Research & Papers

Beyond Static: Related Questions Retrieval Through Conversations in Community Question Answering

New TeCQR model uses conversations to find related questions on Stack Overflow.

Deep Dive

Researchers have developed TeCQR, a novel model that revolutionizes how related questions are retrieved in community question answering (cQA) platforms like Stack Overflow. Unlike traditional static approaches that ignore user interaction, TeCQR leverages conversations—specifically, tag-enhanced clarifying questions (CQs)—to capture fine-grained question representations. This allows the model to better understand user intent and retrieve more relevant questions.

The model incorporates a noise tolerance mechanism to evaluate semantic similarity between questions and tags, effectively handling noisy feedback. It also uses a two-stage offline training process to learn mutual relationships among user queries, questions, and tags. Experimental results show TeCQR significantly outperforms state-of-the-art baselines, offering a more dynamic and accurate retrieval system for cQA platforms.

Key Points
  • TeCQR uses tag-enhanced clarifying questions to build conversational context for question retrieval.
  • A noise tolerance model evaluates semantic similarity between questions and tags to handle noisy feedback.
  • Experimental results show TeCQR outperforms state-of-the-art baselines in retrieval accuracy.

Why It Matters

Improves how professionals find relevant answers on QA platforms, saving time and boosting productivity.