A text-mining and possibility theory based model using public reports to highlight the sustainable development strategy of a city

Nowadays, ecology and sustainable development are priority government's actions. In Europe, and more specifically in France, sustainable development (SD) is generally broken down into several distinct evaluation criteria. Each criterion is a requirement imposed by the government and corresponds to strategic stakes. When SD improvement actions are financed in an economic region or a city of the French territory by the government, a set of measures is usually set up to assess and control the impact of these actions. More precisely, these measures are used to check whether the region or the city has efficiently invested its budget in respect to the SD strategy of the government. This assessment process is a complex task for the government. Indeed, evaluations are only based on reports provided by the financed regions. These very numerous reports are written in natural language and thus, it is a thorny and time-consuming task for the government to efficiently identify the meaningful information in a plethora of reports and then objectively assess all the expected priorities. This project aims at automating the assessment process from the huge corpus of documents. Text-mining and segmentation techniques are introduced to automatically quantify the attention the region or the city pays to a given criterion. Obviously, this quantification can only be imprecisely determined. Then, the possibility theory is used to merge the information related to each criterion prioritization from all the documents. Finally, an application on the 265 largest cities in France shows the potential of the approach.