Leveraging AI for Video Appraisal: A Case Study at the World Bank Group
Abstract
The World Bank Group (WBG) Archives, in partnership with the WBG IT Department, has developed a tool to support the appraisal, selection and disposition of video recordings. The Archives Video Appraiser (AVA) leverages Machine Learning (ML) to make recommendations on which recordings to keep or destroy after having learnt from a trusted set of training data. Decisions are validated or corrected by an archivist so that AVA can continue to learn from its mistakes. AVA´s current predictions are 85% accurate and the use of AVA has resulted in 1.5-person-day savings per month and in the reduction of manual mistakes.Details
- Creators
- Paloma Arias; Jeanne Kramer-Smyth
- Institutions
- World Bank
- Date
- 2022-09-16 00:00:00 +0000
- Keywords
- machine learning; appraisal; digital preservation
- Publication Type
- poster
- License
- CC-BY 4.0 International
- Download
- (unknown) bytes
- Slides
- here
- Video Stream
- here