A study on clustering customer suggestion on online social media about insurance services by using text mining techniques

Now a day Social media communication become to important factor for business operation. Several Customer prefers to post their comment, suggestion, complaints about company's products and services to online media such as Facebook, Twitter, Social web board because it easy way to blast to public and increases pressure to product owner for responding. This is one factor that cooperate need to be concern and manage responding to customer services that match to customer requirements by analyzes customer suggestion on social media vice versa they can detect negative feedback or complaints early which, able to prevent their reputation. This study was collected text that contains customer suggestion on insurance services from various online social media and extract some specific word via Thai text segmentation and coverts text to Vector Space Model (VSM) based on TF-IDF. We performs experiment by used 800 records of textcrawler and implement two clustering models algorithm which include K-Means and Self-Organization Map (SOM) for clustering suggestion text into three cluster groups as follow Cluster_0 is about to customer feedback on Car Insurance Policy, Car Insurance Premium or Insurance Renewal, Cluster_1 is contains customer feedback on insurance claim services, Cluster_2 is about customer enquired general information. We use “Davies-Bouldin index” method[3] for evaluating both clustering algorithms. A result of experiment shows that K-Means has a significant performance higher than SOM. Finally, The benefit of this study able to help insurance company improve their products and services and increase customer satisfaction and retention strategies planning.