CN111597305B

CN111597305B - Entity marking method, entity marking device, computer equipment and storage medium

Info

Publication number: CN111597305B
Application number: CN202010412002.7A
Authority: CN
Inventors: 徐雨薇; 金海峰
Original assignee: Fazheng International Education Investment Co ltd
Current assignee: Fazheng International Education Investment Co ltd
Priority date: 2020-05-15
Filing date: 2020-05-15
Publication date: 2020-12-22
Anticipated expiration: 2040-05-15
Also published as: CN111597305A

Abstract

The application relates to an entity marking method, an entity marking device, a computer device and a storage medium. Wherein, the method comprises the following steps: responding to a natural spelling test request triggered by a tested person terminal, and sending first test data containing natural spelling test questions to the tested person terminal; receiving and storing audio data and spelling information sent by a tested person terminal; sending the audio data, spelling information and second test data containing the natural spelling test questions and answers to a tester terminal, and receiving a test result of each test question in the natural spelling test questions uploaded by the tester terminal; selecting a natural spelling knowledge entity matched with the label from a preset natural spelling knowledge map according to the label associated with the natural spelling test question; and adding a color identifier to the matched natural spelling knowledge entity according to the test result of each test question, and displaying a natural spelling knowledge map containing colors. By adopting the method, the natural spelling and reading test efficiency can be improved.

Description

Entity marking method, entity marking device, computer equipment and storage medium

Technical Field

The present application relates to the technical field of knowledge graph and computer interactive processing, and in particular, to an entity tagging method, apparatus, computer device, and storage medium.

Background

Google corporation introduced artificial intelligence-based knowledge graph technology as early as 2012, and with the enhancement of computer processing capability and the growth of data resources, knowledge graphs are increasingly widely applied as a means for data analysis and decision support, such as search engines, commodity recommendation, automatic question answering and the like.

Knowledge graph technology has also attracted extensive attention in all circles in the field of education. China 'New generation artificial intelligence development planning' particularly emphasizes that a knowledge graph covering hundreds of millions of knowledge entities is constructed by researching key technologies such as knowledge graph construction and learning, knowledge evolution and reasoning and the like.

One attempted application in the field of education today is disciplinary knowledge mapping. In brief, the discipline knowledge graph is a knowledge base composed of knowledge point entities and relations among the knowledge point entities, wherein the knowledge point entities can be understood and stored by a computer. The relations among the knowledge point entities have definite sequence, inclusion and orientation relations, and a knowledge system established according to the semantic relations can provide support for teachers, students and the like in the aspects of teaching, learning, evaluation and the like.

In recent years, the teaching concept of 'centering on students' is increasingly emphasized in the field of education, and how to evaluate the real mastery condition of the knowledge of the students is still a difficult problem for schools, teachers, students and parents. If can be based on subject knowledge map, with the help of the education data acquisition of equipment, record student in the data of each link such as homework, exercise, examination, combine machine learning analysis technique simultaneously, can not only be with the visual mastery degree that demonstrates student's knowledge point of form of knowledge map, also can provide help and guide for teacher's further teaching.

In the current English subject, natural spelling is regarded as basic capability of English reading and is emphasized by more and more families and schools. As a learning method for effectively establishing English reading capability, natural spelling allows English learners to actively judge the pronunciation of words immediately after seeing the spelling of the words by learning the one-to-one correspondence between phonemes and letters or letter combinations, thereby greatly improving the accuracy and efficiency of reading.

How to accurately judge how does a student master the natural spelling ability during learning? The traditional method is to complete the test questions on the paper test paper within a specified time by the testee, and then realize the natural spelling capability test by the way of scoring for the testee. The disadvantage of this form is that the paper-pen test excessively depends on teaching material resources or subjective experience in the process of paper formation, scoring and analysis for teachers, and the single score dimension cannot show the mastering degree of specific knowledge points and the conditions of the prior knowledge points related to the specific knowledge points for students and teachers, and the test cannot provide data references for future accurate teaching, personalized learning and the like, so that the test accuracy is low, and the test efficiency based on the traditional mode is difficult to effectively improve.

Disclosure of Invention

In view of the above, there is a need to provide an entity marking method, an entity marking device, a computer device, and a storage medium, which can improve the testing accuracy to accurately judge the natural spelling ability of the student during the learning process.

A method of entity tagging, the method comprising:

responding to a natural spelling test request triggered by a testee through a testee terminal, and sending first test data containing natural spelling test questions to the testee terminal so that the tester terminal can display the natural spelling test questions; the natural spelling test questions comprise pronunciation test questions and spelling test questions;

receiving and storing audio data sent by the testee terminal, wherein the audio data is obtained by recording when the testee performs pronunciation test according to the pronunciation test question;

receiving and storing spelling information sent by the tested person terminal, wherein the spelling information is information input in the tested person terminal when the tested person carries out spelling test according to the spelling test question;

sending the audio data, the spelling information and second test data containing natural spelling test questions and answers to a tester terminal, and receiving a test result of each test question in the natural spelling test questions uploaded by the tester terminal; the test result comprises a correct result and an incorrect result;

selecting a natural spelling knowledge entity matched with the label from a preset natural spelling knowledge map according to the label associated with the natural spelling test question; the natural spelling knowledge entity comprises a knowledge block entity, a level entity, a test type entity and a test question type entity, wherein the knowledge block entity comprises at least one knowledge point entity;

and adding a color identifier to the matched natural spelling knowledge entity according to the test result of each test question, determining a natural spelling knowledge graph containing the color identifier as a user portrait of the tested person, and displaying the user portrait through the tested person terminal.

A method of entity tagging, the method comprising:

responding to a natural spelling test request triggered by a testee through a testee terminal, and sending first test data containing natural spelling test questions to the testee terminal so that the tester terminal can display the natural spelling test questions;

receiving and storing audio data and spelling information sent by the tested person terminal, wherein the audio data and the spelling information are obtained when the tested person carries out natural spelling test according to the natural spelling test question;

sending the audio data, the spelling information and second test data containing natural spelling test questions and answers to a tester terminal, and receiving a test result of each test question in the natural spelling test questions uploaded by the tester terminal;

selecting a natural spelling knowledge entity matched with the label from a preset natural spelling knowledge map according to the label associated with the natural spelling test question;

A natural spelling test apparatus, the apparatus comprising:

the test data sending module is used for responding to a natural spelling test request triggered by a testee through a testee terminal, and sending first test data containing natural spelling test questions to the testee terminal so that the tester terminal can display the natural spelling test questions; the natural spelling test questions comprise pronunciation test questions and spelling test questions;

the audio data receiving module is used for receiving and storing the audio data sent by the testee terminal, wherein the audio data is obtained by recording when the testee performs pronunciation test according to the pronunciation test questions;

the spelling information receiving module is used for receiving and storing spelling information sent by the tested person terminal, wherein the spelling information is information input in the tested person terminal when the tested person carries out spelling test according to the spelling test question;

the test result determining module is used for sending the audio data, the spelling information and second test data containing natural spelling test questions and answers to a tester terminal and receiving a test result of each test question in the natural spelling test questions uploaded by the tester terminal; the test result comprises a correct result and an incorrect result;

the matching module is used for selecting a natural spelling knowledge entity matched with the label from a preset natural spelling knowledge map according to the label associated with the natural spelling test question; the natural spelling knowledge entity comprises a knowledge block entity, a level entity, a test type entity and a test question type entity, wherein the knowledge block entity comprises at least one knowledge point entity;

and the identification adding module is used for adding a color identification to the matched natural spelling knowledge entity according to the test result of each test question, determining the natural spelling knowledge map containing the color identification as the user portrait of the tested person, and displaying the user portrait through the tested person terminal.

In one embodiment, the apparatus further comprises:

the map correction module is used for acquiring historical test data of a tested person in a preset time period; inputting the historical test data into the trained atlas correction neural network model to obtain a corrected natural spelling knowledge atlas; and replacing the natural spelling knowledge map by the corrected natural spelling knowledge map.

