Movie Summarization via Sparse Graph Construction
Keywords:Summarization, Language Grounding & Multi-modal NLP, Graph-based Machine Learning
AbstractWe summarize full-length movies by creating shorter videos containing their most informative scenes. We explore the hypothesis that a summary can be created by assembling scenes which are turning points (TPs), i.e., key events in a movie that describe its storyline. We propose a model that identifies TP scenes by building a sparse movie graph that represents relations between scenes and is constructed using multimodal information. According to human judges, the summaries created by our approach are more informative and complete, and receive higher ratings, than the outputs of sequence-based models and general-purpose summarization algorithms. The induced graphs are interpretable, displaying different topology for different movie genres.
How to Cite
Papalampidi, P., Keller, F., & Lapata, M. (2021). Movie Summarization via Sparse Graph Construction. Proceedings of the AAAI Conference on Artificial Intelligence, 35(15), 13631-13639. https://doi.org/10.1609/aaai.v35i15.17607
AAAI Technical Track on Speech and Natural Language Processing II