DnDScore: Decontextualization and Decomposition for Factuality Verification in Long-Form Text Generation
This work addresses factuality verification for long-form text generated by LLMs, which is an incremental improvement over existing decompose-then-verify methods.
The paper tackled the problem of verifying factuality in long-form text generation by investigating the interaction between decomposition and decontextualization strategies, finding that strategy choice significantly impacts factuality scores, and introduced DnDScore, a method that improved verification by validating subclaims with contextual information.
The decompose-then-verify strategy for verification of Large Language Model (LLM) generations decomposes claims that are then independently verified. Decontextualization augments text (claims) to ensure it can be verified outside of the original context, enabling reliable verification. While decomposition and decontextualization have been explored independently, their interactions in a complete system have not been investigated. Their conflicting purposes can create tensions: decomposition isolates atomic facts while decontextualization inserts relevant information. Furthermore, a decontextualized subclaim presents a challenge to the verification step: what part of the augmented text should be verified as it now contains multiple atomic facts? We conduct an evaluation of different decomposition, decontextualization, and verification strategies and find that the choice of strategy matters in the resulting factuality scores. Additionally, we introduce DnDScore, a decontextualization aware verification method which validates subclaims in the context of contextual information.