An examination question query system and method

Abstract

The present invention discloses a test question query system and method, including a test question library module, a test question collection module, an ocr module, a test question element separation module, and a test question query module. The test question library module is used to store test question resources. The test question collection module is used to collect images of test questions to be queried. The ocr module is used to perform ocr recognition on the test question images to be retrieved to obtain the text content of the test questions to be queried. The test question element separation module is used to split the test question stem and attached drawings. The test question query module is used to query target test questions according to the characteristics of the separated elements. The present invention effectively combines the text recognition matching feature and the picture similarity matching feature, ensuring both the recognition accuracy rate and the recognition speed, and can effectively improve the user experience of searching for test questions. Through the test question element separation module, it is possible to separate test questions without attached drawings, test questions with text in the attached drawings, and test questions without text in the attached drawings, thereby narrowing the scope of test questions during query and improving the query efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent teaching, and particularly to a test question query system and method. Background Art

[0002] At present, online education products are constantly developing, including question-searching products with functions such as taking pictures to answer questions. The question-searching products are designed to enable student users to obtain an image containing a question when they encounter a difficult problem in their homework, perform image recognition on the image, and search for the questions and answer analyses required by the user in the background question bank based on the results of the image recognition.

[0003] Current test question searches usually involve taking pictures of test question images, performing ocr on the test question stem, and then performing text similarity detection to confirm the target retrieval question. This method has a high accuracy for test questions with a large amount of text, but has a low accuracy for test questions with little text and many attached drawings. Therefore, an optimized test question query system and method are needed. Summary of the Invention

[0004] The purpose of the present invention is to provide a test question query system and method to solve the problems raised in the above background art.

[0005] To achieve the above purpose, the present invention provides the following technical solution: A test question query system includes a test question bank module, a test question collection module, an ocr module, a test question element separation module, and a test question query module;

[0006] The test question bank module is used to store test question resources;

[0007] The test question collection module is used to collect images of test questions to be queried;

[0008] The ocr module is used to perform ocr recognition on the test question image to be retrieved and obtain the text content of the test question to be queried;

[0009] The test question element separation module is used to split the test question stem and attached drawings;

[0010] The test question query module is used to query the target test question according to the separated element features.

[0011] Preferably, the test question bank module includes at least a test question stem text field and a test question attached drawing data field.

[0012] Preferably, the test question bank module further includes one or more fields such as test question answers, analyses, associated videos, knowledge points, examination points, grades, subjects, question types, sources, and inclusion times.

[0013] Preferably, the test question collection module uses a scanner, a high-speed camera, and an intelligent camera to collect test question images.

[0014] A method for querying test questions, comprising the following steps:

[0015] S1. Collect test question images, obtain the test questions to be recognized, perform element separation on the test questions to be recognized, separate the test question stem and the test question attached drawings. If there are no attached drawings for the test question, only return the text result. If the test question includes multiple attached drawings, perform an attached drawing splicing operation to synthesize one attached drawing;

[0016] S2. When the result is a test question without attached drawings, directly use the ocr module to recognize the text similarity for matching, and obtain the query result with the highest matching degree as the target test question result;

[0017] S3. When the result is a test question with text in the attached drawing, obtain the query result with the highest matching degree as the target test question result according to the text content of the attached drawing with text and the text content of the test question stem;

[0018] S4. The text content of the attached drawing with text is recognized by the ocr module and the attached drawing separation operation of the test question element separation module to obtain the coordinates of the ocr text and the coordinates of the attached drawing respectively. The text content of the attached drawing can be obtained through the region where the coordinates coincide.

[0019] Preferably, the test question element separation further includes test questions without attached drawings, test questions with text in the attached drawing, the attached drawing contains text, and test questions with no text in the attached drawing.

[0020] Preferably, the obtaining of the query result with the highest matching degree further means that in the test question bank module, a first result set with attached drawings with text is screened out, and the text similarity of the test question stem and the attached drawing text is calculated respectively in the first result set. After taking the weighted average of the calculation results and sorting, the result with the highest similarity is obtained as the target test question result.

