Epidemiological analysis is currently a very important issue especially in the case of public health policy. The goal Epimining is to reveal the impact of an epidemic in the media and to classify documents according to their content. Realised in partnership with the INVS (Institut National de Veille Sanitaire) Epimining complements traditional health information and provides an analysis of the spread of an epidemic in the media.
Initially news are retrieved via queries containing keywords associated with a disease (eg, "H1N1", "swine flu", "swine flu", "flu", etc..). Different classification algorithms based on patterns extracted automatically and the expert are used to classify documents.
News used a variety of information that can be associated with different classes (eg death, balance, new case ...). Epimining identifies the segments corresponding to the different classes in news. Figure 1 illustrates the display of different information from the documents: location-based classes, statistics, ...
Figure 1 - Epimining
Last update on 20/01/2015