NCN grant
CORE – Computer-based methods for coreference resolution in Polish texts
The project created methods and tools for automated anaphora and coreference resolution in Polish texts, including a manually annotated coreference corpus.
Lead Investigator: Maciej Ogrodniczuk
- Grant Number
- 6505/B/T02/2011/40
- Start Date
- 18 April 2011
- End Date
- 17 July 2014
Project factsheet
| Polish name: | Komputerowe metody identyfikacji nawiązań w tekstach polskich |
| Project type: | A National Science Centre grant 6505/B/T02/2011/40 |
| Project Web page: | http://core.ipipan.waw.pl/ |
Project summary
The aim of the project is to create innovative methods and tools for automated anaphora and coreference resolution in Polish texts. To achieve this goal, the subset of the National Corpus of Polish will be manually annotated with coreferential chains. As the next step, several coreference resolution approaches will be tested on the annotated data (rule-based, statistical, hybrid etc.) Additional resources (such as Polish WordNet and Polish Wikipedia) will be also investigated to improve results.
Selected results
Polish Coreference Tools
Tools created as part of the project:
- Mention detection: MentionDetector
- Coreference resolution: Ruler (rule-based), Bartek (statistical)
- Evaluation: Scoreference
- Coreference annotation: DistSys, MMAX4CORE
- Converters: PCC converters
- Visualisation: brat4core
- Other: Text Selector