Utilizing coarse-grained data in low-data settings for event extraction
2022
0 views
0 downloads
Advisor: Prof. Dr. Deniz Yuret ; Dr. Öğr. Üyesi Ali Hürriyetoğlu
Abstract (EN)
Annotating text data for event information extraction systems is hard, expensive, and error-prone. We investigate the feasibility of integrating coarse-grained data (document or sentence labels), which is far more feasible to obtain, instead of annotating more documents. We utilize a multi-task model with two auxiliary tasks, document and sentence binary classification, in addition to the main task of token classification. We perform a series of experiments with varying data regimes for the aforementioned integration. Results show that while introducing extra coarse-grained data offers greater improvement and robustness, a gain is still possible with only the addition of negative documents that have no information on any event.
Author
Osman Mutlu
Institution
How to Cite
Osman Mutlu (Master Thesis). Utilizing coarse-grained data in low-data settings for event extraction, 2022, Koç University.
Keywords
License
Tüm Hakları Saklıdır
This work is shared under the specified license terms.
More theses from Koç University
- Obje tabanlı akıl danışma-tavsiye iletişimi tasarımına ilham kaynağı olarak Türk kahve falı(2017)
- State-building in multi-ethnic borderlands: Nationalizing Eastern Anatolia and Transylvania in interwar Turkey and Romania(2021)
- Cross-cultural and artistic dialogues in the seventeenth century constantinople/istanbul: The Iconography of Madonna della Misericordia and the Galata Icon(2024)
- Ekom-Eczacıbaşı'nın Rusya piyasasındaki pazarlama stratejileri(1995)
- Barok döneminde Balkanlar Osmanlı Avrupası'nda mimaride, dekorasyonda, himaye ve kültürel üretim modellerinde dönüşüm, 1718-1856(2006)
- De Rham-Witt kompleks(2011)
