A method for automatically answering a test question difficulty evaluation of a benchmark database

Abstract

The application provides a test question difficulty evaluation method for automatically answering a benchmark database, which comprises the following steps: constructing a balanced database based on the benchmark database; obtaining a first feature set based on the balanced database, performing feature screening on the first feature set, and obtaining a second feature set; inputting a label of a current test question and a second feature value obtained based on the second feature set into a preset model to obtain a feature contribution degree of the current test question; obtaining a feature vector of the current test question and a weight corresponding to the feature vector based on the feature contribution degree; performing normalization processing on the second feature value, and calculating the second feature value and the corresponding weight to obtain a difficulty value of the current test question, so as to realize difficulty evaluation of the current test question. The method can directly evaluate and sort the test questions in the benchmark database for automatically answering, and improve the generalization ability of a mechanical learning model trained in the benchmark database for automatically answering.

Images