User:Gor/emm

title: Europe Media Monitor

Research and development efforts at the Joint Research Centre produced novel and unique software that gathers reports from news portals world-wide in 43 languages, classifies the articles, analyses the news texts by extracting information from them, aggregates the information, issues alerts and produces intuitive visual presentations of the information found.

User-driven Multilingual News Gathering and Aggregation

The European Commission's Joint Research Centre (JRC) has developed a number of news aggregation and analysis systems to support EU institutions and Member State organisations. The three Web Portals NewsBrief, NewsExplorer and MedISys are publicly accessible and attract up to 1,2 Million hits per day. The system was initially developed as an in-house application for the European Commission’s Directorate General Communication (DG COMM) to replace their tedious and expensive manual media monitoring and press cutting services. Since then, EMM has become a crucial instrument in the daily work of almost all Commission services and many other public organisations. EMM is the news gathering engine behind a number of applications. EMM monitors the live web, i.e. the part of the web that has ever changing content, such as news sites, discussion sites and publications. All applications are developed, maintained and run by the JRC.

h2. EMM News Brief

The public web application NewsBrief shows the hottest topics discussed during the last few minutes and hours across multiple news sources from around the world in 43 languages. Updated every ten minutes, the site groups related news, shows timelines and displays the biggest stories first.

News by subject

NewsBrief categorises all news into hundreds of customer-oriented subject domains and according to the countries mentioned in the articles. Users can opt to see only news falling into specific subjects or concerning countries of their choice.

Customising the view of the news and tools to produce newsletters

Users can customise their view of the news by filtering the news by language and by selecting specific news sources. Registered organisations can additionally use the Rapid News Service RNS to further filter the news and to produce their own newsletters by grouping news items and by dragging and dropping them in an easy-to-use interface. With a simple click, email or PDF versions can be sent to defined user groups.

Vital statistics Monitors over 4000 sites from 1600 key news portals world-wide plus 20 commercial news feeds and, for some applications, also specialist sites.

Retrieves over 40000 reports per day in 43 languages. Classifies all news according to hundreds of subjects and countries. Access on the web, via email and by RSS. Runs 24 hours per day, 7 days a week.

h2. EMM News Explorer

Daily news summary

At the end of each day and for each of 19 languages, all news articles talking about the same event or subject are grouped into clusters and displayed by cluster size. Users can see the people, organisations and locations mentioned most in today’s news. A map shows where today’s events happened. The calendar allows to read about past events.

Collecting information about people, organisations and countries

To a large extent, news is about people, organisations and countries. NewsExplorer keeps track of who gets mentioned where and how often, who gets mentioned with which other persons or countries, etc. Up to 170 multilingual spelling variants for the same person have been detected automatically so that users can search for information about persons independently of the name spelling. Quotations by or about people are also displayed. The database contains information on close to 700,000 names, extracted and aggregated from tens of thousands of articles every day written in 19 languages.

Linking related news across 19 languages

The perspective on individual issues often differs from one country to the next. Seeing related news coming from different countries or written in different languages next to each other can be very informative. Using unique technology, NewsExplorer automatically detects related news clusters across languages and allows users - with one simple click - to access foreign language news about the same subject or event. For languages spoken in more than one country (e.g. English, German or French), users can additionally opt to read the news produced in only one of these countries.

Detecting trends over time

Each day’s news clusters are automatically linked to the related news of previous days and months. Timelines show developments over time and allow users to jump to the origin or to peaks of each story. Separate sections displaying news stories that started in the last week or in the last month allow users to get up to date with recent developments.

NewsExplorer highlights Produces daily news summaries in 19 languages.

Tracks topics and events over time. Calendar allows to find news from days in the past. Links related news across languages. Extracts multilingual information about people and organisations from the news. Allows to find information on people independently of their name spelling. Detections quotations by and about people.

Pouliquen Bruno, Ralf Steinberger, Camelia Ignat, Irina Temnikova, Anna Widiger, Wajdi Zaghouani & Jan Žižka (2005). Multilingual person name recognition and transliteration. Journal CORELA - Cognition, Représentation, Langage. Numéros spéciaux, Le traitement lexicographique des noms propres. Available online at: http://edel.univ-poitiers.fr/corela/document.php?id=490. ISSN 1638-5748.