Analysis Of Social Interrelationship With Multi-source Interactive Information

Whether in real life or in social networks,social relations are a very important part.Social relations can provide important basis for the study of downstream tasks such as friend recommendation,information dissemination and group discovery,so the study of social relations is very meaningful.In the modern society,where the pace of life is getting faster and faster,people's social activities are gradually shifting from offline to online,so it is more important to study online social relations.This paper attempts to model,calculate and predict the social relationship based on online social network platform.In the aspect of feature selection,this paper chooses four linguistic features and three topological features,and studies the correlation between linguistic features and topological features.The experimental results show that the number of public friends is proportional to the consistency of their views on the strength of their relationship.This paper also studies the interaction balance in the topology analysis,and analyses it from the static and dynamic perspectives respectively,and obtains more meaningful conclusions.Based on Enron's e-mail data and Sina Weibo's data,this paper uses SVM model,CNN model and RNN model to predict the relationship between superiors and subordinates and the gender of the two parties.The validity of the extracted features is verified,and the difference between using the two features alone and their combination effect is also verified.The experimental results show that linguistic information is more accurate than topological information in expressing and describing social relations,and the information contained in the combination of them is more abundant,and the effect is better than using any kind of information alone.