Enables machines to better understand and extract free text, with applications for financial contract management and epidemic policy advisory
OneConnect Financial Technology Co., Ltd., a leading technology-as-a-service platform for financial institutions in China, is pleased to announce that its artificial intelligence research institute, Gamma Lab, topped the rankings in two sub-tasks under ‘Common Sense Knowledge and Reasoning, Knowledge Extraction” at the 14th International Workshop on Semantic Evaluation (SemEval 2020), a series of tests to measure computer understanding of human language.
SemEval is an international vocabulary and semantic competition organized by the Association for Computational Linguistics (ACL), an international scientific and professional society for people working on problems involving natural language and computation.
This year, Gamma Lab’s proprietary free text information extraction technology won the two sub-tasks of “Sentence Classification” and “Sequence Labeling” under “Extracting Definitions from Free Text in Textbooks” with accuracies of 87.83% and 84.71%.
The high accuracy of the technology has led to its commercial rollout in OneConnect’s Smart Contract Cloud Platform, with over 1,000 standard contract templates or banking, fund management, securities, trust and other financial institutions. The smart contract cloud platform enables financial institutions to efficiently review, manage and complete large numbers of contracts with their end-users.
To help smaller firms better react to the fast-changing COVID-19 situation, OneConnect has also deployed the free text information extraction technology into its Epidemic Advisory Program, which provides analysis and classification of the latest epidemic-related policies and announcements to small and medium enterprises (SMEs) for them to stay on top of government interventions and respond accordingly.This year, Gamma Lab’s proprietary free text information extraction technology won the two sub-tasks of “Sentence Classification” and “Sequence Labeling” under “Extracting Definitions from Free Text in Textbooks” with accuracies of 87.83% and 84.71%.