Objectives

Within the general framework of the project, the UniOr NLP Research Group (University of Naples "L'Orientale") aims to develop an application that uses computational linguistics and Artificial Intelligence technologies to detect tweets containing information regarding environmental crimes in action posted on Twitter by private individuals, NGOs or local organizations as well as tweets posted on the Twitter account of local newspapers.

Implementation and results

The research work started off with a focus on the Terra dei Fuochi (literally, the Land of Fires), an area in the South of Italy, sadly famous for being an illegal site for the disposal of toxic waste. The focus was then extended to the whole peninsula and included more types of environmental crimes.

A number of resources were created: i) the Environmental Crimes Glossary of Terms; ii) the UNIOR Eye corpus, consisting of 228,412 tweets including reports on environmental crimes downloaded from Twitter from 01/01/2013 to 06/08/2020; iii) an interactive map of Italy showing the locations of illegal fires; iv) a model for the analysis of tweets.

The first phase of the project concerned the creation of the Environmental Crimes Glossary of Terms: a glossary (with information on semantic field, definition and sources) related to various environmental crimes (waste and illegal fires, water and soil pollution, hazardous substances and materials, etc.), developed on the basis of various institutional linguistic and terminological resources.

The second phase of the project consisted in the development of the UNIOR Eye corpus, downloading from Twitter all the tweets containing the keywords in the corpus preceded by hashtags.The hashtags were used to collect the information needed to detect crimes against the environment. The corpus was subdivided into semantic areas, each one related to a specific environmental crime: waste and Terra dei fuochi; water crimes; hazardous materials and substances; fires and environmental fires. These areas are further categorised into more specific subsets, e.g. the water crimes area contains the subsets wastewater, sewage, polluted rivers, effluent.

In order to monitor the occurrence of environmental crimes reported in real time by private individuals, NGOs or local organizations on social media, a tweet analysis model ─ LiTEM (Linguistic Tools for Environment Monitoring) ─ was developed. With the use of advanced technologies of computational linguistics and artificial intelligence, this model is able to discriminate, among the flow of tweets, tweets that report an environmental crime in action (alerts) from those who do not perform this function (non-alert). Although non-alerts still deal with environmental issues, they do not perform a reporting function and generally take the form of tweets of political opinion, hate speech and so on.

The model also allowed the creation of an interactive map, where it is possible to visualize the location of all fires reported on Twitter from 2013 to 2020.

The innovativeness of the project consists in the use of advanced tools in the field of Artificial Intelligence to contribute to the protection of the environment through social media monitoring, favouring a timely action on site by the operators in charge of environmental protection. The project also aims to raise people's awareness and responsibility for the environment, highlighting the importance of citizen's active participation and the potential offered by a proper use of social media networks.

Project partners: Analist Group s.r.l. (lead partner), CIRA S.c.p.A. (technical-scientific coordination), Università degli Studi di Cassino e del Lazio Meridionale, Università degli Studi di Napoli “L’Orientale”, Expert Systems S.p.A., Major Bit Consulting s.r.l., AI Tech s.r.l., MapSat s.r.l.

Endorsement: the project has been considered of particular interest by SMA Campania spa (Campania Region's in house company for environmental monitoring) and ARPA Campania (Regional Environmental Protection Agency). The two organisations are members of the Project Advisory Board.