A MULTI-AGENT BASED SOCIAL CRM FRAMEWORK FOR EXTRACTING AND ANALYSING OPINIONS

Abstract

Social media provide a wide space for people from around the world to communicate, share knowledge and personal experiences. They increasingly become an important data source for opinion mining and sentiment analysis, thanks to shared comments and reviews about products and services. And companies are showing a growing interest to harness their potential, in order to support setting up marketing strategies. Despite the importance of sentiment analysis in decision making, there is a lack of social intelligence integration at the level of customer relationship management systems. Thus, social customer relationship management (SCRM) systems have become an interesting research area. However, they need deep analytic techniques to transform the large amount of data “Big Data” into actionable insights. Such systems also require an advanced modelling and data processing methods, and must consider the emerging paradigm related to proactive systems. In this paper, we propose an agent based social framework that extracts and consolidates the reviews expressed via social media, in order to help enterprises know more about customers’ opinions toward a particular product or service. To illustrate our approach, we present the case study of Twitter reviews that we use to extract opinions and sentiment about a set of products using SentiGem API. Data extraction, analysis and storage are performed using a framework based on Hadoop MapReduce and HBase.

Keywords

• Big data
• Social CRM
• Hadoop
• Sentiment analysis
• Multi-agent system.