Harnessing the Wikipedia Graph for Effective Multi-Entity Question Answering
Keywords: Multi-Entity QA, Wikipedia Graph, Structured QA, RAG
Abstract: Wikipedia serves as a rich repository of well-curated knowledge, making it a popular source for information retrieval through question answering (QA). Often, these inquiries involve multiple entities, such as ``How many Turing Award winners are Canadian?'', necessitating the consolidation of information from various Wikipedia pages. Multi-entity question answering typically comprises two steps: multi-entity retrieval and subsequent reasoning using large language models (LLMs). The pre-defined connections within Wikipedia, known as the wiki-graph, facilitate relatively straightforward multi-entity retrieval. However, traditional solutions leveraging retrieval-augmented generation (RAG) encounter limitations, as LLMs often struggle to aggregate insights from multiple pages effectively. In response, we propose a Structured QA (SQA) approach that first organizes extracted entities into a relational table (e.g., a table schema with columns (name, nationality) for Turing Award winners) and then employs table-based methods such as TableQA or NL2SQL for answering. Extensive experiments demonstrate the superior effectiveness of SQA in addressing multi-entity QA challenges, improves the overall accuracy 29.6% over the SOTA solutions, paving the way for more robust information retrieval from Wikipedia.
Primary Area: generative models
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 6515
Loading