A text mining framework for advancing sustainability indicators

Assessing and tracking sustainability indicators (SI) is challenging because studies are often expensive and time consuming, the resulting indicators are difficult to track, and they usually have limited social input and acceptance, a critical element of sustainability. The central premise of this work is to explore the feasibility of identifying, tracking and reporting SI by analyzing unstructured digital news articles with text mining methods. Using San Mateo County, California, as a case study, a non-mutually exclusive supervised classification algorithm with natural language processing techniques is applied to analyze sustainability content in news articles and compare the results with SI reports created by Sustainable San Mateo County (SSMC) using traditional methods. Results showed that the text mining approach could identify all of the indicators highlighted as important in the reports and that the method has potential for identifying region-specific SI, as well as providing insights on the underlying causes of sustainability problems. Region specific sustainability indicators were identified by analyzing news.Text mining allowed the tracking and reporting of the sustainability indicators.Text mining provided insights into sustainability issues in San Mateo County, CA.Developed a document classification algorithm to handle overlapping indicators.Chronic problems considered less newsworthy proved more difficult to track.