Open Access

Research on the Financial Event Extraction Method Based on Fin-BERT

 and   
Dec 31, 2024

Cite
Download Cover

Figure 1.

Model framework
Model framework

Figure 2.

The trained model of Fin-Bert
The trained model of Fin-Bert

Figure 3.

RAAT structure
RAAT structure

Figure 4.

Types of Events Extracted Resulting F1 Values
Types of Events Extracted Resulting F1 Values

experimental extraxtion of chfinann datasets

P R F1
Doc2EDAG 80.3 70.5 77.5
GIT 82.3 78.4 80.3
PTPCG 88.2 69.1 79.4
Fin-RAAT 84.0 79.9 81.9

comparison of event extraction experiment results in chfinann datasets

F1
Single Multi All
Doc2EDAG 81.0 67.4 77.5
GIT 87.6 72.3 80.3
PTPCG 88.2 691 79.4
Fin-RAAT 87.9 75.3 81.9

datasets information

Training data Validation data Test data Event types
ChFinAnn 25632 3204 3204 5
Duee-fin 7015 1171 About 3500 13

duee-fin datasets

Dev Online test
P R F1 P R F1
Doc2EDAG 73.7 59.8 66.0 67.1 51.3 58.1
GIT 75.4 61.4 67.7 70.3 46.0 55.6
PTPCG 71.0 61.7 66.0 66.7 54.6 60.0
Fin-RAAT 76.1 71.3 73.0 70.3 56.1 62.8

experimental enviroment

Hardware Information Configure
Operating system Windows 10
Internal storage 32GB
CPU AMD R9-5900HX
Video card RTX-3080-LAPTOP
Memory 16GB
Exploitation environment Python 3.7
Language:
English
Publication timeframe:
4 times per year
Journal Subjects:
Computer Sciences, Computer Sciences, other