A computer device comprising a memory and a processor, the memory storing a computer program, the processor implementing the following steps when executing the computer program:

A computer-readable storage medium, on which a computer program is stored which, when executed by a processor, carries out the steps of:

According to the entity marking method, the entity marking device, the computer equipment and the storage medium, the whole testing process is completed through interaction of the tester terminal, the tested tester terminal and the server, the user portrait which reflects the testing result and is based on the natural spelling knowledge map is given, the purpose of on-line testing is achieved, the testing accuracy is improved, the spelling and spelling abilities of students are accurately judged, and meanwhile the testing efficiency can be effectively improved when the testing scale is large. And secondly, in the test process, modes such as recording, spelling and the like which are easy to be operated by the testee are adopted, the operability is strong, the test burden of the testee is not additionally increased, and the remote, delayed and unlimited test can be realized. Moreover, the method adopts the test types of multiple dimensions such as pronunciation test, spelling test and the like, so that the natural spelling ability level of the testee can be tested more comprehensively, and the test accuracy is improved. And the obtained user portrait can more visually reflect the natural spelling ability level of the testee by adding the color identification to the natural spelling knowledge map, so that the visualization effect is more visual and accurate, and the user can quickly distinguish whether the natural spelling knowledge points are mastered. In addition, the mode of labeling the test questions is adopted, so that the accuracy of test question statistics corresponding to each knowledge point can be ensured, and the test credibility is further improved.

Drawings

FIG. 1 is a diagram of an embodiment of an application environment for an entity tagging method;

FIG. 2 is a flow diagram illustrating a method for entity tagging in one embodiment;

FIG. 3 is a schematic flow chart illustrating a complementary scheme for adding color labels to matched natural spelling knowledge entities according to the test results of each test question in one embodiment;

FIG. 4 is a schematic diagram illustrating the process of automatically modifying the scoring threshold by big data according to an embodiment;

FIG. 5 is a flowchart illustrating a process of generating and displaying a test report according to a test result of each test question and a knowledge point tag associated with each test question in one embodiment;

FIG. 6 is a schematic flow diagram illustrating automated refinement of a knowledge graph structure using a deep reinforcement learning model using artificial intelligence techniques in one embodiment;

FIG. 7 is a flowchart illustrating a method for entity tagging in another embodiment;

FIG. 8 is a block diagram of an entity tagging apparatus in one embodiment;

FIG. 9 is a diagram illustrating an internal structure of a computer device according to an embodiment.

Detailed Description

In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.

The entity marking method provided by the application can be applied to the application environment shown in fig. 1. Wherein the tester terminal 102 and the tester terminal 104 communicate with the server 106 through a network. Specifically, when the testee initiates a natural spelling test request through the tester terminal 102, the server 106 responds to the natural spelling test request and sends first test data containing natural spelling test questions to the tester terminal 102, so that the testee performs a natural spelling test according to the natural spelling test questions displayed by the tester terminal 102. Then, the server 106 receives the audio data and spelling information given by the testee for the natural spelling test questions, and sends the audio data and spelling information and the second test data containing the natural spelling test questions and answers to the tester terminal 104, so that the tester can determine the test result of each test question in the natural spelling test questions through the tester terminal 104 and upload the test result to the server 106, the server 106 selects the natural spelling knowledge entity matched with the label in the preset natural spelling knowledge map according to the label associated with the natural spelling test questions, and adds color identification to the matched natural spelling knowledge entity according to the test result of each test question. Then, the server determines the natural spelling knowledge map containing the color identifier as the user portrait of the tested person and sends the user portrait to the tested person terminal 102, so that the tested person terminal 102 displays the user portrait, and thus, the tested person can intuitively know the current natural spelling mastering level according to the natural spelling knowledge map.

The testee terminal 102 and the tester terminal 104 may be, but not limited to, various personal computers, notebook computers, smart phones, tablet computers, and portable wearable devices, and the server 104 may be implemented by an independent server or a server cluster formed by a plurality of servers.

In an exemplary embodiment, as shown in fig. 2, an entity tagging method is provided, which is described by taking the application of the method to the server in fig. 1 as an example, and specifically can be implemented by the following steps:

step 202, responding to a natural spelling test request triggered by a testee through a testee terminal, sending first test data containing natural spelling test questions to the testee terminal so that the testee terminal can display the natural spelling test questions.

The natural spelling test questions comprise pronunciation test questions and spelling test questions. In other embodiments, the natural spelling questions may also include speech-aware questions and comprehensive application questions. The pronunciation test question, spelling test question, voice awareness test question and comprehensive application test question all contain natural spelling knowledge points, such as letter name a and consonant concatenate dr. The pronunciation test question is used for the testee to carry out pronunciation test according to the seen natural spelling question so as to reflect the pronunciation capability mastered by the testee; the spelling test question is used for the tested person to carry out spelling test according to the heard natural spelling question so as to reflect the spelling ability mastered by the tested person; the voice consciousness test question is used for a tested person to carry out a voice consciousness test according to the seen natural spelling questions (such as selection questions) so as to reflect the voice consciousness ability mastered by the tested person; the comprehensive application test question is used for providing a segment of character description, so that the testee answers the character description based on the comprehensive application test question to reflect the comprehensive application capability mastered by the testee.

The terminal of the testee is terminal equipment used by the testee for performing natural spelling test.

Specifically, when the testee starts the natural spelling test, a natural spelling test request is sent to the server through the tester terminal. And after receiving the natural spelling test request, the server responds to the natural spelling test request and sends first test data containing natural spelling test questions to the terminal of the tested person. After receiving the first test data, the terminal of the testee shows the natural spelling question contained in the first test data to the testee, so that the testee can perform natural spelling test.

And step S204, receiving and storing the audio data sent by the tested person terminal.

The audio data is obtained by recording when the testee performs pronunciation test according to the pronunciation test question. The test question type of the pronunciation test question may be a reading question.

Specifically, when the testee performs a pronunciation test according to the pronunciation test question, the testee terminal or other audio acquisition devices acquire audio data of the testee. The tested person terminal or other audio collecting equipment can send collected audio data of the tested person to the server at preset time intervals, and can also send complete audio data of the tested person to the server after the pronunciation test of the tested person is finished. After receiving the audio data, the server stores the audio data and the corresponding ID of the testee in an associated manner, so as to facilitate subsequent calling at any time.

And step S206, receiving and storing spelling information sent by the terminal of the testee.

Wherein, the spelling information is the information input in the terminal of the testee when the testee carries out spelling test according to the spelling test question. Alternatively, the spelling information may be obtained through an image recognition technique through a spelling image obtained by intercepting an image in a terminal of the testee when the spelling image is used for a spelling test of the testee according to a spelling test question. The spelling information may also be information written in the screen of the test subject terminal by the test subject or entered information. The question type of the spelling question may be a dictation question.

As an implementation manner, when a testee performs a spelling test according to a spelling test question, the testee may write a spelling answer in a tester terminal, and may also write a spelling answer on another product (e.g., an answer sheet), and the tester terminal or another image capturing device may capture a spelling image by means of screen capturing or shooting. The terminal or other image acquisition equipment of the testee can send the acquired spelling image of the testee to the server at intervals of preset time, and can also send the complete spelling image set of the testee to the server after the spelling test of the testee is finished. After receiving the spelling image, the server stores the spelling image and the corresponding ID of the tested person in an associated manner, so as to facilitate subsequent calling at any time.

Step S208, sending the audio data, spelling information and second test data containing the natural spelling test questions and answers to the tester terminal, and receiving the test result of each test question in the natural spelling test questions uploaded by the tester terminal.

Wherein the test result comprises a correct result and an incorrect result.

