Test paper generation system capable of avoiding test question similarity

Abstract

The invention relates to the technical field of computers, in particular to a test paper generation system capable of avoiding test question similarity. Comprising an intelligent question bank unit, a parameter design unit, an LSH algorithm unit, an intelligent question extraction unit and a typesetting generation unit. Wherein the intelligent question bank unit is used for differentially storing test question contents; the parameter design unit is used for designing environment parameters and test paper parameters; the LSH algorithm unit is used for quickly identifying similar test questions; the intelligent question extraction unit is used for automatically extracting test questions and answers; the typesetting generation unit is used for automatically typesetting and generating test paper. According to the design of the invention, teachers can complete test paper composition in a short time according to requirements, the test paper composition time is saved, and the method has the advantages of high flexibility, customizability, what you see is what you get, high matching with teachers and the like; the test paper can be generated only by office, the test paper can be conveniently formed at any place, and the use convenience is enhanced; according to the generated test paper, the repetition rate of similar test questions can be effectively reduced, and the test paper composition complexity is reduced.

Description

technical field

[0001] The invention relates to the field of computer technology, in particular to a system for generating test papers that avoids similar test questions. Background technique

[0002] Repetition of test questions is one of the important factors affecting the quality of test questions and test papers, so it is particularly important to detect and remove similar test questions. The current method for discovering similar test questions mainly uses the text similarity detection method in natural language processing (NLP). These methods can be roughly divided into the following categories: word co-occurrence method, similarity method based on lexical database, and recurrent neural network and deep neural network method based on word vectors. But in general, all the above methods are often not suitable for similarity detection of test questions. These methods are too expensive to pursue precise semantic similarity, and they ignore the time complexity of pairwise...

Images