What is the Topic?

The topic is to analyze the coalition agreement [1] using Natural Language Processing (NLP) methods. This can involve applying or developing methods to identify the most surprising statements in the text (e.g., text classification for surprise detection, see [2]). Additionally, the goal of the thesis could be to compare which topics and statements, previously mentioned in news articles before the federal election, made it into the coalition agreement and which ones did not. Those that made it into the agreement should be examined in greater detail. For instance, the sources of the news articles that covered these statements before the election can be revealed. The thesis may also examine changes between the parties’ pre-election statements and what was ultimately included in the coalition agreement.

It is expected that the student will submit the work as a joint scientific publication with the supervisor at a later stage. The required data (including news articles from before the election) are already available. The core tasks include processing the data (e.g., with Python) and applying and evaluating methods for automatic sentence comparison (e.g., SentenceBERT, ABCNN [3], BM25).

What Should You Bring?

• Interest in Text Mining
• An independent working style
• Basic programming knowledge (e.g., in Python or R)