Article Preview
Top1. Introduction
Internet public opinion is mainly composed of two parts: the Internet and public opinion. It is a new concept gradually formed with the development of computer information technology, the emergence and popularization of the Internet, and more and more people participating in network activities. The initial social public opinion exists in a small range in personal minds and oral communication among many people. Therefore, the initial method of obtaining public opinion information can only be obtained by integrating it into the people's lives and conducting public opinion surveys. This method of acquisition is very inefficient. Samples are small, expensive, and have obvious regionalization and cannot be widely spread. The Internet has many characteristics such as openness, globality, extensiveness, richness, diversification so that Internet public opinion is often spread in an explosive state of communication, such as a small event, plus emotional opinions. It is often easy to attract the attention of the public, become a hot spot of public opinion, and trigger the butterfly effect. Therefore, relevant departments need to quickly grasp the relevant public opinion trends and monitor and guide the online public opinion through the network public opinion monitoring system.
According to the development process of network events, to enable the network public opinion monitoring system to influence the event at the “attention” stage, the most critical process is first to find hot topics from the mass Internet public opinion information, and then based on the hot topics relevant departments and units have targeted and effective control of the development of public opinion (Xu J, Ma B., 2014, Fan D.P. et al, 2015). The key technology for hot topic discovery is the selection and application of similarity calculation between texts and clustering methods. SinglePass algorithm is a classic unsupervised incremental clustering algorithm, which is often used in the discovery of hot topics (Cody E M et al. 2015, Liesbeth Mollema et al. 2015). However, in modern times, Internet public opinion information is exploding, and the amount of data is substantial. It is challenging to meet the monitoring and management needs of Internet public opinion relying on the traditional single-pass SinglePass algorithm. Cloud computing is a new computing model. It uses the Internet to meet the needs of different users and to share and access the original resources in resource pools (such as computer processors, storage devices, and applications) on-demand and conveniently. At the same time, users can quickly release the occupied resources for later use by other users in need (Samaresh Bera et al. 2015). This technology can process GibaByte(GB) or even terabytes of data in just a few seconds, so by using advanced cloud computing technology. It can improve the processing process of the traditional hot topic discovery SinglePass algorithm, and quickly and efficiently find hot topics (Yumin Wang et al. 2019).
It is of great significance to study the information dissemination based on user relationships in a social network environment. Ji-Ming analyzedthe development of information dissemination models, including non-linear propagation based on user relationships, low threshold and balanced distribution, information cascading, and value-added propagation. Based on the network information dissemination model, a social information dissemination model was proposed and explained from the perspective of user relations and social media. This method is very accurate, but it is not efficient to implement (Ji-Ming, 2015). He et al., introduced a public opinion monitoring model based on the cloud computing environment. This model can mine and analyze data collected on a large scale, detect and track hot topics, conduct social network analysis on forums, and visualize the analysis results. The cloud-based public opinion monitoring system can provide sensitive information in a timely manner and effectively respond to public crises (He Z T et al, 2013). The traditional Internet public opinion monitoring methods have some shortcomings, such as lack of strong data collection and analysis capabilities. To solve this problem, Li proposed a cloud computing framework for Internet public opinion monitoring. The specific method is to use MapReduce's massive data processing capabilities to implement a new network information acquisition method (Li G et al,2013).