Identification And Analysis Of Hot Topics In Online Communities Based On Complex Complex Networks

While community governance is an important part of grassroots democracy,grid is a new model of community governance that is being carried out.Nowadays,adopting a new model of “Internet +” community governance based on grid,many grid communities have established an “Internet +” governance platform,allowing users to submit topics on the platform.Due to the complexity of the people in grid communities,the problems submitted by each person are different,and are closely related to their life,resulting in many different types and large quantity of,and complex topics.Therefore,filtering out the hot issues that the residents really care about and urgently need to be solved from these topics is particularly important for improving the efficiency of community governance.Focusing on the filtering,identification and analysis of hot topics from the topic data submitted by residents,the paper mainly conducts the following work and innovation:(1)A keyword-topic complex network model is established to identify and filter hot topics.The paper establishes a keyword co-occurrence network based on the keyword cooccurrence relationship and a topic similarity network based on the topic similarity relationship,loads the relationship between the two networks[1],and finally establishes a keyword-topic complex network model.The network can quickly establish models and effectively map the topic data,thereby helping to improve the efficiency of grid community governance.(2)A method of selecting topic words and calculating the hot value of topic words are designed.In data preprocessing,this paper proposes a short text classification optimization algorithm.Aiming at the problems of complex content,short text and small amount of data,this algorithm can better classify the data,filter out irrelevant topic data,and make the experimental data more accurate;in the extraction of keywords,this paper adopts an integrated algorithm to make up for the deficiency of a single extraction algorithm.A method of selecting topic words based on connected subgraphs and co-occurrence is proposed,and a set of hot value calculation formulas are designed to calculate the hot value of topic words to characterize the degree of attention of a class of topics.(3)The research contents of this paper are verified by examples.According to the background topic data of APP submitted by the grid communities in Qingdao,and with regard to the many strokes and pinyin errors in the data,this paper uses pycorrector to establish and train a Chinese error correction model for topics,and establishes a keywordtopic complex network model for specific topic data.Based on the above model and algorithm,this paper preprocesses and models topic data generated by the grid communities in Qingdao,selects topic words,and calculates topic word hot value.The experimental result shows that the keyword-topic complex network model and topic selection method are effective.