Specifically, during or after the test of the testee, the server sends the uploaded audio data and spelling information and the second test data containing the natural spelling test questions and answers to the tester terminal, so that the tester can determine the test result of each test question in the natural spelling test questions by comparing the test answer (audio data, spelling image and/or option information) uploaded by the testee with the correct answer of the natural spelling test questions, and upload the test result of each test question to the server through the tester terminal, so that the server performs subsequent entity marking.

And step S210, selecting a natural spelling knowledge entity matched with the label from a preset natural spelling knowledge map according to the label associated with the natural spelling test question.

Wherein, the natural spelling knowledge map is composed of natural spelling knowledge entities. The natural spelling knowledge entity comprises a knowledge block entity, a level entity, a test type entity and a test question type entity, wherein the knowledge block entity comprises at least one knowledge point entity. For example, the knowledge point entities include diacritics ch, sh, and consonant concatenations dr, fr, and the corresponding knowledge block entities include diacritics and consonant concatenations. The level entities include a first level, a second level, a third level, and so on. The test type entities include pronunciation tests and spelling tests. In other embodiments, the test-type entity further includes a speech awareness test and a comprehensive application test. The test question type entities comprise reading questions, dictation questions and selection questions.

And the natural spelling knowledge entity and the label associated with the natural spelling test question have a corresponding relation. For example, the natural spelling test question a is associated with a diacritical ch label, a diacritical label, a second level label, and so on, in which case, the label diacritical ch corresponds to the entity diacritical ch, the label diacritical corresponds to the entity diacritical, and the label second level corresponds to the entity second level.

Specifically, the server selects a natural spelling knowledge entity matched with the label from a preset natural spelling knowledge map according to the label associated with the natural spelling test question. For example, the server may match the consonant-containing ch entity, the consonant-containing entity, and the second-level entity in the natural spelling knowledge graph according to the tag associated with the natural spelling test question a.

Step S212, according to the test result of each test question, adding a color identifier to the matched natural spelling knowledge entity, determining the natural spelling knowledge map containing the color identifier as the user portrait of the tested person, and displaying the user portrait through the terminal of the tested person.

Specifically, the server classifies the labels associated with the natural spelling test questions, and counts the accuracy of the natural spelling test questions corresponding to each label according to the test result of each test question. And then, the server adds color identification to the matched natural spelling knowledge entity according to the corresponding accuracy of each label. Optionally, the color indicator includes a plurality of color indicators, each color indicator representing a different natural spelling degree of the testee. In order to enable a testee or other users to know the current natural spelling mastering degree of the testee, the server sends the natural spelling knowledge map containing the colored natural spelling knowledge entity to the testee terminal, so that the testee terminal displays the natural spelling knowledge map containing the colored natural spelling knowledge entity to the testee or other users, and the natural spelling knowledge map containing the colored natural spelling knowledge entity can also be called as a user portrait and is used for reflecting the test result and the capability evaluation level of the testee.

In the entity marking method, the whole testing process is completed through interaction of the tester terminal, the tested tester terminal and the server, the user portrait which reflects the testing result and is based on the natural spelling knowledge map is given, the purpose of on-line testing is achieved, the testing accuracy is improved, the spelling and spelling abilities of students are accurately judged, and meanwhile, the testing efficiency can be effectively improved when the testing scale is large. And secondly, in the test process, modes such as recording, spelling and the like which are easy to be operated by the testee are adopted, the operability is strong, the test burden of the testee is not additionally increased, and the remote, delayed and unlimited test can be realized. Moreover, the method adopts the test types of multiple dimensions such as pronunciation test, spelling test and the like, so that the natural spelling ability level of the testee can be tested more comprehensively, and the test accuracy is improved. And the obtained user portrait can more visually reflect the natural spelling ability level of the testee by adding the color identification to the natural spelling knowledge map, so that the visualization effect is more visual and accurate, and the user can quickly distinguish whether the natural spelling knowledge points are mastered. In addition, the mode of labeling the test questions is adopted, so that the accuracy of test question statistics corresponding to each knowledge point can be ensured, and the test credibility is further improved.

In an exemplary embodiment, a possible implementation process is involved for selecting a natural spelling knowledge entity matching a tag in a preset natural spelling knowledge graph according to the tag associated with the natural spelling test question. On the basis of the above embodiment, step S214 can be specifically implemented by the following steps:

step S2142, matching the label associated with the pronunciation test question with a natural spelling knowledge entity in a preset natural spelling knowledge map, and determining the natural spelling knowledge entity corresponding to the pronunciation test question.

Step S2144, matching the label associated with the spelling test question with a natural spelling knowledge entity in a preset natural spelling knowledge map, and determining the natural spelling knowledge entity corresponding to the spelling test question.

The labels associated with the pronunciation test questions comprise knowledge point labels, knowledge block labels, knowledge point level labels, pronunciation labels and question type labels. The labels associated with the spelling test questions comprise knowledge point labels, knowledge block labels, labels of the levels of the knowledge points, spelling labels and question type labels.

Specifically, assuming that the label associated with the pronunciation test question a includes a consonant continuum dr (knowledge point label), a consonant continuum dr (knowledge block label), a second level (knowledge point level label), a spelling test (spelling label) and a reading question (question type label), the server can search the consonant continuum dr entity (knowledge point entity), the consonant continuum entity (knowledge block entity), the second level entity (level entity), the pronunciation test entity (test type entity) and the reading question entity (question type entity) which are matched with the label in the natural spelling knowledge map according to the correspondence between the label and the entity.

Similarly, it is assumed that the tags associated with the spelling test question b include a binary sound oa (knowledge point tag), a binary sound (knowledge block tag), a third level (knowledge point level tag), a spelling test (spelling tag), and a dictation question (question type tag), and the server can search the binary sound oa entity (knowledge point entity), the binary sound entity (knowledge block entity), the third level entity (level entity), the spelling test entity (test type entity), and the dictation question entity (question type entity) matched with the tags in the natural spelling knowledge map according to the correspondence between the tags and the entities.

In the embodiment of the application, the labels associated with the natural spelling test questions are adopted to quickly find the natural spelling knowledge entities matched with the labels, and then the test results of the natural spelling test questions corresponding to each label can be integrated into the natural spelling knowledge map through induction analysis, so that the natural spelling knowledge map can intuitively reflect the natural spelling ability level of the tested person, and the test accuracy is higher.

In an exemplary embodiment, a possible implementation process involving adding color identification to the matching natural spelling knowledge entity based on the test results of each test question. On the basis of the above embodiment, as shown in fig. 3, step S216 may be specifically implemented by the following steps:

step S2162, if the pronunciation test questions and the spelling test questions are associated with the same first label, the test accuracy of the pronunciation test questions and the spelling test questions associated with the first label is counted according to the test result of each test question;

step S2164, if the test accuracy is greater than or equal to a first preset threshold, adding a first color identification to the natural spelling knowledge entity matched with the first label;

step S2166, if the test accuracy is greater than a second preset threshold and less than a first preset threshold, adding a second color identification to the natural spelling knowledge entity matched with the first label;

step S2168, if the test accuracy is less than or equal to the second preset threshold, adding a third color identification to the natural spelling knowledge entity matched with the first label.

The first color identification is used for representing that the natural spelling knowledge corresponding to the natural spelling knowledge entity is mastered. The second color identification is used for representing that the natural spelling knowledge corresponding to the natural spelling knowledge entity needs to be reviewed. And the third color identification is used for representing that the natural spelling knowledge corresponding to the natural spelling knowledge entity is not mastered.

