Domain-specific Text Summarization

Master Thesis

Text summarization aims at distilling the essential information from a text to produce a shorter version, such as generating headlines for news and subject lines for emails. Recent summarization methods use deep learning models that are trained and evaluated on English and standard corpora. However, to what extent these models can generalize to other languages like German and technical domains with rare words have not been explored. Specifically, summarizing technical texts is challenging as we require expertise and deep understanding in the domain of interest.

This project aims to explore and develop transfer learning methods to generate a sentence summary from a given textual snippet (a news article or an email written in German on a technical domain).