An integrated personalized decision approach with probabilistic linguistic context for grading restaurants in India

Online reviews from the web are rich data sources that promote tourism analytics. Restaurants make a significant contribution to the growth of tourism in India. The literature studies on restaurant selection show that most decision frameworks do not handle uncertainty effectively and pay subtle attention to heterogeneous sources. Additionally, the extant models (i) cannot accept missing entries and its imputation; (ii) reliability of data source agents are not methodically determined; (iii) attributes’ interactions are not properly considered; and (iv) personalized restaurant ranking is unavailable. The research problem considered in this study is the rational selection of restaurants based on online reviews from heterogeneous sources to support travelers in the tourism process. The main objective of this study is to circumvent the challenge in the literatures by proposing a novel integrated decision framework that collects data from heterogeneous rating sources and transforms them into ‘probabilistic linguistic information (PLI)’, which effectively handles uncertainty by relating occurrence probability to each linguistic term. Due to the uncertain nature of online reviews, missing data are inevitable. For rational imputation of data, a case-based method is proposed. Later, the relative significance of each attribute and the reliability of each rating source are determined using ‘criteria importance through intercriteria correlation (CRITIC)’ and Dempster–Shafer-based Bayesian approximation methods. Furthermore, the PLIs from each source are aggregated by using the newly proposed discriminative weighted Muirhead mean operator. Personalized prioritization of restaurants is achieved by using the newly proposed probabilistic linguistic comprehensive (PLC) method that acquires expectation queries from customers. Lastly, the practicality of the developed framework is testified by a real-case example of restaurant selection based on the data collected from online sources via web crawlers. Results infer that (i) the proposed framework is innovative/original, personalized, significant, and mitigates human intervention compared to the extant models, (ii) robust in terms of ranking of restaurants even after adequate weight alterations, and (iii) finally, supports stakeholders to effectively plan their tourism process and attain win-win conditions for effective growth of hospitality sector. •Web crawlers are used to extract data from online sources.•Web source data a