Key Node Identification And Link Prediction For Virtual Brand Community

With the rapid development of information technology and the increasingly developed network functions,cyberspace provides a good platform for people with the same interests and ideas to share information,exchange opinions and express their feelings through the network.Therefore,virtual communities have been set up one after another and led to the rise of virtual brand communities.From the perspective of consumers,virtual brand community enables consumers to share product experience,design ideas and so on through direct communication with enterprises or other consumers.From the perspective of enterprise operators,enterprises can reduce costs from various aspects by means of interaction between consumers in the virtual brand community.With the booming development of virtual brand community,more problems have been exposed: the number of users has increased,which is difficult to manage;The user participation is not enough,the community viscosity is low.Existing researches on "virtual brand community" focus on the aspects of member participation,value co-creation,brand loyalty,community cohesion,etc.There are few literatures on the identification of key nodes in virtual brand community and link prediction of virtual brand community which takes key nodes into consideration.But the key node identification of virtual brand community will help enterprises to better manage the community.The link prediction based on the key node of the virtual brand community will be beneficial to enhance the user's community identity and brand identity,so as to improve the user stickiness of the virtual brand community.Therefore,this paper mainly identifies the key nodes of the virtual brand community and studies the link prediction under the participation of the key nodes in the virtual brand community.On the basis of sorting out relevant literature,this paper combines theory with practice to construct network based on specific virtual brand community--"what is worth buying"(thirdparty platform)xiaomi(brand)community.The statistical results of the network show that the interactive network of virtual brand community constructed in this paper is scale-free and small world,and further analyzes and demonstrates the possibility of the existence of key nodes of virtual brand community.Aiming at the interactive network of virtual brand community,this paper analyzes the characteristics of members in the virtual brand community,describes the value of members from the two aspects of influence and loyalty,and puts forward the evaluation index of node individual value.Combined with the influence of nodes in the community,the key node recognition algorithm in the virtual brand environment is proposed.When identifying the key nodes,it is also found that some nodes have greater influence,but the removal of this node has less influence on the overall network,that is,the loyalty of this node is too low.Therefore,by combining the influence index,loyalty index and all the influences of nodes,it is more conducive to selecting the key nodes with great value in the community and facilitating the future management of enterprises.Based on the identification of key nodes in the virtual brand community,this paper combines the absolute node synthesis value of key node identification in virtual brand community with the tightness of node similarity network structure,further studies the link prediction process with the participation of key nodes in virtual brand community,and compares the prediction effects of different link prediction algorithms in the same situation.The results show that the link prediction algorithm proposed in this paper is more accurate for virtual brand community with the intervention of node comprehensive value.The innovation points of this paper mainly include the following points:?1 Proceeding from the reality,the data of "xiaomi"(brand)on the community platform of "what is worth buying" was collected,and the user interaction network in the virtual brand community was constructed by using the real data set.The influence and loyalty are combined on the basis of predecessors,and the influence of nodes is considered comprehensively,so as to select the key nodes of the virtual brand community.?2 Combined with the absolute index of key node identification of virtual brand community--comprehensive value of nodes,similarity of nodes and tightness of network structure,this paper explores the link prediction process in the environment of virtual brand community and compares it with other link prediction algorithms to provide suggestions for enterprises to guide users in the future.