NOTICIAS: Improved recovery news and access to financial information: text retrieval on documentary sources of news agencies

This research project aims to provide advanced access to information contained in textual document bases of news agencies. A single event or piece of news is usually covered by many different sources. The presence of multiple documents reporting on the same subject causes that usual mechanisms and search systems provide redundant information. This is because the model “state of the art" of information retrieval tend to prioritize the importance of the texts for the user's query, ignoring important issues such as diversity.

Moreover the news redundancy between different or the same source there is the problem of locating the relevant information within a document. Users have to review in depth the contents of texts to meet their information needs. However, in many cases the information required by the user is a small extract from the news and the process carried out unnecessarily tedious.

Objectives

  • Investigate further models and techniques of information retrieval applied to textual news, resulting in an advanced system of access to information that works on the basis of news documentaries.
  • Design advanced search systems, able to retrieve passages within news, detect novelty with respect to other texts, build summaries and identify subtopics.