Open Domain Question Answering with Conflicting Contexts
This addresses accuracy issues in open domain QA systems for users relying on web-based information, representing an incremental improvement through model finetuning.
The paper tackles the problem of open domain question answering systems producing untruthful answers due to conflicting information in retrieved contexts, finding that 25% of unambiguous questions lead to such conflicts, and demonstrates that finetuning large language models to explain answers improves their reasoning with conflicting contexts.
Open domain question answering systems frequently rely on information retrieved from large collections of text (such as the Web) to answer questions. However, such collections of text often contain conflicting information, and indiscriminately depending on this information may result in untruthful and inaccurate answers. To understand the gravity of this problem, we collect a human-annotated dataset, Question Answering with Conflicting Contexts (QACC), and find that as much as 25% of unambiguous, open domain questions can lead to conflicting contexts when retrieved using Google Search. We evaluate and benchmark three powerful Large Language Models (LLMs) with our dataset QACC and demonstrate their limitations in effectively addressing questions with conflicting information. To explore how humans reason through conflicting contexts, we request our annotators to provide explanations for their selections of correct answers. We demonstrate that by finetuning LLMs to explain their answers, we can introduce richer information into their training that guide them through the process of reasoning with conflicting contexts.