Incremental Update Summarization: Adaptive Sentence Selection based on Prevalence and Novelty
- Submitting institution
-
University of Glasgow
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 11-00751
- Type
- E - Conference contribution
- DOI
-
10.1145/2661829.2661951
- Title of conference / published proceedings
- CIKM '14: 23rd ACM International Conference on Conference on Information and Knowledge Management
- First page
- 301
- Volume
- -
- Issue
- -
- ISSN
- -
- Open access status
- -
- Month of publication
- November
- Year of publication
- 2014
- URL
-
http://eprints.gla.ac.uk/101448/
- Supplementary information
-
-
- Request cross-referral to
- -
- Output has been delayed by COVID-19
- No
- COVID-19 affected output statement
- -
- Forensic science
- No
- Criminology
- No
- Interdisciplinary
- No
- Number of additional authors
-
2
- Research group(s)
-
-
- Citation count
- -
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- ORIGINALITY: Proposes a new event summarization model that avoids undesirable summary dilution over time by learning the necessary number of updates needed to convey the evolution of an event to users; RIGOUR: Evaluated over an extended standard TREC test collection across 10 events against 8 existing baselines for both retrospective and live summarization, while providing quantitative and qualitative feature analysis; SIGNIFICANCE: Reduces the amount of non-relevant content the user sees (statistically significant 46% improvement). Deployed for intelligence gathering from tweets within a Flash Flooding Live Deployment (https://hubs.mymeedia.com/super-fp7/post/38012439) during the SUPER project (http://super-fp7.eu/);
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -