Using Word Embedding and Community Discovery to Understand the Market for Remanufactured and Refurbished Products

Abstract

This paper draws on word embedding and community discovery algorithms to allow researchers to easily and effectively gather information on markets for remanufactured and refurbished products. Off-the-shelf, pre-trained vectors created using word embedding are employed and neither input on remanufacturing or related concepts (e.g., data or corpus containing specific information on product remanufacturing) nor human interactions are required. The study’s outcomes have implications for academics and practitioners alike. First, environmental concerns only weakly appear in the discourse on product remanufacturing and refurbishing. Second, the terms remanufactured and refurbished are strongly associated with product type. Third, product remanufacturing is associated with industries such as car parts and printers, in both the B2B and B2C markets. Fourth, product refurbishing is associated mostly with consumer electronics and predominantly the B2C markets.Fifth, in the remanufacturing/refurbishing business, OEMs’ presence is more salient than that of independent remanufacturers. Sixth, the salience of Japanese and US OEMs is higher than that of OEMs based elsewhere. OEMs that manufacture printing equipment, cars and consumer electronics are the most salient. By contrasting the CSLC academic literature with the clusters obtained through big data analysis, this study also identifies products and brands that are understudied in the academic literature.