Research On Forecasting The Number Of UGC In Online Question Answering Community Based On Time Series

The popularity of mobile internet technology has made User Generated Content(UGC)develop rapidly in recent years,and has achieved coverage in all fields and in all directions.In this context,the carrier of user-generated content,the online UGC community,has also evolved and matured.The online Q&A community,a new type of online community that evolves from the traditional Q&A platform with the UGC model,is a typical UGC-derived application.For the online Q&A community,UGC content publishing behavior is a direct manifestation and representation of community users' activity.Analysis and excavation of community users' active laws and encouraging users to create UGC content are important for platform operation.Evaluating and predicting the quantity and quality of user UGC published content is one of the key tasks to assist community operation management.The research in this paper aims to build a relatively better forecasting strategy to improve the forecasting efficiency and accuracy.In this paper,a UGC quantity prediction model based on time series is established,and the performance of the model is tested using user data in Zhihu,a large online Q&A community.Finally,a higher prediction accuracy is obtained.First,based on the characteristics of the UGC quantity prediction problem,this paper analyzes the applicability and limitations of the forecasting strategy from the theoretical and application levels,and selects the time series method as the main forecasting method.At the same time,by analyzing the characteristics of the online Q&A community and combining the past research results,this article regards the UGC content output of the online Q&A community users as a kind of knowledge contribution behavior.On the basis of summarizing the past research on simulating the willingness to contribute knowledge through hidden Markov model,this paper proposes to use hidden Markov model to mine more features of such time series to predict the number of UGC releases.Finally,this paper combines the idea of transforming time series into supervised series,and uses the nonlinear integrated learning model XGBoost to predict the number of UGC releases.Compared with the traditional time series analysis method(ARIMA model)for the prediction of UGC quantity,the method proposed in this paper has obtained the improvement of prediction accuracy and efficiency.