amazon-science / handling-ontology-gaps-in-semantic-parsing

Repository of the paper Handling Hallucinations In Neural Semantic Parsing.

Handling Hallucinations In Neural Semantic Parsing

Repository of the paper Handling Hallucinations In Neural Semantic Parsing.

Structure of the Repository

The repository contains three main folders

• Hallucination Simulation Framework (HSF)
  • Located in HSF-KQA-PRO/HallucinationSimulationFramework
  • It is used to generate the dataset which will be used by the NSP model to train and extract hallucination detection features
• NSP model
  • Located in HSF-KQA-PRO/Bart_Program
  • With that code you can train and extract
• Hallucination Detection Model
  • Located in HallucinationDetectionModel
  • This folder contains the code to train the HDM model
  • And it contains the end-to-end scorer

Code of Conduct

This project has adopted the Amazon Open Source Code of Conduct. For more information see the Code of Conduct FAQ or contact [e-mail] with any additional questions or comments.

Security issue notifications

If you discover a potential security issue in this project we ask that you notify AWS/Amazon Security via our vulnerability reporting page. Please do not create a public github issue.

Licensing

See LICENSE and THIRD-PARTY-LICENSES in the root of the project.

• CC BY NC: all folders except HSF-KQA-PRO/HallucinationSimulationFramework/out_folder
• CC BY SA: HSF-KQA-PRO/HallucinationSimulationFramework/out_folder

Authors

• Andrea Bacciu
• Marco Damonte
• Marco Basaldella
• Emilio Monti