Research On Knowledge Transfer Between Enterprises Based On Link Prediction

Since Barabasi and Albert discovered scale-free networks,more and more scholars have analyzed the real-world networks,and studying knowledge transfer from the perspective of social networks has become a trend in the field of knowledge management research.The research on the influencing factors of knowledge transfer mainly focuses on the transfer willingness,transfer ability,absorption and transformation ability of transfer subjects and the difference of basic knowledge among subjects,as well as the transfer situation and knowledge characteristics.Based on the perspective of social network,this paper studies the network structure The impact on the effect of knowledge transfer.On the basis of constructing a multi-layer patent citation multi-layer network,this paper proposes a hybrid link prediction model combining the patent citation network structure and patent node attribute information to predict the possibility of knowledge transfer of Chinese integrated circuit enterprises in the future.Selection and strategy formulation have important practical significance.This article first sorts out the research data at home and abroad,clarifies the research purpose,content and method,and then defines the concepts and theoretically expounds the multi-layer network,knowledge transfer and link prediction.Secondly,it analyzes the direction and purpose of patent citation and knowledge transfer,and then analyzes the necessity of constructing a multi-layer patent citation network to study knowledge transfer.Thirdly,by collecting patent citation data of integrated circuit enterprises,select time nodes 1970 and 2010 to construct a three-layer patent citation network,analyze the distribution of node degrees in the multi-layer network,and analyze the technical fields and patents contained in the patent documents Analysis of the time of acquisition of rights,patent holders and other information and the possibility of knowledge transfer between enterprises.The study found that the patent citation network conforms to the scale-free network feature,and the high node out-degree or in-degree indicates that the patentee has advanced technology or strong knowledge absorption and transfer capabilities.Chinese companies have higher out-of-node patent nodes and higher in-degree Low,which is related to the recent patent acquisition time of Chinese companies and the time lag of patent citations;the more similar the technical field,the greater the opportunities for future cooperation;then,this paper also proposes a hybrid link prediction based on a multi-layer network across time Model to predict the possibility of future cooperation between patentees.The study found that the possibility of a connection between patent owners at the same layer is higher than that between patent owners at different layers.This article also focuses on the position of Chinese companies in the multi-layer network of knowledge transfer.Not only do Chinese companies and foreign companies have similar knowledge or technology,they can become potential collaborators in the future,and they can use patent citation networks to identify foreign competitors.There is a lot of room for future cooperation and development between domestic companies.At present,we should pay attention to changes in patent applications of foreign competitors,grasp the direction of technological development in a timely manner,and cooperate with potential partners to jointly build R & D centers to promote knowledge transfer.The government should promote the construction of technological R & D and innovation platforms,promote the complementarity of regional advantages,and create favorable conditions for talent exchange and cooperation.The research in this paper helps to deepen the combination of link prediction methods and multi-layer networks,and at the same time provides theoretical guidance for enterprise partner selection,competitor identification,and development strategy formulation.