Automatic Extraction of Suppliers’ ESG Compliance Information from Textual Sources: A Literature Review

This paper presents a literature review regarding the automatic extraction of meaningful information regarding suppliers’ ESG and sustainability compliance from textual sources. Assessing suppliers’ ESG compliance has become a key challenge for procurement managers. Given the large number of suppliers and required data points, traditional approaches such as questionnaires and audits are inefficient, ineffective and difficult to scale. To solve this problem, we investigate whether the required information can be automatically harvested from suppliers’ textual sources. Our structured literature review identified 82 papers on which we performed a descriptive analysis, finding a rich and flourishing body of literature produced by a heterogeneous scientific community. We further reduced our sample to 73 full-text articles that supported a more in-depth content-based analysis. We investigated which data sources can be used in particular, which technologies can be leveraged, and which types of outputs can be generated. Even though they could provide much of the required information, corporate websites are rarely utilized as data sources, partly due to the limited adoption of large language models (LLMs). LLMs are less diffused than traditional Natural Language Processing (NLP) techniques due to their recent introduction and some gaps that still limit their performance. This represents both a constraint and an opportunity for future research.