At a glance
The OPTIMA action provides scientific and technical support to EU policies in the field of global security and crisis management. The objective is to improve the EU’s prevention, preparedness and response capabilities to a wide range of threats, including humanitarian crises, natural and man-made disasters, endangerment of public health, conflict, terrorism and organised crime. OPTIMA achieves this by providing users with the results of an automatic analysis of open sources from the internet (mostly textual traditional media and social media) in order to support users in their need to monitor information, detect threats, follow trends and analyse impact.
The beneficiaries (or ‘clients’) of our work include all EU institutions (e.g. the Council and the European External Action Service), EU agencies, as well as national authorities in the individual EU Member States (e.g. for public health and law enforcement). OPTIMA also works closely with international organisations (e.g. the United Nations and the Global Health and Security Action Group) and with regional organisations (e.g. the African Union and the Pan-African Parliament). Additionally, over 30,000 internet users per day access the public web pages of OPTIMA’s main product, the Europe Media Monitor (EMM), consisting of NewsBrief, the Medical Information System MedISys, NewsExplorer and EMM-Labs.
Due to the highly multilingual setting of the EU – there are 23 official EU languages – and the EU’s involvement with partners world-wide, OPTIMA focuses on highly multilingual applications and on tools that allow cross-lingual information access and multilingual information aggregation. EMM analyses an average of 150,000 online news articles per day from thousands of sources in about fifty languages.
The technologies used in OPTIMA are drawn from Computer Science and Computational Linguistics (also referred to as Language Technology, Natural Language Processing or Text Mining). More specifically, we use techniques from web technology, machine learning, statistics and symbolic (rule-based) language processing. The JRC-developed text analysis components used in EMM include: information extraction; named entity recognition and disambiguation; co-reference resolution; geo-tagging; event extraction; quotation recognition; the automatic generation of social networks; machine translation; document clustering; multi-label document classification; opinion mining (sentiment analysis); topic detection and tracking within the same and across different languages; fusion of information extracted from different languages. There is an extensive list of scientific publications describing these tools in detail.
OPTIMA also supports R&D efforts by releasing publicly large-scale multilingual language resources produced in-house or provided by other EU organisations. These include highly multilingual parallel corpora, translation memories, categorisation software, name variant lists and more.