Exploring User Comments on Online News
What is ComEx?
The comment sections of online news platforms have shaped the way in which people express their
To foster more interactive and engaging discussions, we present our ComEx system for the
exploration of reader comments on online news platforms.
Central to this system is that potential discussion participants get a quick overview and are
not discouraged by an abundance of comments.
About the Project
How it works
1. Build Discussion Graph
Each comment is split into snippets and represented by nodes in a graph.
Weighted edges are added based on structural or semantic features.
- Text similarity (embedding model)
- Reply to another comment
- Part of the same comment
- Timely related
- Cluster assignment
2. Reduce Discussion Graph
There are numerous edge generators populate the graph and filters to reduce the complexity.
You can have a look at the parameters in the API for /api/graph and play with different settings.
Generated graphs are typically highly connected and need to be simplified not only for better readability, but also since processing and transmission of the graph would be too slow.
3. Display Discussion Graph
The fully processed graph is shown in the user interface using a force layout.
Edges are omitted for better readability.
Julian Risch, Tim Repke, Lasse Kohlmeyer, Olena Vyshnevska and Ralf Krestel
ConvEx: Conversation Exploration for Users of Online News Platforms. 2020 (unpublished)
Julian Risch, Tim Repke, Lasse Kohlmeyer and Ralf Krestel
ComEx: Exploring User Comments on Online News Platforms. 2020 (unpublished demo)