Iterative Semi-supervised Learning: Helping the User to Find the Right Records

This paper proposes extending semi-supervised learning by allowing an ongoing interaction between a user and the system. The extension is intended to not only to speed up search for relevant aircraft engine maintenance records but also to help in improving the user’s understanding of the problem domain. After the user has identified a small number of relevant records, the system produces a description which generalizes their common properties. If the user is satisfied with the description, the system retrieves more potentially relevant records. The user critiques the items returned, labeling them as relevant or not. The system updates the description using this new labeling information and retrieves more records. The process continues until the user is satisfied that most relevant records have been found. To validate the efficacy of the approach, a set of related maintenance records are collected using the system. These records are compared to those collected without system support.

• Type: Conference
• Year: 2004
• Where: IEAAIE
• Authors: Chris Drummond
• Keywords: IEAAIE 2004, Maintenance Records, Relevant Records, User