to resolve this difficulty.To promote effective university-industry-research cooperation,this paper primarily develops an expert recommendation system based on latent semantic text similarity calculation according to the construction of scientific and technical dictionary,the building of technical demand model,and the building of expert knowledge model.The concrete research in this paper is shown as follows:(1)Construction of scientific and technical dictionary.Owing to the rapid development of science and technology,and the constant emergence of new words,new concepts and new relationships,the general dictionary and traditional lexical analysis are already unable to meet the requirements of semantic analysis in the field of science and technology,so the constriction of a scientific and technical dictionary is a foundation for expert semantic modeling and accurate recommendation.First,new scientific and technical terms are extracted using topical crawler;second,new terms are sorted out by statistical method;finally,new terms are added continuously after being sorted up based on the traditional dictionary.(2)Research of technical expert knowledge representation model.First,an expert matter-element knowledge representation model is built according to expert features;an expert vector model is built based on scientific and technical dictionary and TF-IDF method;a latent semantic-based expert concept model is proposed based on mass expert data;meanwhile,the potential research-related semantic relationship between lexical items is mined through machine learning,and then expert clustering is realized based upon a latent semantic model.(3)Research of semantic similarity calculation-based expert recommendation method.The most similar expert clustering is first determined based on semantic similarity calculation in accordance with technical needs,and then the text similarity between expert vectors and technical demand vectors in this clustering is computed;Considering experts' practical ability in technical demand,the above expert recommendation algorithm is further optimized,and finally top N experts are recommended in the manner of linear weighted combination.Experiments show that the new method has a higher accuracy.An industry-university-research expert recommendation system,which consists of Web management server,mobile APP and expert recommendation server,is developed according to the above research output.The gradual popularization and use of this system will certainly effectively promote the effect of scientific and technological achievement transformation.