Article Preview
Top1. Introduction
The social network is a highly complex and interconnected structure of online communities that offers an interactive platform to exchange ideas and ideologies. The socially interacting networks are regarded as complex not only due to the number of profiles (nodes) and links (edges) but because of the existence of profile-centric features (aspects) and language associated (Ji et. al., 2020; Lu et. al., 2014). Such association requires context-based language analysis to distinguishes the profiles as well as the nature of links connecting those online profiles. For instance, profile name, date, time, posts, etc. are regarded as multiple aspects in the social network. Social network analysis mainly targets the nodes and edges linking in any specific context. It focuses on the type of connection between the nodes and the relationship defining how the nodes are interconnected with each other. We use nodes to identify contexts, however, one may also use contexts to identify nodes. Analyzing the contextual information helps to gain better insights that can be used for measuring, monitoring, and evaluating contents to improve the operability of any organization.
Behavioral analysis, also generally referred to as sentimental analysis, combines the usage of text processing and mining techniques to thoroughly evaluate the subjective mood of users about extracted contexts (Pradhan et. al., 2020; Kumar et. al., 2020). It represents the process of analyzing people's sentiments or attitudes towards any product, policy, topic, or event. Any shift in the online opinion of people in different contexts directly affects the performance parameters of large-scale social, political, or profit organizations. Behavioral analysis is one of the most buzzed filed in social network analysis. With the increase in social networking sites, e-commerce sites, and online platforms for expressing user opinions like Facebook, Twitter, etc. competition among the products and policy has increased a lot. User satisfaction and user loyalty is the utmost concern of the policy and product makers for assuring the quality of service. Business analysts, customer support personnel, human resource managers, company directors, and other stakeholders make use of sentiment analysis to understand the problems and challenges being presently confronted. User feedback has been taken very seriously for the quality improvement of the product and policy. This could further assist in creating space for designing a wide range of feasible solutions for organizational progress. Behavioral analysis is applied to user feedback, comments, reviews, or any other text message to identify the polarity of the message as positive, negative, or neutral. Such analysis is generically performed on the document level as a whole, thereby resulting in overall polarity which leads to very high sparsity in the result. To improve this problem, our research is focussed on contextual behavioral analysis (Zhao et. al., 2018).
One of the significances of analyzing user sentiment is to perform segregate textual content into smaller parts with polarity. This performs classification of the associated sentiment of text as positive, negative, or neutral. There are several graphical ways of representing the sentimental aspects, of which signed graphs provide richer visualization. A signed network belongs to the network theory that assists in modeling real-world phenomena. A signed edge in the social network graph represents an edge with a positive, negative, or neutral sign. Social networks offer a rich blend of positive and negative connections. This network architecture helps in identifying the pattern of interaction that is only feasible by analyzing the data with finer granularity. In the present day scenario, besides understanding the thought process of people regarding newly launched products, existing policies, etc.; it is equally important to analyze feedbacks for each of the associated features and classify whether the experience is positive, negative, or neutral. Some of the significant terms related to our research are highlighted as follows: