Learning by Semantic Similarity Makes Abstractive Summarization Better
This work addresses the gap between automatic metrics like ROUGE and human judgment in abstractive summarization, providing insights for researchers and practitioners, though it is incremental as it builds on existing models and datasets.
The paper compares model-generated summaries from BART with reference summaries from CNN/DM using human evaluation, finding that model-generated summaries receive higher scores, highlighting issues with dataset characteristics and model generalization.
By harnessing pre-trained language models, summarization models had rapid progress recently. However, the models are mainly assessed by automatic evaluation metrics such as ROUGE. Although ROUGE is known for having a positive correlation with human evaluation scores, it has been criticized for its vulnerability and the gap between actual qualities. In this paper, we compare the generated summaries from recent LM, BART, and the reference summaries from a benchmark dataset, CNN/DM, using a crowd-sourced human evaluation metric. Interestingly, model-generated summaries receive higher scores relative to reference summaries. Stemming from our experimental results, we first argue the intrinsic characteristics of the CNN/DM dataset, the progress of pre-trained language models, and their ability to generalize on the training data. Finally, we share our insights into the model-generated summaries and presents our thought on learning methods for abstractive summarization.