CONFERENCE (INTERNATIONAL) Diverse and Non-redundant Answer Set Extraction on Community QA based on DPPs
Shogo Fujita (Tokyo Institute of Technology), Tomohide Shibata, Manabu Okumura (Tokyo Institute of Technology)
The 28th International Conference on Computational Linguistics (COLING2020)
December 07, 2020
In community-based question answering (CQA) platforms, one problem is that it takes time for a user to get useful information among many answers. Although one solution is an answer ranking method, the user still needs to read through carefully the top-ranked answers. This paper proposes a new task of selecting a diverse and non-redundant answer set rather than ranking the answers. Our proposed method is based on Determinantal Point Processes (DPPs), and calculates the answer importance and the similarity between answers using BERT. We built a dataset focusing on a Japanese CQA site, and the experiments on this dataset demonstrated the proposed method outperformed several baseline methods.
Paper : Diverse and Non-redundant Answer Set Extraction on Community QA based on DPPs (external link)