Specifically, if the pronunciation test question and the spelling test question are associated with the same first label, for example, the pronunciation test question a1 and the spelling test question b1 are both associated with the knowledge point label of the consonant link dr, it indicates that the pronunciation test question a1 and the spelling test question b1 are both used to test the grasping level of the testee for the consonant link dr. Then, in order to know the grasping level of the subject for the consonant concatemer dr, the server counts the correct rate of all the test questions (not limited to the pronunciation question a1 and the spelling question b1) associated with the first label according to the test result of each test question, wherein the correct rate is the number of correct test questions containing a certain knowledge point/the number of all test questions containing the knowledge point. After the data summarization, data interpretation is performed: the server adds different color identifiers to the natural spelling knowledge entity by comparing the accuracy with a preset threshold, specifically: if the test accuracy is greater than or equal to a first preset threshold, the server adds a first color identifier, such as a blue identifier, to the natural spelling knowledge entity matched with the first tag. And if the test accuracy is greater than a second preset threshold and smaller than a first preset threshold, adding a second color identifier, such as an orange identifier, to the natural spelling knowledge entity matched with the first label by the server. And if the test accuracy is less than or equal to a second preset threshold, adding a third color identifier, such as a yellow identifier, to the natural spelling knowledge entity matched with the first label by the server. It should be clear that the present application is not limited to the specific examples of the aforementioned color markings, for example, the first color marking may also be a green marking, a purple marking, etc., and the same applies to the second color marking and the third color marking.

Optionally, the first preset threshold is any one of 60% to 100%. The second preset threshold is any one of 0% to 60%. The first preset threshold is different from the second preset threshold.

In the embodiment of the application, the answer accuracy of the tested person under the label can be determined by counting the accuracy of the natural spelling test questions associated with the same label, the answer accuracy of the natural spelling entity corresponding to the label can be further determined, different color marks are added to the natural spelling entity by comparing the answer accuracy with the preset threshold, so that the mastery degree of the tested person on different knowledge points can be more easily distinguished, the tested person can more intuitively and comprehensively know the mastered knowledge and deficiency of the tested person, the tested person can make up the deficiency of the tested person in a targeted manner, and the natural spelling level of the tested person can be effectively improved.

In order to ensure that the first preset threshold and/or the second preset threshold can more accurately represent the natural spelling knowledge level held by the testee, in an exemplary embodiment, the scoring threshold is automatically modified by big data, and on the basis of the above embodiment, as shown in fig. 4, the method further includes the following steps:

step S222, obtaining a plurality of test accuracy rates of a preset number of testees for the natural spelling test questions corresponding to the knowledge point entities;

step S224, if the average test accuracy corresponding to the testees exceeding the preset number threshold is determined to be different from the first preset threshold or the second preset threshold according to the plurality of test accuracy among the preset number of testees, the first preset threshold and/or the second preset threshold corresponding to the knowledge point entity is changed according to the average test accuracy.

Specifically, the server obtains a plurality of test correctness rates of a preset number (for example, 2000) of testees for natural spelling test questions corresponding to knowledge point entities in the natural spelling knowledge entities, wherein each tester corresponds to one test correctness rate. And then, the server judges whether the first preset threshold and/or the second preset threshold needs to be changed or not according to the threshold changing condition. Specifically, if there are more than a predetermined number of subjects, for example, more than 95% of subjects, among all the subjects, and the average test accuracy calculated from their test accuracy is different from the first predetermined threshold or the second predetermined threshold, the first predetermined threshold and/or the second predetermined threshold needs to be changed. The server may directly change the first preset threshold and/or the second preset threshold with the average test accuracy, or may replace the first preset threshold with the average test accuracy, and then replace the second preset threshold with a number obtained by subtracting a value. Therefore, the goal of revising the scoring threshold value of each knowledge point is achieved. For example, the first preset threshold of the "26-letter-name" knowledge block entity is 85%, that is, the test accuracy of all the test questions containing the "26-letter-name" label reaches 85% or more, and the first color identification mark is used to mark that the knowledge block is mastered. However, by analyzing the test correctness of 2000 users in the "26-letter-name" tag test question, 90% of the users achieve 89% of test correctness, and therefore, the server changes the first preset threshold of the "26-letter-name" knowledge block entity to 89%. In a practical application scenario, after a preset time interval, for example, 5 days, the server may modify the scoring threshold periodically for each knowledge point, so that the scoring threshold is dynamically variable, and the scoring threshold is guaranteed to be updated in time. On the other hand, considering that the whole level of the testee is effectively improved due to the change of objective conditions such as a learning method and the like of the testee in different periods, the real ability level of the testee cannot be accurately reflected by using the preset scoring threshold, and on the basis, the scoring threshold is periodically corrected by combining the whole test results of the testees in different batches, so that more accurate ability evaluation can be given.

In the embodiment of the application, the scoring threshold is automatically corrected through big data, so that the scoring threshold can more accurately represent the mastery level of the testee on the knowledge points.

Further, in order to reduce the calculation amount of the score threshold automatic correction, in an exemplary embodiment, the step S224 may be specifically implemented by the following steps: step S224a, if it is determined that, according to the multiple test accuracy rates, the average test accuracy rate corresponding to the test subject exceeding the threshold of the preset number is different from the first preset threshold or the second preset threshold, and the difference between the average test accuracy rate and the first preset threshold or the second preset threshold is greater than or equal to a preset value, the first preset threshold and/or the second preset threshold corresponding to the knowledge point entity is changed according to the average test accuracy rate.

It can be understood that, in this embodiment, it is further required to determine whether the difference between the average test accuracy and the first preset threshold or the second preset threshold is greater than or equal to the preset value, and the server changes the first preset threshold and/or the second preset threshold only when the condition that the difference between the average test accuracy and the first preset threshold or the second preset threshold is greater than or equal to the preset value is further satisfied. Otherwise, if the difference between the average test accuracy and the first preset threshold or the second preset threshold is smaller than the preset value, the server reserves the first preset threshold and/or the second preset threshold corresponding to the knowledge point entity.

In an exemplary embodiment, as shown in fig. 5, the method further comprises the steps of:

step S232, according to the test result of each test question and the knowledge point label associated with each test question, counting the test accuracy of the natural spelling test question corresponding to the knowledge point label;

step S234, analyzing the ability mastering information of the testee to the knowledge point corresponding to the knowledge point label according to the testing accuracy of the knowledge point label and the natural spelling test question corresponding to the knowledge point label;

and step S236, generating and displaying a test report according to the test accuracy, the capability mastering information and the user portrait of the natural spelling test questions corresponding to the knowledge point labels and the knowledge point labels.

Specifically, the server counts the accuracy of the natural spelling test questions corresponding to the knowledge point labels according to the test result of each test question and the knowledge point labels associated with each test question, analyzes the ability grasping information (such as grasped, required to be reviewed and not grasped) of the testee for the knowledge points corresponding to the knowledge point labels according to the accuracy of the natural spelling test questions corresponding to the knowledge point labels and the knowledge point labels, and finally generates a test report according to the ability grasping information of the accuracy of the natural spelling test questions corresponding to the knowledge point labels and the user portrait and displays the test report to the user. Optionally, the user portrait can be displayed as a part of the test report, and a learning suggestion can be accurately given according to the corresponding relation between each knowledge point and each knowledge block contained in the user portrait, so that the user portrait can effectively learn the shortages.

Table 1 is an exemplary test report form:

TABLE 1

In the embodiment of the application, the tested person can intuitively know the self level by generating the test report.

Further, in an exemplary embodiment, the server pushes the exercise content to the tested person terminal through natural spelling of the knowledge graph and the user portrait, so that the tested person can complete corresponding exercises according to the exercise content. For example, when the CVC word is determined to be reviewed and the diphthong is determined to be not mastered, the server automatically pushes a corresponding exercise question to the testee. In an exemplary embodiment, the test type entities include a pronunciation entity and a spelling entity, and the test question type entities include a reading question entity and a dictation question entity. Based on the method, the construction process of the natural spelling knowledge graph can be specifically realized through the following steps:

step S242, dividing the initial knowledge graph into a plurality of levels according to level entities;

step S244, setting a plurality of knowledge block entities in the level corresponding to each level entity, and selecting a connection line representing the correspondence between the plurality of knowledge block entities to connect the plurality of knowledge block entities according to the correspondence between the plurality of knowledge block entities;

step S246, a plurality of knowledge point entities are set in each knowledge block entity, and the inclusion relationship that the knowledge block entity includes the plurality of knowledge point entities is established;

in step S248, in each knowledge point entity, a pronunciation entity and a spelling entity are set, and an inclusion relationship between the knowledge point entity including the pronunciation entity and the spelling entity is established.

