We present the system we built for participating in SemEval-2016 Task 3 on Community Question Answering. We achieved the best results on subtask C, and strong results on subtasks A and B, by combining a rich set of various types of features: semantic, lexical, metadata, and user-related. The most important group turned out to be the metadata for the question and for the comment, semantic vectors trained on QatarLiving data and similarities between the question and the comment for subtasks A and C, and between the original and the related question for Subtask B.
@article{arxiv.2109.15120,
title = {SUper Team at SemEval-2016 Task 3: Building a feature-rich system for community question answering},
author = {Tsvetomila Mihaylova and Pepa Gencheva and Martin Boyanov and Ivana Yovcheva and Todor Mihaylov and Momchil Hardalov and Yasen Kiprov and Daniel Balchev and Ivan Koychev and Preslav Nakov and Ivelina Nikolova and Galia Angelova},
journal= {arXiv preprint arXiv:2109.15120},
year = {2021}
}
Comments
community question answering, question-question similarity, question-comment similarity, answer reranking