Broadcast news story segmentation using sticky hierarchical dirichlet process

Yu, Jia; Shao, Hongxiang

doi:10.1007/s10489-021-03098-4

Cited by 4 publications

(1 citation statement)

References 54 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the experimental section, we compare our proposed HDA approach with two baseline methods: Broadcast news story segmentation using sticky hierarchical Dirichlet process (SHDP-HMM) [38] and Semantic Segmentation of Text using Deep Learning (SSOT-DL) [39]. SHDP-HMM is chosen as a baseline due to its effectiveness in segmenting broadcast news stories.…”

Section: B Baseline Modelmentioning

confidence: 99%

Text Segmentation via Hierarchical Document Attention Model

Wang,

Min

2023

IEEE Access

View full text Add to dashboard Cite

With the rapid development of natural language processing technology, text segmentation has become an important task in text processing. However, existing text segmentation methods often perform poorly when faced with long texts and complex structures, requiring a more efficient and accurate approach. In this paper, we propose a new text segmentation method based on the Hierarchical Document Attention (HDA), which automatically identifies and segments different paragraphs in the text by analyzing and weighting the hierarchical structure of the text sequence data. Compared with existing methods, the model has higher accuracy and efficiency, and better supports tasks such as text analysis and information extraction. The main contribution of this paper is the proposal of a text segmentation method based on the HDA, which effectively models text sequences through multi-level attention mechanisms. Experimental verification on public datasets shows that this model exhibits good performance in text segmentation tasks.

show abstract

Section: B Baseline Modelmentioning

confidence: 99%