A method and system for intelligent identification and correction of questions

Abstract

The invention discloses a method and system for intelligent identification and correction of questions, which realizes automatic identification of question stems and answer contents of various types of questions in test papers and automatic correction of answer contents through artificial intelligence. The technical solution is: pre-training a neural network model (such as a CNN model) into a model for topic type identification based on a large number of test paper samples, and using the trained model to detect the position of each topic in the test paper and the corresponding type of the topic , different components in the question (question stem, answer question, picture), and then recognize the printed character information of the question stem and the handwritten character information of the answer part based on the recognition model, and finally perform automatic correction based on the recognized character information.

Description

technical field [0001] The invention relates to artificial intelligence technology, in particular to a technology for automatically correcting answer sheets containing various types of questions. Background technique [0002] With the continuous advancement of computer technology and educational informatization, computer technology has been gradually applied to daily educational and teaching activities, such as in the teaching evaluation scene. Existing basic education in China and the main examination forms of students' learning status are still various types of examinations or tests. In this situation, teachers are under great pressure to correct homework and test papers. In response to this phenomenon, various automatic marking systems have been gradually promoted and adopted in large and medium-sized or high-importance examinations or tests. Such systems can reduce the workload of teachers marking papers to a certain extent. [0003] In the existing automatic marking sy...

AI Technical Summary

Problems solved by technology

[0003] In the existing automatic marking system, most of the marking is completed by computer, which is mainly for marking objective questions (such as multiple choice questions). The processing method is to read the data on the chart card and compare it with the standard answer Yes, although this method effectively improves the review speed, it requires students to use 2B pencils to answer objective questions when answering, which is not only inconvenient, but also prone to mistakes, omissions, etc.

In addition, the printed test papers written by students are heavily dependent on manual inspection and correction by teachers or parents, especially for questions such as filling in the blanks or subjective questions. The position and specific content of the answers cannot be effectively automated through the existing marking system Recognition, or the recognition effect is poor and the efficiency is low

Moreover, subjective factors such as different teachers' styles, emotions, and psychological states may cause unfairness and injustice in the results of manual marking.

Images