Wherein, the corresponding relationship comprises an inclusion relationship, a progressive relationship and an extension relationship.

Specifically, in constructing the natural spelling knowledge-graph, the server first divides the initial knowledge-graph into a plurality of levels, for example, a first level, a second level, a third level, and a fourth level, according to the level entities. After the levels are divided, the server sets a plurality of knowledge block entities in each level, the knowledge block entities in each level are arranged in a column from top to bottom without any special sequence, for example, the first level contains 3 knowledge block entities such as 26 letter names, 26 letter sounds, CVC words, and the like. The server selects connecting lines representing the corresponding relations among the knowledge block entities to connect the knowledge block entities according to the corresponding relations among the knowledge block entities, for example, an inclusion relation (connected by a solid line), a progressive relation (connected by a dotted line), and an extension relation (connected by a dotted line), so that the association of all the knowledge block entities is completed, and the thumbnail is formed. For example, there is a progressive relationship between the 26-letter sounds and the CVC words, i.e., the CVC words can be spelled by applying the knowledge of the 26-letter sounds. Next, each knowledge block is further expanded into knowledge points, specifically: in each knowledge block entity, the server sets a plurality of knowledge point entities, and establishes an inclusion relationship that the knowledge block entity includes a plurality of knowledge point entities, for example, 26 knowledge point entities such as "a-letter sound", "b-letter sound" … … "z-letter sound" are included under "26-letter sound" of the knowledge block entity. Next, in each knowledge point entity, the server sets a pronunciation entity and a spelling entity, and establishes an inclusion relationship that the knowledge point entity includes the pronunciation entity and the spelling entity, so as to represent the test condition of each knowledge point in each ability dimension.

Alternatively, the user may enter the expanded knowledge point entity interface by clicking on the knowledge block entity interface, and may enter the expanded pronunciation entity and spelling entity interface by clicking on the knowledge point entity interface.

In the embodiment of the application, the natural spelling knowledge map is formed by establishing the relation between the natural spelling knowledge points in an entity mode.

In an exemplary embodiment, in order to gradually improve the structure of the natural spelling knowledge graph in the server, the present embodiment uses a deep reinforcement learning model of an artificial intelligence technology, so that the server can automatically improve the structure of the knowledge graph according to the test performance of the user on the basis of a preset knowledge graph. Referring to fig. 6, the method further includes the following steps:

step S272, obtaining a test data sample of the tested person; the test data sample comprises a test result of each test question, a test accuracy rate corresponding to each knowledge point entity and a user portrait;

step S274, analyzing whether the corresponding relation between the knowledge point entities and/or the corresponding relation between the knowledge block entities in the user portrait is accurate or not according to the test result of each test question, the test accuracy of each knowledge point entity and the user portrait;

step S276, if not, adding correction marks to the corresponding relation between inaccurate knowledge point entities in the user portrait and/or the corresponding relation between knowledge block entities, and inputting the user portrait added with the correction marks, the test result of each test question and the test accuracy corresponding to each knowledge point entity into the atlas correction neural network model to be trained for model training to obtain the trained atlas correction neural network model.

Specifically, first, the server collects test data samples of a certain number of testees, for example, the test data samples may be test data samples of 1000 testees, where the test data samples include test results of each test question of the user, test accuracy corresponding to each knowledge point entity, and a user portrait generated based on the current knowledge graph. Alternatively, the server may group the test data samples, for example, 1000 people test data samples into 100 groups, each group of test data samples being subsequently used as a training sample. And then, automatically analyzing the test result of each test question of the user and the test accuracy corresponding to each knowledge point entity in each group of test data samples by the server, and deducing whether the corresponding relation between the knowledge point entities in the user portrait and/or the corresponding relation between the knowledge block entities is accurate according to the current knowledge graph. And if the inaccurate corresponding relation exists, carrying out correction marking on the inaccurate corresponding relation, and recording a correction instruction. In other examples, the above-described marking process may also be performed manually. Finally, it is understood that the test data samples of 1000 testees will result in 100 training samples.

An example is given here to illustrate whether the user representation is accurate. For example, in the current knowledge-graph, it is shown that the consonant-run-along knowledge block entity and the consonant-doublet knowledge block entity are not directly related to each other, although they are the entities at the second level. However, according to the test data samples of a group of testees, when the test accuracy corresponding to the consonant conjoint knowledge block entity of the testees exceeds 90%, the test accuracy corresponding to the consonant conjoint knowledge block entity is about 70%; when the testing accuracy corresponding to the consonant conjoint knowledge block entity of the tested person is less than 70%, the testing accuracy corresponding to the consonant conjoint knowledge block entity is usually lower than 60%, so that the conclusion is drawn that a positive correlation possibly exists between the consonant conjoint knowledge block entity and the consonant conjoint knowledge block entity, namely, the consonant conjoint knowledge block entity is the first-order knowledge of the consonant conjoint knowledge block entity, and a progressive relation is added between the consonant conjoint knowledge block entity and the consonant conjoint knowledge block entity.

And then, inputting the user portrait added with the correction marks, the test result of each test question and the test accuracy corresponding to each knowledge point entity into the atlas correction neural network model to be trained by the server for model training to obtain the trained atlas correction neural network model. The server analyzes the relevance between the knowledge blocks and between the knowledge points and compares the relevance with the current knowledge map, and if the corresponding relation between inconsistent knowledge point entities and/or the corresponding relation between knowledge block entities exist, the server can automatically correct the knowledge blocks. Because the training samples are marked in the steps, the server compares the manual marking with the automatic correction part, records the behavior process of the manual marking and the automatic correction part and optimizes a knowledge graph correction strategy. Thus completing the model training once. It will be appreciated that the atlas-correct neural network model may acquire the ability to automatically correct the knowledge-atlas after repeated training of the atlas-correct neural network model using all training samples, with the steps described above. Therefore, in the subsequent test, the server can intelligently check and optimize the relation between the entities in the knowledge graph according to the user data.

In an exemplary embodiment, the method further comprises the steps of: acquiring historical test data of a tested person in a preset time period; inputting historical test data into the trained atlas correction neural network model to obtain a corrected natural spelling knowledge atlas; and replacing the natural spelling knowledge map with the corrected natural spelling knowledge map. In the embodiment, the natural spelling knowledge graph is optimized by correcting the natural spelling knowledge graph.

In an exemplary embodiment, the method further comprises the steps of:

step S251, responding to the generation request of the pronunciation test question, and analyzing a target investigation point contained in the generation request of the spelling test question; the target survey points are natural spelling knowledge points and comprise target phonemes such as alpha-letter sounds, consonant concatenative dr, ong combinations and the like;

step S252, searching word combination conditions matched with the target investigation point, and reading phonemes to be combined according to the word combination conditions;

step S253, combining the target investigation point and the phoneme to be combined according to a preset word combination condition to obtain a plurality of combined words;

step S254, screening target words which accord with word screening conditions in a plurality of combined words; the word screening conditions include discarding words having actual meanings;

and step S255, associating the target word with a label corresponding to the pronunciation test question to obtain the pronunciation test question.

The phoneme is the smallest unit in the english phonetic system, and is divided into two major categories, namely consonant and vowel. The word combination condition comprises a condition for combining the target investigation point and the phoneme to be combined. Optionally, the sum of the target survey point and the number of phonemes to be combined is less than or equal to three.

The word screening condition comprises discarding the combined words which do not accord with the English voice rule and/or discarding the words which contain the actual meaning.

