Better Early than Late: Fusing Topics with Word Embeddings for Neural Question Paraphrase Identification
This work addresses the need for more accurate duplicate question detection in online communities, though it is incremental as it builds on existing neural and topic modeling approaches.
The paper tackled the problem of identifying duplicate questions in Community Question Answering by proposing a neural architecture that merges topics with word embeddings through early and late fusion methods. The results showed that their system outperformed neural baselines on multiple datasets, with early fusion being particularly effective.
Question paraphrase identification is a key task in Community Question Answering (CQA) to determine if an incoming question has been previously asked. Many current models use word embeddings to identify duplicate questions, but the use of topic models in feature-engineered systems suggests that they can be helpful for this task, too. We therefore propose two ways of merging topics with word embeddings (early vs. late fusion) in a new neural architecture for question paraphrase identification. Our results show that our system outperforms neural baselines on multiple CQA datasets, while an ablation study highlights the importance of topics and especially early topic-embedding fusion in our architecture.