Topic Models: Post-Processing and Applications

The main objective of the workshop is to bring together researchers who are interested in applications of topic models and improving their output. Our goal is to create a broad platform for researchers to share ideas that could improve the usability and interpretation of topic models. We expect this will promote topic model applications in other research areas, making their use  more effective.

Workshop at a Glance:

Topic modelling is a popular tool for modelling document collections and has seen applications in a variety of domains, from medical science to digital humanities. Topic models provide an effective means of extracting topics from large document collections and have the advantage that they require no supervision or document labels. While topic models can be trained automatically, judicious processing of their output is often required in applications where human users interact directly with the topics generated by a model. For example, the raw output from topic models can be difficult to interpret and consequently labels (manually assigned or automatically generated) are often added to increase readability; visualisation of the relationship between documents and topics is not intuitively obvious and its design requires thoughtful consideration. In addition, topics can contain unrelated words which may make the navigation of the document collection difficult; a typical approach to tackle this is to filter out low quality topics by computing topic coherence. This workshop aims to bring together a wide range of tools and methodologies used in post-processing the output of topic models, with the goal of improving their usability and interpretation within user applications.

Of particular interest are ideas that relate to:
  • Topic coherence

  • Extrinsic topic models evaluation

  • Topic representation (generation of topic labels)

  • Topic similarity

  • Applications of topic models

  • Visualisation of document collections with topic models (topic browsers)

  • Multimodal topic models

  • Data sets for evaluating topic qualities, including coherence, labels and similarity

Important Dates:

  • Submissions Due: 
    • June 5June 19, July 6 2015 (Extended)
  • Acceptance Notification: July 17, 2015
  • Camera-ready Due: July 31, 2015
  • Workshop Date: October 19, 2015