Park. Optimizing Domain Specificity of Transformer-based Language Models for Extractive Summarization of Financial News Articles in Korean

Frequent usage of complex expressions with numbers and of the terms that require domain knowledge makes it more difficult to comprehend and summarize financial news articles than that of other daily news articles. We present a transformer-based model for the automatic summarization of the financial news articles in Korean and address related issues, and in particular analyze the interplay between the domain of the dataset used for pre-training and that for fine-tuning. We find that the summarization model performs much better when the two coincide, even when they are different from that of the target task, which is the financial domain in our work.