Research And Application On The Mapping Relationship Between Network Public Opinion Information And Real Transaction Behavior

With the rapid development of Internet and social media, the network public opinion information and real transaction behaviors are becoming highly integrated. Internet is evolving from a simple technology platform of information release to a major carrier of social media which becomes an interactive social network with special characteristics of information release, sharing, exchange and cooperation. Notably, it has greatly changed the way of us observing the society and economic. Along with the ceaselessly rising participation, people no longer passively obtain knowledge, but actively make various opinions and comments. Those opinions and comments can not only express people's true thoughts instanty, but also impact the activity of the real world through affecting the audiences' psychology. At present, researchers often study the mapping of real transactional behaviors based on the attention of network public opinion information and neglect the expression on the implied content of the network public opinion information on Internet and the transactional behaviors.Consider the limitation of cross discipline theory and information technology, it is mentioned a lot but empirically research much few. Aims to those problem above, this paper takes financial market as background and focuses on stock transaction behavior data and network public opinion information associated with the stock. Mapping relationship between network public opinion information and real transaction behaviors is independently studied based on influence of network public opinion information and temporal segmentation and anomaly of real transaction behaviors. Research on mapping relationship between them is divided into two aspects. On the one hand, we detect the events in network public opinion information and tracing events in each specific area, divide the events according to time granularity, then give measuring method of event influence, and turn events influence into time series, at last mapping relationship between network public opinion information and real transaction behaviors is discovered based on event influence. On the other hand, we segment time series and acquire anomaly in transaction behaviors, find out mapping relationship between network public opinion information and real transaction behaviors based on time series segmentation and anomaly, and identify causes of these rules and exceptions through mapping relationship with network public opinion information. The main contributions and innovations of this paper are listed as follows:(1) Aiming at the lack of ability of current methods on accurately detecting and tracking events in special areas, this paper presents a matching method of weighted maximum bipartite graph to track the events in special areas. It confines the weight of partial keywords by the association rules to remove the noises in special areas.(2) Mapping relationship between network public opinion information and real transaction behaviors is studied based on event influence of network public opinion information. This paper proposes a method to general measurement of event influence via combining the heat of event itself with the users participated in the propagation of event, which avoid the unreal result of influence caused by the false heat and useless. This method can also divide the event influence into the time series based on the time granularity, and find out the mapping relationship between network public opinion information and real transitional behaviors by the consistency of temporal correlation and space consistency based on influence.(3) Mapping relationship between network public opinion information and real transaction behaviors is studied based on time series segmentation. This paper proposes a method based on context-aware edge similarity algorithm to segment time series. The algorithm overcomes the machinery between traditional pattern recognition and traditional pattern matching and has stronger adaptive capacity in time series that produced by real transaction behaviors. The results of the experiment show that the method can reduce the error rate of segmentation caused by only considering the pattern matching and has better anti-interference performance in noisy environments and finds the mapping relationship between real transaction behaviors and network public opinion information more accuracy.(4) Mapping relationship between network public opinion information and real transaction behaviors is studied based on time series anomaly. This paper proposes a method of self-adaptive section to detect the anomaly with the consideration of time-varying characteristics and unpredictability. According to the characteristics of the data itself, it measures the data relationship and builds the self-adaptive section with the degree of separation, and filters the abnormal data based on the self-adaptive section. Finally, there are establishment and analysis on the mapping relationship between network public opinion information and real transitional behaviors based on the abnormal time series. The results of the experiment prove that the method of this paper not only effectively detects the abnormity of time series data, but also finds the mapping relationship between the network public opinion information and the real transitional behaviors, which can help the users know the reasons for abnormity.