Specifically, when a question is required, the server responds to a pronunciation test question generation request, analyzes the pronunciation test question generation request to obtain a target examination point, such as a long vowel o, matches a corresponding word combination condition for the target examination point from a word combination condition library, further reads a phoneme to be combined from a knowledge point phoneme library based on the phoneme to be combined in the word combination condition, and combines the target examination point and the phoneme to be combined according to the matched word combination condition to obtain a plurality of combined words. Then, the server matches the plurality of combined words with the words in the dictionary database, if the matching is successful, the combined words are abandoned, and the words which are not successfully matched are reserved as target words, so that the server associates the labels corresponding to the pronunciation test questions with the target words to obtain the pronunciation test questions.

Wherein the dictionary database contains all words with practical meaning.

Wherein the server determines that the matching is successful when the combined word is searchable in the dictionary database.

For example, if a test question of a long vowel o is requested to be generated, the target examination point is the "long vowel o", the matching word combination condition of the long vowel o is 1 consonant + e ", according to the word combination condition of the long vowel o, the server needs to randomly extract 2 consonant phonemes as the phonemes to be combined, the formed words may be r + o + p + e, i.e., rope, t + o + g + e, i.e., tage, and the like, and for the possible words rope, toe, joke, fobe … …, rope and joke, the remaining words such as rope or fobe are screened out as the really existing words, and the matching labels such as the" long vowel o "," long vowel "," second level "," pronunciation question "," test question "," reading question ".

For another example, if a test question of an ar combination is requested to be generated, the target examination point is an "ar combination", the matching word combination condition is that "ar combination word combination condition 1 is 1 consonant + ar combination, ar combination word combination condition 2 is 1 consonant + ar combination +1 consonant", according to the word combination condition 1 of the ar combination, the server randomly extracts 1 consonant phoneme as a phoneme to be combined, and the formed words may be c + ar, i.e., car, s + ar, i.e., sar, and the like; then according to the word combination condition 2 of ar combination, the server randomly extracts 2 consonant phonemes as phonemes to be combined, the formed words may be p + ar + t, namely part, l + ar + g, namely larg, and the like, true words car, part, and the like are screened out from the generated words car, sar, part, and larg … …, and the remaining earning question banks of sar and larg are provided with matching labels of "ar combination", "r-controlled volwels", "third level", "pronunciation test questions", and "reading questions".

In the embodiment of the application, the purpose of automatically generating pronunciation test questions is realized by setting word combination conditions and word screening conditions and adding labels to finally obtained words, so that a large number of pronunciation test questions can be generated quickly to meet the increasing test requirements.

In an exemplary embodiment, the method further comprises the steps of:

step S261, responding to the spelling test question generation request, and analyzing target investigation points contained in the spelling test question generation request; the target investigation point is a natural spelling knowledge point and comprises a target phoneme;

step S262, searching word combination conditions matched with the target investigation point, and reading phonemes to be combined according to the word combination conditions;

step S263, combining the target investigation point and the phoneme to be combined according to a preset word combination condition to obtain a plurality of combined words;

step S264, screening target words which accord with word screening conditions in a plurality of combined words; the word screening conditions include discarding words having actual meanings;

step 265, acquiring pronunciation data of each phoneme contained in the target word, and combining the pronunciation data of each phoneme to obtain pronunciation information of the target word;

step S266, associate the pronunciation information with the label corresponding to the spelling test question to obtain the spelling test question.

Specifically, when a question is required, the server responds to a spelling test question generation request, analyzes the spelling test question generation request to obtain a target examination point, such as a consonant sh, matches a corresponding word combination condition for the target examination point from a word combination condition library, further reads a phoneme to be combined from a knowledge point phoneme library based on the phoneme to be combined in the word combination condition, and combines the target examination point and the phoneme to be combined according to the matched word combination condition to obtain a plurality of combined words. Then, the server matches the plurality of combined words with the words in the dictionary database, and if the matching is successful, the combined words are discarded, and the words which are not successfully matched are reserved as target words. Then, the server reads and combines the pronunciation data of each phoneme in the target word from the phoneme pronunciation library to obtain the pronunciation information of the target word, and the server associates the pronunciation information of the target word with a label corresponding to the spelling test question to obtain the spelling test question.

The pronunciation database stores pronunciation data of all single English phonemes and stores the pronunciation data in the form of an audio file.

For example, if a test question of a consonant sh is requested to be generated, the target investigation point is the consonant sh, the matching word combination condition is that the consonant sh word combination condition is that consonant sh +1 vowel +1 consonant, according to the consonant sh word combination condition, the server needs to randomly extract 1 vowel phoneme and 1 consonant phoneme as phonemes to be combined, the formed words may be sh + o + p, that is, a shot, sh + u + g, that is, etc., for the possible words shot, shot … …, and shot as really existing words, the word is screened out, and the word such as shot, etc. is left, and the phonetic notation is/shot/etc., the word and audio information item library of the word such as shot +/shot/etc. is collected and matched with the label of the consonant sh, the consonant level, the second spelling test question, and the second question level, "dictation questions".

For another example, if an ell combined test question is requested to be generated, the target survey point is "ell combination", the matching word combination condition is "ell combined word combination condition is 1 consonant + ell combination", according to ell combined word combination condition, the server randomly extracts 1 consonant phoneme as a phoneme to be combined, the formed word may be r + ell (all, t + ell (all), etc., for the possible words, all, pell … …, the true word tell, etc. is screened out, the rest words are annotated, for example, all is/all/, the words of all +/all/etc. are imported into the question bank, and the matching labels are "ell combination", "complex letter combination", "fourth level", "spelling test question", "dictation question".

In the embodiment of the application, the purpose of automatically generating spelling test questions is realized by setting word combination conditions and word screening conditions and adding labels to finally obtained words, so that a large number of spelling test questions can be generated quickly to meet the increasing test requirements.

In an exemplary embodiment, the natural spelling test comprises a plurality of different test scenarios. Specifically, the test scenarios may be classified into diagnostic test scenarios, formative test scenarios, and final test scenarios, and the test paper generation modes of the test scenarios are different.

In one example, all knowledge points at a certain level are covered for both diagnostic test scenarios and terminal test scenarios. Based on the above, the test paper generation process comprises the following steps:

step S282, selecting a type and a level of a test scenario, for example, a tester may select a scenario "diagnostic test scenario" and a level "second level";

step S284, the server generates a set of test paper according to the selected test scenario type and level, where the test paper may include the above-described 4 test types, and the ratio of the number of questions of each test type may be the number of pronunciation questions: number of spelling questions: number of speech awareness questions: the number of the comprehensive application test questions is 4:4:1: 1.

in another example, for the formative test scenario, since the formative test occurs in the learning phase of the testee and the main purpose of the test is to play a reverse role in teaching, the tester can select the test range autonomously, and the test paper generation process thereof includes the following steps:

step S292 is to select a test scenario type and a test scenario level, for example, a tester may select a scenario "formative test scenario" and a test scenario level "second level";

in step S294, the tester may set the test type, knowledge block/knowledge point of the test, and may select one or more test points, such as "pronunciation test" + "consonant concatenation" or "speech awareness test" + "consonant concatenation bl" + "short vowel a" or "pronunciation test" + "spelling test" + "consonant concatenation bl".

Step S296, the server extracts corresponding test questions from the question bank to generate a set of test paper according to the range selected by the tester, and if a plurality of test types are involved, the number of the test questions of each test type can be designed according to the proportion of 4:4:1: 1; if there is only one test type, the number of test questions can be set as a default value, but not less than 40.

In the embodiment of the application, different test scenes are set, so that the test functions are diversified, different test requirements of users are met, and the improvement of user experience is facilitated.

