Computer-Implemented Systems and Methods for Predicting Performance of Automated Scoring
- Author(s):
- Sukkarieh, Jana Z.
- Patent Issued:
- Jun 25, 2019
- Patent Number:
- 10,332,411
- Source:
- ETS Patent
- Document Type:
- Patent
- Family ID:
- 1000004106506
- Subject/Key Words:
- Patent, Active Patent, Automated Scoring and Natural Language Processing, Test Items, Computer-Implemented, Prediction, Performance, Scoring Models
Abstract
Systems and methods are provided for selecting a proposed test item for inclusion in an examination where a non-multiple choice response to the proposed test item will be automatically scored. A proposed test item is analyzed to generate a proposed test item metric, where the proposed test item is a non-multiple choice test item. The proposed test item metric is provided to a proposed test item scoring model, where the proposed test item scoring model outputs a likelihood score indicative of a likelihood that automated scoring of a response to the proposed test item would be at or above a quality level. The proposed test item is selected for inclusion in the examination based on the likelihood score.