- Paper (ICWSM'21): Embeddings-Based Clustering for Target Specific Stances: The Case of a Polarized Turkey
- Paper Presentation from ICWSM'21: papertalk
- Thesis (MSc August 2020): Embeddings-Based Clustering For Target Specific Stances
- We propose an unsupervised user stance detection method to capture fine grained divergences in a community across various topics. We employ pre-trained universal sentence encoders to represent users based on the content of their tweets on a particular topic. User vectors are projected into a lower dimensional space using UMAP, then clustered using HDBSCAN.
- Our method is able to capture stances to the party-affiliation level in a completely unsupervised manner.
- Given the resultant user stances, we are able to observe correlations between topics and compute topic polarization.
- We identify the most prominent terms in
each cluster to show how people talk about the same issue
in different contexts.
Note: This work was tested using umap-learn 0.3.x. Newer versions might not work as expected.