Extraction And Analysis Of Technology Thread Information Based On Scientific Literatures

Scientific literature is one of the important carriers for the development of scientific and technological innovation,the scientific literature network is a special complex information network.Each literature has its corresponding title,authors,publication,keywords,citations and other information,thus forming a literature network structure centered on the literature with other objects linking together through the literature.Whether in practical engineering or scientific research,exploring technology-based assisted generation technology based on scientific literature is particularly important.Firstly,dynamically collect literature data in the computer field based on existing literature databases.The specific collection content includes the bibliographic information of the citing literature and the cited literature,and the citation relations between the literatures.Then clean the collected literature data,including deleting non-literature data,incomplete data and combining repeated data.Secondly,the literature data is clustered by using the citation relationship unique to the literature network.The bibliographic information belongs to short text.Based on the traditional clustering algorithm,this paper combines the co-citation relationship between literatures.Based on the computer domain literature dataset,three different methods for calculating the distances between two clusters are used for clustering.The mean Silhouette Coefficient is improved obviously,which effectively improves the clustering effect between the cited literature and the citing literature.The clustered literatures are projected onto the time axis,representative literature are selected,a linear technology thread is obtained,and the technology thread graph is displayed using the BDP visualization tool.Finally,a technology thread analysis system based on scientific literature is designed and implemented.The system can realize the process of online literature data collection based on input search terms,literature data preprocessing,literature clustering,technology thread analysis and technology thread visualization.