The Analysis On Scientific Collaborative Network Of Guangdong-Hong Kong-Macao Great Bay Area And The Collaborative Relationship Prediction

Nowadays,the information technology revolution represented by computers,the Internet and the World Wide Web has completely changed people’s information sharing,interconnection and other interactive ways.Various social information networks(social network)have emerged as the times require,such as Facebook,YouTube,Flicker,LiveJournal,Twitter,Douban,Sina Weibo,Tik Tok,etc.Social network has both the characteristics of information network and social network,which not only reflects the social relations among users,but also becomes the main carrier of information interaction among users.People can share news,pictures,videos,express personal opinions and comments anytime and anywhere,which leads to the rapid growth of data scale.It has high theoretical significance and practical application value to conduct qualitative or quantitative analysis and explore its common law and essential characteristics.Scientific collaborative network is an important network in social network,which describes scholars’ research fields,research affiliations,collaborative time and collaborative topics.The analysis of scientific collaborative network is an important way to understand scientific cooperation at regional,national,regional,university and individual levels.The relevant research results can be directly applied to expert recommendation,relationship prediction,scholar’s active degree,influence analysis and many other fields.Therefore,scientific collaborative network has become a hot research field.The Guangdong,Hong Kong and Macao Great Bay Area is one of the districts with the highest degree of openness and the strongest economic vitality in China.It has an important strategic position in the overall situation of national development and has been highly regarded by the state.The analysis of scientific collaborative network in the Great Bay Area can help us understand the current situation of scientific cooperation of the Great Bay Area and provide decision support for Great Bay Area discipline development,which has important research significance and application value.This thesis mainly studies the scientific collaborative network of the Great Bay Area,explores its collaborative mode from different angles,and predicts the collaborative relationship.Specifically,the main contribution of this thesis includes the following two aspects:(1)The scientific collaborative network of Guangdong,Hong Kong and Macao is analyzed and the cooperative mode is excavated.Firstly,we grasped the information of papers of Great Bay Area.which published in the computer field.Then we constructed the scientific cooperative network of universities in the Great Bay Area.On this basis,we deeply explored the situation of cooperation among universities in the Great Bay Area.We also analyzed the characteristics and modes and the impact of papers published by universities(or affiliations)in Great Bay Area.The number of cooperation between universities is increasing,the cooperation between universities in the Great Bay Area is developing rapidly,and the trend of integration is more obvious.(2)Scientific cooperative relationship prediction based on attention mechanism in deep learning.Based on the analysis of the scientific collaborative network of universities in the Great Bay Area,we combined the structural information of the network and the attribute information of the nodes to analyze and predict which universities will have cooperation in the future.Firstly,the dynamic heterogeneous scientific collaborative network is constructed,and the data model of the network is built.The matrix of cooperation based on the influence of the paper and the matrix of cooperation based on the number of cooperation are obtained.Based on the traditional neural network,we proposed an attention and deep learning model to predict the cooperative relationship.Experiments on real data and comparisons with mainstream methods show that the proposed model has higher prediction accuracy.The model can effectively combine the topological information of the network and the attributes of the nodes in the network to model the real scientific collaborative network problem,which provides a new way to predict the scientific cooperative relationship among universities.