An Unsupervised Graph-Based Hybrid Approach for Opinion Summarization

Huge volume of opinions is posted daily on the internet due to the rapid growth of merchant sites, social media sites and other discussion forums. This contribution creates an information overload scenario forcing analysts and bosses to look for efficient opinion summarization software but developing an optimal summarization system is an uphill task. In this paper, a hybrid approach towards opinion summarization is investigated. Extractive and abstractive summarization techniques are merged with structured summarization approaches to produce an informative, concise and readable summary of the opinionated text. An unsupervised graph-based approach is used for extractive and abstractive summarization. The results are compared with existing state-of-the-art summarization approaches using ROUGE-N. It is concluded that the proposed system produced more ideal summaries than the existing traditional approaches.