An end-to-end ranking system based on customers reviews: Integrating semantic mining and MCDM techniques

Considering customer reviews is one of the challenges of real-world decision models. These reviews can be on different platforms and include a large amount of information and may also include incomprehensible and unrelated phrases. The main advantage of customer reviews is the realistic view that it provides a realistic view of the product or service. Therefore, converting unstructured and incoherent customer-based reviews into machine learning language and ultimately turning it into a decision model is very important. In this paper, we propose an end-to-end ranking method for integrating mechanisms such as text processing, sentiment analysis and the multi-criteria decision-making technique. The proposed ranking method relies on the integration of three methods, namely, the aspect-based sentiment analysis (ABSA) method, the Dawid-Skene algorithm and the Best Worst Method (BWM). In other words, the proposed work encompasses four major steps: i) crawling customer reviews, ii) preprocessing, iii) aspect term extraction, aspect category detection and polarity detection, and iv) designing a decision-making model. The main contribution of this study is to consider ABSA at three levels simultaneously and integrate ABSA and BWM in designing an end-to-end ranking method for ranking the quality of hotel services, facilities and amenities based on customer reviews. The ability of the proposed end-to-end ranking framework is evaluated using a real data set of user reviews of Sydney hotels.