In order to improve the accuracy of audio acquisition of a testee, noise reduction needs to be performed on acquired audio data. In an exemplary embodiment, a server obtains preset environment sound data and audio data, then analyzes the audio data and the environment sound data respectively to obtain audio codes and environment sound codes, then performs audio mixing processing based on the audio codes and the environment sound codes to obtain audio mixing codes, and finally performs noise reduction processing on the audio mixing codes based on the environment sound codes to obtain noise reduction data.

Specifically, in order to avoid audio data distortion and noise data, the terminal of the testee collects environmental sound data within a preset range through a built-in recording module. The environmental sound data refers to sound of a real environment around a position where the current test subject terminal is located. The acquired environmental sound data can reflect real and natural noise-free situations to the maximum extent, the environmental sound data in a preset range is acquired and can be used as reference calling data of subsequent audio data to relate to the noise reduction effect of the audio data, so that the accuracy of acquiring the environmental sound data has high requirements on hardware of the recording module, and the recording effect of the recording module can restore the real effect. Next, after obtaining the audio data and the environmental sound data, the server analyzes and analyzes the audio data and the environmental sound data to analyze the audio data and the environmental sound data into audio codes and environmental sound codes in coded forms. The audio data and the ambient sound data are converted into audio coding and ambient sound coding of the signal frequency band type, for example, by audio data parsing. The audio frequency coding and the environment sound coding of the signal frequency band type can visually display the audio data characteristics and the environment sound data characteristics, and meanwhile, the identification degree of the audio data and the environment sound data is favorably improved, and the subsequent operation and processing are facilitated. Next, the server mixes the audio coding and the environment sound coding, that is, mixes the audio coding and the environment sound coding into the same audio track, and presents the audio coding and the environment sound coding in an encoding form, thereby obtaining mixed coding. Next, after the server obtains the audio mixing code, the server may perform noise reduction processing on the audio mixing code through the environment sound code, so as to filter, isolate or soften a part of discrete codes that do not conform to the characteristics of the audio mixing code in the audio mixing code, thereby reducing or eliminating the discrete noise codes.

In order to improve the recognition accuracy of English letters in a spelling image, in an exemplary embodiment, a server acquires the spelling image, extracts an image block containing a spelling object from the spelling image, performs character recognition on characters in the image block, acquires an initial recognition result, acquires a description file, wherein the description file comprises constraint information used for indicating rule requirements that the characters in the image block need to meet, aligns a coordinate system of the image block with an image coordinate system specified in the description file, and finally corrects the initial recognition result by using at least part of the constraint information in the description file to acquire a final recognition result.

The description file is a predefined document, which can describe the format and attributes that the characters in the object to be recognized usually follow, that is, it specifies the rule requirements that the characters in the image block need to conform to. The description file may be used to provide a priori information about the object to be identified, thereby assisting in the correction of the primary recognition result. Constraint information refers to information in the description file indicating rule requirements that the characters in the image block need to conform to.

Specifically, the server divides the spelling image into image blocks containing the spelling object by means of cropping, and after extracting the image blocks containing the spelling object, the server can perform certain preprocessing on the image blocks, such as inclination correction, contrast adjustment and the like on the image blocks, so that the preprocessed image blocks can be more easily subjected to character recognition. Next, the server may employ Optical Character Recognition (OCR) to recognize the text in the image block. Next, the server aligns the coordinate system of the image block with the image coordinate system specified in the description file, and associates the sub-image blocks in the image block with the fields in the description file according to the coordinates of the aligned image block and the coordinates of each field in the description file. In general, the position of a character in an object to be recognized such as a test paper is approximately fixed, and therefore the accuracy of determination of such a correspondence relationship is high.

It should be understood that although the various steps in the flow charts of fig. 2-6 are shown in order as indicated by the arrows, the steps are not necessarily performed in order as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least some of the steps in fig. 2-6 may include multiple steps or multiple stages, which are not necessarily performed at the same time, but may be performed at different times, which are not necessarily performed in sequence, but may be performed in turn or alternately with other steps or at least some of the other steps.

Based on the same inventive concept, in an exemplary embodiment, as shown in fig. 7, an entity tagging method is further provided, which is described by taking the application of the method to the server in fig. 1 as an example, and specifically can be implemented by the following steps:

step S302, responding to a natural spelling test request triggered by a testee through a testee terminal, and sending first test data containing natural spelling test questions to the testee terminal so as to enable the tester terminal to display the natural spelling test questions;

step S304, receiving and storing the audio data and spelling information sent by the tested person terminal, wherein the audio data and spelling information are obtained when the tested person carries out natural spelling test according to the natural spelling test question;

step S306, sending the audio data, spelling information and second test data containing the natural spelling test questions and answers to a tester terminal, and receiving the test result of each test question in the natural spelling test questions uploaded by the tester terminal;

step S308, selecting a natural spelling knowledge entity matched with the label from a preset natural spelling knowledge map according to the label associated with the natural spelling test question;

step S310, according to the test result of each test question, adding a color identifier to the matched natural spelling knowledge entity, determining the natural spelling knowledge map containing the color identifier as the user portrait of the tested person, and displaying the user portrait through the terminal of the tested person.

Specifically, the specific implementation process of step S302 to step S310 may refer to the specific implementation process of step S202 to step S212, which is not described herein again.

In the entity marking method, the whole testing process is completed through interaction of the tester terminal, the tested tester terminal and the server, the user portrait which reflects the testing result and is based on the natural spelling knowledge map is given, the purpose of on-line testing is achieved, the testing accuracy is improved, the spelling and spelling abilities of students are accurately judged, and meanwhile, the testing efficiency can be effectively improved when the testing scale is large. And secondly, the remote, time-delay and unlimited testing can be realized. Moreover, the natural spelling knowledge map is used for recording and reflecting the test result, the test accuracy can be improved, the obtained user portrait can more visually reflect the natural spelling capability level of the tested person by adding the color identification to the natural spelling knowledge map, the visual effect is more visual and accurate, and the user can quickly distinguish whether the natural spelling knowledge point is mastered. In addition, the mode of labeling the test questions is adopted, so that the statistical accuracy of the test questions corresponding to each knowledge point can be ensured, and the test accuracy is further improved.

In an exemplary embodiment, the natural spelling questions comprise pronunciation questions and spelling questions; adding color marks to the matched natural spelling knowledge entities according to the test result of each test question, comprising the following steps:

if the pronunciation test questions and the spelling test questions are associated with the same first label, counting the test accuracy of the pronunciation test questions and the spelling test questions associated with the first label according to the test result of each test question;

if the test accuracy is greater than or equal to a first preset threshold value, adding a first color identifier to the natural spelling knowledge entity matched with the first label; the first color identification is used for representing that the natural spelling knowledge corresponding to the natural spelling knowledge entity is mastered;

if the test accuracy is greater than a second preset threshold and smaller than a first preset threshold, adding a second color identifier to the natural spelling knowledge entity matched with the first label; the second color identification is used for representing that the natural spelling knowledge corresponding to the natural spelling knowledge entity needs to be reviewed;

if the test accuracy is smaller than or equal to a second preset threshold value, adding a third color identifier to the natural spelling knowledge entity matched with the first label; and the third color identification is used for representing that the natural spelling knowledge corresponding to the natural spelling knowledge entity is not mastered.

Specifically, the specific implementation process of each step in this embodiment may refer to the specific implementation process of steps S2162 to S2168, which is not described herein again.

In an exemplary embodiment, the method further comprises the steps of:

acquiring a plurality of test accuracy rates of a preset number of testees for the natural spelling test questions corresponding to the knowledge point entities;

if the average test accuracy corresponding to the tested persons exceeding the preset number threshold is different from the first preset threshold or the second preset threshold according to the plurality of test accuracy among the preset number of tested persons, the first preset threshold and/or the second preset threshold corresponding to the knowledge point entity are/is changed according to the average test accuracy.

Specifically, the specific implementation process of each step in this embodiment may refer to the specific implementation process of step S222 to step S224, which is not described herein again.

