Development and testing of the media monitoring tool MedISys for the monitoring, early identification and reporting of existing and emerging plant health threats

Abstract

MedISys is a media monitoring system initially intended for news items related to human health. The tool has how been extended by the Joint Research Centre, Universitat de Lleida and IRTA to also deal with plant health threats. This EFSA-funded project was based on a knowledge representation approach that generated an ontology, a formal representation of knowledge related to plant health threats. The ontology models plant pests and diseases, together with other concepts related with them: affected crops, hosts, vectors and symptoms. First of all, a collection of news sources related to plant health threats was collected to be monitored by MedISys. These sources included already known manually curated Web pages but also additional ones discovered by performing global Web searches using terms appearing in the ontology. Then, the news items coming from these sources were filtered using MedISys using a set of categories with keywords to identify those actually about plant health threats. Most of these categories focused on known threats and used terms associated with the 117 pests and diseases selected at the beginning of the project. Additionally, categories for unknown threats were also developed. In this case the categories included keywords that are usually used by experts to describe unknown threats and keywords related with symptoms expressions. All these MedISys categories combined provide mechanism to monitor plant health threats mentions in media, from newspapers to social media, ranging from those that explicitly mention a named threat (useful to monitor re-emerging threats or their spread) to those related to unknown ones (to monitor potential new threats). The project concluded with an evaluation of the e-mail alerts and reports generated by MedISys based on the previous categories. A survey and tests with real users were conducted and the results analysed to generate a set of recommendations and improvements to facilitate the use of MedISys as a plant health threats monitoring tool.

Horizon Scanning for Plant Health: report on 2017‐2020 activities

“…Additional sources, both from specialist websites and general news media, were added and a total number of 131 categories corresponding to pests or groups of pests based on a multilingual ontology 3 were defined. Therefore, a first plant health threat ontology with pest and disease names coming from sources such as UniProt Taxon (The UniProt Consortium, 2016), European and Mediterranean Plant Protection Organization (EPPO) (EPPO 2018), and Wikipedia was built (Alomar et al, 2015(Alomar et al, , 2016. Multilingual keywords were added to enrich the categories.…”

“…For editing and publishing newsletters, a customized version of the Newsdesk tool was set up for EFSA. Furthermore, daily e-mail alerts were put in place that point to recent reports on plant pests (Alomar et al, 2016).…”

“…Sinoxylon spp.) or for identifying media and scientific articles about emerging plant health threats.3 As defined inAlomar et al, 2016, an ontology formally represents knowledge within a domain using a concrete vocabulary to denote the concepts, properties and interrelationships among them. In this report ontology indicates all categories corresponding to pests or groups of pests.…”

PADI-web 3.0: A new framework for extracting and disseminating fine-grained information from the news for animal disease surveillance

“…The definitions and concepts from both ECDC and WHO were elaborated for public health. However, they have been successfully transferred to other domains, such as plant health [ 7 ] and both terrestrial and aquatic animal health [ [8] , [9] , [10] ]. Both EBS and IBS can be formally represented as consecutive steps, corresponding to the flow of epidemiological information from its detection to its communication to the relevant authorities [ 4 , 11 , 12 ] (see Fig.…”

EFSA's Activities on Emerging Risks in 2016

“…Unfortunately, MedISys is a search engine based on keywords and not word meanings so it seems not possible to go beyond this relevance ratio. In any case, the work carried out might be an interesting starting point for future studies on how symptoms are expressed across pests and diseases or affected crops to generate a detailed network of plant health threats (Alomar et al, 2016).…”