Cargo company recommendation study based on probabilistic linguistic term set

Abstract

The global economic structure is the main reason for changes in consumption habits and consumer behavior. Developing information technologies direct producers and consumers to e-commerce. Cargo services are an important link in the chain in the fast and effective operation of e-commerce. The growth in e-commerce has a driving force in the development of cargo services and cargo companies. Cargo companies can survive in global competition by being preferred by customers and increasing their number of customers. The change in the number of customers occurs by communicating the satisfaction or dissatisfaction with the cargo company to potential customers. This study deals with the preference levels of cargo companies serving in Turkey according to customer suggestions. The data obtained from the survey evaluations are processed and recommendation ranking calculations are made for cargo companies. Probabilistic Linguistic Term Sets (PLTS) are used to eliminate customer ambiguities in survey evaluations. Alternative cargo company recommendations are ranked based on the customers' past service experiences from cargo companies. Aras Cargo, MNG Cargo, PTT Cargo, Surat Cargo, UPS Cargo, Yurtiçi Cargo companies are evaluated according to price, personnel, speed, reliability and network attributes. The maximum deviation optimization method based on the Lagrangian function is used to calculate the weights of the cargo companies' attributes. The probabilistic linguistic cosine similarity method compares cargo companies pairwise under attributes and a similarity matrix is obtained for six cargo companies. The similarity matrix defines the alternative cargo company recommendation ranking based on customers' past experiences. UPS, SURAT and MNG cargo companies stand out as the most prioritized companies according to the evaluation results. The effects of attribute weights are observed by designing six different scenarios and it is observed that the differentiating attribute weights affect the recommendation ranking. Spearman correlation coefficient evaluation based on recommendation rankings indicates a high relationship between attributes.