Solution Report for the KDD 2024 OAG-Challenge: Academic Question Answering Task
Keywords: OAG-Challenge, Academic Question Answering, Large Language Models, Semantic Search
Abstract: This report details our solution for the OAG-Challenge, a competition aimed at advancing the state of academic knowledge graph mining technologies. We focused on the Academic Question Answering (AQA) task, which requires retrieving relevant papers that answer specialized questions. Our solution involves using pretrained large language models (LLMs) for generating text embeddings and employing similarity-based retrieval to identify the top 20 matching papers for each question. It is worth noting that our solution is built upon open-sourced LLMs for text embedding, making it training-free and resource-friendly for participants. Despite this, we achieved a top-9 rank in both the public and private leaderboards.
Submission Number: 11
Loading