Visual Analysis Of Institutional Research Based On Topic Model

In the Internet era,due to the rapid development of information technology,data shows an explosive growth.For the academic research field,the increase of researchers leads to the increasing of academic literature.Using and understanding these data can better understand the structure and dynamics of science itself,which requires the combination of data analysis and visualization.Different research institutions involve different research fields,methods and models.Subject modeling and visual analysis of academic literature can reflect the development and trend of an institution in various research fields.Topic model becomes the main text mining algorithm for academic documents,which is used to extract the hidden information — topic.Then,the theme and other data information are displayed and analyzed through visualization elements,so as to understand and master the research development trend of an institution.In view of the above requirements,this thesis proposes a visual analysis method of mechanism research status based on topic model.Help users to understand the current research situation of the organization,find the most influential scientific research fields,and find the disciplinary advantages.This thesis mainly does the following research work:1.The research method of mechanism research status analysis based on topic model is proposed.The main process of this method is as follows: firstly,the collected SCI academic literature is cleaned to get the corpus of subject modeling;secondly,the subject of the corpus is extracted by the topic model algorithm,and then the literature is clustered;finally,the other dimension information of the literature data is fused with the extracted subject,and the research situation of the institution is visually analyzed and predicted from different perspectives.2.A visual analysis system of mechanism research status based on topic model is designed and implemented in combination with the method proposed in this thesis.This system consists of three parts: overall research trend,topic hierarchy and discovery of disciplinary advantages.The sub modules of each part contain different information dimensions and provide corresponding interactive operations.In this thesis,all these functional modules are integrated in the visualization system.The visual analysis system of institutional research based on topic model designed in this thesis provides a new idea for the application of visualization in the direction of academic data.The effectiveness and feasibility of this system are evaluated through the case analysis of real data sets.