AUTHOR=Ehnert Philip , Schröter Julian 

TITLE=Key point generation as an instrument for generating core statements of a political debate on Twitter

JOURNAL=Frontiers in Artificial Intelligence

VOLUME=Volume 7 - 2024

YEAR=2024

URL=https://www.frontiersin.org/journals/artificial-intelligence/articles/10.3389/frai.2024.1200949

DOI=10.3389/frai.2024.1200949

ISSN=2624-8212

ABSTRACT=Identifying key statements in large volumes of short, user-generated texts is essential for decision makers to quickly grasp their key content. To address this need, this research introduces a novel abstractive Key Point Generation (KPG) approach applicable to unlabelled text corpora, using an unsupervised approach, a feature not yet seen in existing abstractive KPG methods. The proposed method uniquely combines topic modeling for unsupervised data space segmentation with abstractive summarisation techniques to efficiently generate semantically representative Key Points from text collections. This is further enhanced by hyperparameter tuning to optimise both the topic modeling and abstractive summarisation processes. The hyperparameter tuning of the topic modeling aims at making the cluster assignment more deterministic, as the probabilistic nature of the process would otherwise lead to high variability in the output. The abstractive summarisation process is optimised using a Davies-Bouldin Index specifically adapted to this use case, so that the generated Key Points more accurately reflect the characteristic properties of this cluster. In addition, our research recommends an automated evaluation that provides a quantitative complement to the traditional qualitative analysis of KPG. This method 31 regards KPG as a specialised form of Multi 32 Document Summarisation (MDS) and employs 33 both word-based and word embedding-based 34 metrics for evaluation. These criteria allow for 35 a comprehensive and nuanced analysis of the 36 KPG output. Demonstrated through application 37 to a political debate on Twitter, the versatility 38 of this approach extends to various domains, 39 such as product review analysis and survey 40 evaluation. This research not only paves the 41 way for innovative development in abstractive 42 KPG methods but also sets a benchmark for 43 their evaluation.