Learning by Semantic Similarity Makes Abstractive Summarization Better

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.
Comment: The initial version of the manuscript includes a model design (semsim), experimental results, and discussions on the results. We found that our model has flaws in its implementation and design. This final version of the manuscript is from the rest of the initial paper; we included our findings on the benchmark dataset, BART generated results and human evaluations, and we excluded our model semsim