[0021] Preferably, the obtaining of the query result with the highest matching degree further means that in the test question bank module, a second result set with attached drawings without text is screened out, and the number of words in the test question stem is further judged in the first result set. If the number of words is greater than the preset threshold, the target test question result is retrieved from the second result set based on the text similarity. If the number of words is less than or equal to the preset threshold, the target test question result is retrieved based on the attached drawing similarity.

[0022] Compared with the prior art, the beneficial effects of the present invention are:

[0023] 1. The present invention effectively combines the text recognition and matching characteristics and the picture similarity matching characteristics, while ensuring the recognition accuracy rate, also ensuring the recognition speed, and can effectively improve the user's experience of searching for test questions.

[0024] 2. Through the question element separation module, the present invention can separate questions without attached drawings, questions with attached drawings having text, and questions with attached drawings without text, thereby narrowing the scope of questions during query and improving the query efficiency. Detailed implementation mode

[0025] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0026] Embodiment 1:

[0027] The present invention provides a technical solution: a question query system, including a question bank module, a question collection module, an ocr module, a question element separation module, and a question query module;

[0028] The question bank module is used to store question resources. The question bank module at least includes a question stem text field and a question attached drawing data field. The question bank module also includes one or more fields among question answers, analysis, associated videos, knowledge points, examination points, grades, subjects, question types, sources, collection times, etc.;

[0029] The question collection module is used to collect the images of the questions to be queried. The question collection module uses a scanner, a high-speed camera, and an intelligent camera to collect the question images;

[0030] The ocr module is used to perform ocr recognition on the question image to be retrieved and obtain the text content of the question to be queried;

[0031] The question element separation module is used to split the question stem and attached drawings;

[0032] The question query module is used to query the target question according to the separated element features.

[0033] A question query method includes the following steps:

[0034] S1. Collect the question image, obtain the question to be recognized, perform element separation on the question to be recognized, separate the question stem and the question attached drawing. If the question has no attached drawing, only return the text result. If the question includes multiple attached drawings, perform the attached drawing splicing operation to synthesize one attached drawing;

[0035] S2. When the result is a question without an attached drawing, directly use the ocr module to identify the text similarity for matching, and obtain the query result with the highest matching degree as the target question result.

[0036] Embodiment 2:

[0037] The present invention provides a technical solution: a test question query system, including a test question library module, a test question collection module, an ocr module, a test question element separation module, and a test question query module;

[0038] The test question library module is used to store test question resources. The test question library module includes at least a test question stem text field and a test question drawing data field. The test question library module also includes one or more fields such as test question answers, analysis, associated videos, knowledge points, examination points, grades, subjects, question types, sources, and collection times;

[0039] The test question collection module is used to collect images of test questions to be queried. The test question collection module uses a scanner, a high-speed camera, and an intelligent camera to collect test question images;

[0040] The ocr module is used to perform ocr recognition on the test question image to be retrieved to obtain the text content of the test question to be queried;

[0041] The test question element separation module is used to split the test question stem and the drawing;

[0042] The test question query module is used to query the target test question according to the separated element features.

[0043] A test question query method includes the following steps:

[0044] S1. Collect a test question image to obtain the test question to be recognized, perform element separation on the test question to be recognized, separate the test question stem and the test question drawing. If the test question has no drawing, only return the text result. If the test question includes multiple drawings, perform a drawing splicing operation to synthesize one drawing;

[0045] S2. When the result is a test question with text and drawing, obtain the query result with the highest matching degree as the target test question result according to the text content of the test question with text and drawing and the text content of the test question stem.

[0046] The obtaining of the query result with the highest matching degree further means screening out a first result set with test questions with text and drawing in the test question library module, calculating the similarity of the test question stem text and the drawing text respectively in the first result set, sorting after taking the weighted average of the calculation results, and obtaining the result with the highest similarity as the target test question result

[0047] Embodiment 3:

[0048] The present invention provides a technical solution: a test question query system, including a test question library module, a test question collection module, an ocr module, a test question element separation module, and a test question query module;

[0049] The question bank module is used to store question resources. The question bank module includes at least a question stem text field and a question attached drawing data field, and the question bank module further includes one or more fields among question answers, analysis, associated videos, knowledge points, examination points, grades, subjects, question types, sources, collection times, etc.;

[0050] The question collection module is used to collect images of questions to be queried. The question collection module uses a scanner, a high-speed camera, and an intelligent camera to collect question images;

[0051] The ocr module is used to perform ocr recognition on the question image to be retrieved and obtain the text content of the question to be queried;

[0052] The question element separation module is used to split the question stem and attached drawings;

[0053] The question query module is used to query target questions according to the separated element features.

