Projekt

Zur Projekt-Website

This project combines the latest scientific findings in natural language processing and information visualization in a powerful exploration tool that provides accelerated access to text document collections for investigative data-driven journalism. Whereas today's data-driven journalism is mostly based on structured data, i.e. data available in spreadsheets or databases, this project applies the principles of data journalism to unstructured text documents. Using statistical methods of language processing, such as named entity recognition and keyphrase extraction, important elements from these text documents can be identified and visualized in a network. A journalist can browse this network to quickly grasp the content of the text document collection, view the developments of groups of entities over time, annotate the elements of the network with labels and comments to attach a layer of interpretation to the data, and eventually publish part of the network along with an (online) article that is developed in a data-driven fashion from the findings in the document collection. The project consortium combines complementary expertise from "Der Spiegel" and the TU Darmstadt: investigative journalism, information visualization and text mining.

• Prof. Dr. Chris Biemann

  Technische Universität Darmstadt
  Fachbereich Informatik
  Sprachtechnologie
  Darmstadt
  

• Kathrin Ballweg

  Technische Universität Darmstadt
  Fachbereich Informatik
  Grafisch Interaktive Systeme
  Darmstadt
  

• Dr. Alexander Panchenko

  Technische Universität Darmstadt
  Fachbereich Informatik
  FG Language Technology
  Darmstadt
  

• Marcel Rosenbach

  Der SPIEGEL
  Berlin
  

• Dr. Michaela Regneri

  Der SPIEGEL
  Hamburg
  

• Dr.-Ing. Tatiana von Landesberger

  Technische Universität Darmstadt
  Fachbereich Informatik
  Interactive Graphics Systems Group
  Darmstadt
  

• new/s/leak Information Extraction and Visualization for an Investigative Data Journalists

• new/s/leak A Tool for Visual Exploration of Large Text Document Collections in the Journalistic Domain

• new/s/leak Anforderungsanalyse einer interaktiven Visualisierung für Data-Driven Journalism.

• Guidance for Multi-Type Entity Graphs from Text Collections