User Centric Mobile Based Decision-Making System Using Natural Language Processing (NLP) and Aspect Based Opinion Mining (ABOM) Techniques for Restaurant Selection

Abstract

Opinions about restaurants from diners who visit them are essential in restaurant selection process. Crowd-sourced platforms bridge both diners and restaurants through opinions. These platforms suggest straightforward rating based options; however, they do not address the confusion among diners in differentiating restaurants as a rating is a diverse characteristic. Diners resort to reading reviews as a solution since it has a human touch over calculated ratings. Nevertheless, reasons such as prolong time to read, discrepancies in reviews, non-localized platforms, biased opinions and information overload have obstructed diner perspective in picking a restaurant. This research proposes a mobile based intelligent system that collaborates Natural Language Processing (NLP) and Aspect Based Opinion Mining (ABOM) techniques, to assist user centric decision making in restaurant selection. A third party public data provider was used as the data source. The research automates reading of reviews and processes each of them using a custom algorithm. Initially, the paper focuses on a Part-of-Speech (POS) Tagger based NLP technique for aspect identification from reviews. Then a Naïve Bayes (NB) Classifier is used to classify identified aspects into meaningful categories. Finally, NB based supervised and lexicon based unsupervised sentiment allocation techniques are used to find the opinion from categorized aspects. Decision making accuracies and variations were measured over three main areas: aspect extraction, aspect classification and sentiment allocation. Overall results are in acceptable level; however unsupervised method always outperformed supervised method during sentiment allocation. It was proven that NLP and ABOM techniques can be used for the restaurant domain.