CROSS-MEDIA DATA MINING

The Netherlands Institute for Sound and Vision is well-known about its large Radio and Television collection, consisting of audio and video data of the Dutch Public Broadcasters. But also, so-called context collections are available such as Viewing and Listening Ratings, Program Guides and all sorts of documents that were created during the production process such as auto-cues, contracts and scripts. Interrelating the information across these collections would be very interesting. For example, to increase knowledge about individual items in the collections (e.g., descriptions, popularity), but also to study representations of topics or people in different media. As we are collaborating with other institutions such as the National Library and Eye Film Institute (see e.g. Media Suite where you can access the data with your university login), interrelating collections is also interesting from a broader, cross-institutional perspective.

NISV and the users of the collections such as researchers and journalists are very interested in methods and techniques to mine data across collections that we have available. Example questions that we have are:

  • How can we use the Program Guides to add missing information in our audiovisual collections items or to estimate archive coverage over certain periods?
  • How can we combine information from the Radio and Television Ratings and the audiovisual archive to visualize the popularity of programs over time?
  • How can we link topics discussed in newspaper data to the same topics discussed on Radio and Television?

If you are interested to work on these or related questions please contact us and collaborate in our LABS!