Corporate Behavior Recognition And Competitive Research Based On Social Media

Faced with the rapidly changing market environment,business competition is inten-sifying.Companies not only need to focus on the development of their own capabilities,but also pay attention to the behavior of competitors in the market at any time.Monitor-ing the behavior of competitors in the market is an important part of strategic decision-making[1,2].The behavior of competitors may even affect the company's strategic choices.However,it is difficult to obtain complete and timely information on corporate behavior in the traditional way.Fortunately,more and more companies maintain a timely con-nection and communication of information with their stakeholders through social media platforms,resulting in a large amount of enterprise-generated content,which contains a wealth of corporate behavior information,has important business value and research value.Therefore,this paper first identifies corporate behavior from massive social media data,and then studies business competition based on corporate behavior in social media.However,identifying corporate behavior from massive social media data and study-ing business competition is not easy.There are the following challenges:1)Corporate behavior in social media lacks a clear definition;2)It's difficult to extract information from unstructured social media data;3)The dynamic competitive relationship of com-panies lacks index measurement.In order to cope with the above challenges,this paper first uses algorithms to identify corporate behaviors from massive social media data,and then identifies corporate dynamic competitive relationships based on corporate behaviors in social media,and finally analyzes the behavioral evolution of the company itself or competing companies.The specific solutions are as follows:Firstly,corporate behavior recognition based on massive social media data.This arti-cle clearly defines corporate behavior based on the combination of trigger words in social media,and proposes a text trigger word feature optimization algorithm(FAE).Based on the optimized corporate behavior trigger words,this paper further proposes a constrained non-negative matrix factorization algorithm(CNMF-Info)that uses side information to identify corporate behaviors in massive social media data.Secondly,competition research based on corporate behavior.This article first defines a dynamic competitive relationship based on corporate behavior and builds a corporate-behavior time series matrix from three dimensions:time,corporate,and behavior,and then uses Pearson coefficients to quantify the degree of competition of the companies,and ranks the competition index to obtain the company unique and dynamic competitive relationship,and finally analyzes the behavior evolution process of the company itself or among competing companies.In summary,this article takes the mobile phone market as an example to study busi-ness competition based on the behavior of mobile phone companies in social media.The research results show that:1)The FAE proposed in this paper can not only reduce the cost of expert labeling,but also effectively improve the classification effect of CNMF-Info;2)The CNMF-Info proposed in this paper can effectively identify corporate behaviors from massive social media data;3)The competition research based on corporate behavior pro-posed in this article can help companies clarify the competitive environment,realize cor-porate behavior strategy monitoring,and provide support for the formulation of corporate competition strategies.