How Can a Teacher Make Learning From Sparse Data Softer? Application to Business Relation Extraction - Archive ouverte HAL Access content directly
Conference Papers Year :

How Can a Teacher Make Learning From Sparse Data Softer? Application to Business Relation Extraction

(1, 2) , (1, 3) , (3) , (1, 4)
1
2
3
4

Abstract

Business Relation Extraction between market entities is a challenging information extraction task that suffers from data imbalance due to the over-representation of negative relations (also known as No-relation or Others) compared to positive relations that corresponds to the taxonomy of relations of interest. This paper proposes a novel solution to tackle this problem, relying on binary soft-labels supervision generated by an approach based on knowledge distillation. When evaluated on a business relation extraction dataset, the results suggest that the proposed approach improves the overall performances, beating state-of-the art solutions for data imbalance. In particular, it improves the extraction of under-represented relations as well as the detection of false negatives.
Fichier principal
Vignette du fichier
4.pdf (394.88 Ko) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-03730414 , version 1 (24-08-2022)

Identifiers

  • HAL Id : hal-03730414 , version 1

Cite

Farah Benamara, Hadjer Khaldi, Camille Pradel, Nathalie Aussenac-Gilles. How Can a Teacher Make Learning From Sparse Data Softer? Application to Business Relation Extraction. 4th Workshop on Financial Technology and Natural Language Processing (FinNLP @ IJCAI 2022), Jul 2022, Vienna, Austria. pp.22-28. ⟨hal-03730414⟩
54 View
4 Download

Share

Gmail Facebook Twitter LinkedIn More