Mitigating Hallucination in Abstractive Summarization with Domain-Conditional Mutual Information
This addresses the problem of generating inaccurate summaries for users relying on abstractive summarization systems, though it is an incremental improvement.
The paper tackled hallucination in abstractive summarization by introducing a decoding strategy based on domain-conditional mutual information, which improved faithfulness and source relevance on the XSUM dataset.
A primary challenge in abstractive summarization is hallucination -- the phenomenon where a model generates plausible text that is absent in the source text. We hypothesize that the domain (or topic) of the source text triggers the model to generate text that is highly probable in the domain, neglecting the details of the source text. To alleviate this model bias, we introduce a decoding strategy based on domain-conditional pointwise mutual information. This strategy adjusts the generation probability of each token by comparing it with the token's marginal probability within the domain of the source text. According to evaluation on the XSUM dataset, our method demonstrates improvement in terms of faithfulness and source relevance. The code is publicly available at \url{https://github.com/qqplot/dcpmi}.