Rapid methods of text analysis are increasingly important tools for efficiently extracting and understanding communication within the food and beverage space. This study aimed to use frequency-based text mining and biterm topic modeling (BTM) as tools for analyzing how cider products are communicated and marketed on cider-producer websites for products made in Virginia, Vermont, and New York. BTM has been previously used to explore topics in small corpora of text data, and frequency-based text mining is efficient for exploring patterns of text across different documents or filters. The present dataset comprised 1115 cider products and their website descriptions extracted from 124 total cider-producer websites during 2020 and 2021. Results of the text mining analyses suggest that cider website descriptions emphasize food-pairing, production, and sensory quality information. Altogether, this research presents the text mining approaches for exploring food and beverage communication.
This research will be valuable to stakeholders in the United States' cider industry by providing relevant insight as to how cider marketing and sensory communication varies based on extrinsic product factors, such as geography and packaging. This research also demonstrates the efficiency and potential of text mining tools for exploring language and communication related to foods, beverages, and sensory quality. Further, this research provides a framework for extracting sensory-specific language from a large corpus of data, which may be adopted by other researchers wishing to apply rapid descriptive methods in the sensory, quality, and consumer research fields.