Reproduce “A Survey on Event-based News Narrative Extraction” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Brian Keith Norambuena, Tanushree Mitra, Chris North
A systematic literature review comprehensively maps the landscape of event-based news narrative extraction, categorizing 54 relevant articles by representation models, extraction criteria, and evaluation approaches, thereby establishing a structured understanding of the field and charting future research directions.
Key insights
- Narrative representations vary from simple linear timelines to complex graph-based structures, with graph-based offering greater expressiveness but less exploration for tree structures.
- Extraction criteria leverage diverse features including relevance, content similarity, coherence, temporal relationships, and structural information, with a recent shift towards neural embeddings.
- Evaluation relies heavily on computational metrics (e.g., ROUGE, precision/recall) for supervised tasks, but highlights a critical need for more robust metrics, especially for complex, non-linear narrative structures and unsupervised methods.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.