[0054] A question query method includes the following steps:

[0055] S1. Collect a question image, obtain the question to be recognized, perform element separation on the question to be recognized, separate the question stem and the question attached drawing. If the question has no attached drawing, only return the text result. If the question includes multiple attached drawings, perform an attached drawing splicing operation to synthesize one attached drawing;

[0056] S2. For the text content of the attached drawing with text, through the ocr module recognition and the attached drawing separation operation of the question element separation module, obtain the coordinates of the ocr text and the coordinates of the attached drawing respectively. Through the overlapping area of the coordinates, the text content of the attached drawing can be obtained. The question element separation further includes questions without attached drawings, questions with attached drawings with text, and questions with attached drawings without text.

[0057] The obtaining of the query result with the highest matching degree further means that in the question bank module, a second result set with attached drawings without text is screened out, and in the first result set, further judge the number of words in the question stem text. If the number of words is greater than the preset threshold, retrieve the target question result from the second result set according to the text similarity. If the number of words is less than or equal to the preset threshold, retrieve the target question result according to the attached drawing similarity.

[0058] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirits of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for querying test questions, characterized in that, it includes the following steps: S1. Collect test question images, obtain the test questions to be recognized, separate the elements of the test questions to be recognized, separate the test question stem and the test question drawings. If there are no drawings in the test question, only return the text result. If the test question includes multiple drawings, perform a drawing splicing operation to synthesize one drawing; S2. When the result is a test question without drawings, directly use the ocr module to recognize the text similarity for matching, and obtain the query result with the highest matching degree as the target test question result; S3. When the result is a test question with text in the drawing, according to the text content of the drawing with text and the text content of the test question stem, obtain the query result with the highest matching degree as the target test question result; S4. The text content of the drawing with text is recognized by the ocr module and the drawing separation operation of the test question element separation module to obtain the coordinates of the ocr text and the drawing coordinates respectively. The text content of the drawing can be obtained through the overlapping area of the coordinates; The obtaining of the query result with the highest matching degree further means that in the test question bank module, a first result set with text in the drawing is screened out, and the text similarity of the test question stem and the drawing text is calculated respectively in the first result set. After taking the weighted average of the calculation results and sorting, the result with the highest similarity is obtained as the target test question result; The obtaining of the query result with the highest matching degree further means that in the test question bank module, a second result set with no text in the drawing is screened out, and the number of words in the test question stem is further judged in the first result set. If the number of words is greater than the preset threshold, the target test question result is retrieved from the second result set according to the text similarity. If the number of words is less than or equal to the preset threshold, the target test question result is retrieved according to the drawing similarity.

2. A method for querying test questions according to claim 1, characterized in that: The separation of test question elements further includes test questions without drawings, test questions with text in the drawing, the drawing contains text, and test questions without text in the drawing, the drawing has no text.

3. A test question query system using the test question query method according to any one of claims 1-2, characterized in that, it includes a test question bank module, a test question collection module, an ocr module, a test question element separation module, and a test question query module; The test question bank module is used to store test question resources; The test question collection module is used to collect images of the test questions to be queried; The ocr module is used to perform ocr recognition on the test question image to be retrieved and obtain the text content of the test question to be queried; The test question element separation module is used to split the test question stem and the drawing; The test question query module is used to query the target test question according to the separated element features.

4. A test question query system according to claim 3, characterized in that: The test question bank module at least includes a test question stem text field and a test question drawing data field.

5. A test question query system according to claim 3, characterized in that: The test question bank module further includes one or more fields of test question answers, analysis, associated videos, knowledge points, examination points, grades, subjects, question types, sources, collection times, etc.

6. A test question query system according to claim 3, characterized in that: The test question collection module uses a scanner, a high-speed camera, and an intelligent camera to collect test question images.