A machine learning approach for analyzing Free JAR data. (July 2022)

• Record Type: Journal Article
• Authors: Luc, Alexiane; Lê, Sébastien; Philippe, Mathilde; Mostafa Qannari, El; Vigneau, Evelyne
• Is Part Of: Food quality and preference. Volume 99(2022)
• Publication Date: 2022-07
• Subjects: Free JAR -- Labeled textual data -- Valency score -- Interpretability -- Preference mapping -- Cheese; Food preferences -- Periodicals; Food -- Quality -- Periodicals; Food industry and trade -- Quality control -- Periodicals; Préférences alimentaires -- Périodiques; Aliments -- Qualité -- Périodiques; Aliments -- Industrie et commerce -- Qualité -- Contrôle -- Périodiques; Food industry and trade -- Quality control; Food preferences; Food -- Quality; Periodicals; 664
• DOI: 10.1016/j.foodqual.2022.104581
• Languages: English
• ISSNs: 0950-3293

Abstract

Highlights: The link between hedonic categories and Free JAR comments is investigated. This link is used to build classifiers, providing valency score for each comment. The concept of interpretability applied to classifiers highlights liking drivers. The general approach is illustrated by means of a case study pertaining to cheeses. Abstract: This paper presents a methodology based on a machine learning approach for the processing of Free JAR data. The development of this new analysis method is grounded on the fact that Free JAR data can be seen as labeled textual data, since a hedonic category is associated to each Free JAR comment. In particular, this two-step methodology aims to take advantage of the link between these two pieces of information, to disclose the products' assets and weaknesses. The first step consists of assigning valency scores to the Free JAR comments, considering their link to the hedonic categorization provided by the consumers. This is achieved by setting up a classifier based on Random Forest. The second step consists of highlighting the attributes that characterize the product under study based on the concept of interpretability. The interest of the proposed procedure is illustrated through a case study pertaining to cheeses.