In an exemplary embodiment, the method further comprises the steps of:

according to the test result of each test question and the knowledge point label associated with each test question, the test accuracy of the natural spelling test question corresponding to the knowledge point label is counted;

analyzing the ability mastering information of the testee to the knowledge points corresponding to the knowledge point labels according to the testing accuracy of the knowledge point labels and the natural spelling test questions corresponding to the knowledge point labels;

and generating and displaying a test report according to the test accuracy, capability mastering information and user portrait of the natural spelling test questions corresponding to the knowledge point labels and the knowledge point labels.

Specifically, the specific implementation process of each step in this embodiment may refer to the specific implementation process of steps S232 to S236, which is not described herein again.

In an exemplary embodiment, as shown in fig. 8, there is provided an entity tagging apparatus including: a test data sending module 302, an audio data receiving module 304, a spelling information receiving module 306, a test result determining module 308, a matching module 310, and an identification adding module 312, wherein:

the test data sending module 302 is configured to send first test data including natural spelling test questions to a testee terminal in response to a natural spelling test request triggered by the testee through the testee terminal, so that the tester terminal displays the natural spelling test questions; wherein, the natural spelling test questions comprise pronunciation test questions and spelling test questions;

the audio data receiving module 304 is configured to receive and store audio data sent by a terminal of a testee, where the audio data is obtained by recording when the testee performs a pronunciation test according to a pronunciation test question;

the spelling information receiving module 306 is used for receiving and storing spelling information sent by the tested person terminal, wherein the spelling information is information input in the tested person terminal when the tested person carries out spelling test according to spelling test questions;

the test result determining module 308 is configured to send the audio data, the spelling information, and the second test data including the natural spelling test questions and the answers to the tester terminal, and receive a test result of each test question in the natural spelling test questions uploaded by the tester terminal; the test result comprises result correct and result error;

the matching module 310 is configured to select a natural spelling knowledge entity matched with a tag in a preset natural spelling knowledge map according to the tag associated with the natural spelling test question; the natural spelling knowledge entity comprises a knowledge block entity, a level entity, a test type entity and a test question type entity, wherein the knowledge block entity comprises at least one knowledge point entity;

the mark adding module 312 is configured to add a color mark to the matched natural spelling knowledge entity according to a test result of each test question, determine a natural spelling knowledge graph containing the color mark as a user portrait of the testee, and display the user portrait through the terminal of the testee.

Among the above-mentioned entity mark device, accomplish whole test flow and give the user portrait that reflects the test result based on the knowledge map of spelling naturally through the interaction of tester terminal, testee terminal and server, reach the purpose of on-line test, be favorable to improving the test accuracy degree, and then accurate judgement student spells out and read and spelling ability, can effectively promote efficiency of software testing when the test scale is great simultaneously. And secondly, in the test process, modes such as recording, spelling and the like which are easy to be operated by a testee are adopted, so that the operability is strong, the operation is easy, the test burden of the testee is not additionally increased, and the remote, delayed and unlimited test can be realized. Moreover, the method adopts the test types of multiple dimensions such as pronunciation test, spelling test and the like, so that the natural spelling ability level of the testee can be tested more comprehensively, and the test accuracy is improved. And the obtained user portrait can more intuitively reflect the natural spelling ability level of the tested person by adding the color identification to the natural spelling knowledge map, the visualization effect is more intuitive and accurate, and the user can quickly distinguish whether the natural spelling knowledge point is mastered. In addition, the mode of labeling the test questions is adopted, so that the statistical accuracy of the test questions corresponding to each knowledge point can be ensured, and the test accuracy is further improved.

For the specific definition of the entity labeling apparatus, reference may be made to the above definition of the entity labeling method, which is not described herein again. The various modules in the physical token described above may be implemented in whole or in part by software, hardware, and combinations thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.

In an exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as shown in fig. 9. The computer device includes a processor, a memory, and a network interface connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The database of the computer device is used for storing natural spelling test questions and answers thereof. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to implement a method of entity tagging.

Those skilled in the art will appreciate that the architecture shown in fig. 9 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.

In one embodiment, a computer device is further provided, which includes a memory and a processor, the memory stores a computer program, and the processor implements the steps of the above method embodiments when executing the computer program.

In an exemplary embodiment, a computer-readable storage medium is provided, on which a computer program is stored which, when being executed by a processor, carries out the steps of the above-mentioned method embodiments.

It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, database or other medium used in the embodiments provided herein can include at least one of non-volatile and volatile memory. Non-volatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical storage, or the like. Volatile Memory can include Random Access Memory (RAM) or external cache Memory. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM), among others.

The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.

The above-mentioned embodiments only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present patent shall be subject to the appended claims.

Claims

1. A method of entity tagging, the method comprising:

selecting a natural spelling knowledge entity matched with the label from a preset natural spelling knowledge map according to the label associated with the natural spelling test question; the natural spelling knowledge entity comprises a knowledge block entity, a level entity, a test type entity and a test question type entity, wherein the knowledge block entity comprises at least one knowledge point entity, and the test type entity comprises a pronunciation entity and a spelling entity;

adding a color identifier to the matched natural spelling knowledge entity according to the test result of each test question, determining a natural spelling knowledge map containing the color identifier as a user portrait of the tested person, and displaying the user portrait through the tested person terminal;

wherein, the construction process of the natural spelling knowledge graph comprises the following steps:

dividing the initial knowledge graph into a plurality of levels according to the level entities;

setting a plurality of knowledge block entities in the level corresponding to each level entity, and selecting connecting lines representing the corresponding relations among the knowledge block entities to connect the knowledge block entities according to the corresponding relations among the knowledge block entities; the corresponding relation comprises an inclusion relation, a progressive relation and an extension relation;

setting a plurality of knowledge point entities in each knowledge block entity, and establishing an inclusion relation of the knowledge block entities including the plurality of knowledge point entities;

in each knowledge point entity, a pronunciation entity and a spelling entity are set, and an inclusion relationship that the knowledge point entity includes the pronunciation entity and the spelling entity is established.

2. The method of claim 1, wherein selecting a natural spelling knowledge entity matching the tag in a preset natural spelling knowledge graph according to the tag associated with the natural spelling test question comprises:

matching the label associated with the pronunciation test question with a natural spelling knowledge entity in a preset natural spelling knowledge map, and determining the natural spelling knowledge entity corresponding to the pronunciation test question; the labels associated with the pronunciation test questions comprise knowledge point labels, knowledge block labels, knowledge point level labels, pronunciation labels and question type labels;

matching the label associated with the spelling test question with a natural spelling knowledge entity in a preset natural spelling knowledge map, and determining the natural spelling knowledge entity corresponding to the spelling test question; the labels associated with the spelling test questions comprise knowledge point labels, knowledge block labels, labels of the levels of the knowledge points, spelling labels and question type labels.

3. The method according to claim 2, wherein adding color identification to the matched natural spelling knowledge entity according to the test result of each test question comprises:

if the test accuracy is greater than or equal to a first preset threshold value, adding a first color identifier to the natural spelling knowledge entity matched with the first label; the first color identification is used for representing that natural spelling knowledge corresponding to the natural spelling knowledge entity is mastered;

if the test accuracy is larger than a second preset threshold and smaller than the first preset threshold, adding a second color identifier to the natural spelling knowledge entity matched with the first label; the second color identification is used for representing that the natural spelling knowledge corresponding to the natural spelling knowledge entity needs to be reviewed;

if the test accuracy is smaller than or equal to the second preset threshold, adding a third color identifier to the natural spelling knowledge entity matched with the first label; the third color identification is used for representing that the natural spelling knowledge corresponding to the natural spelling knowledge entity is not mastered;

the method further comprises the following steps:

acquiring a plurality of test accuracy rates of a preset number of testees for the natural spelling test questions corresponding to the knowledge point entity;

and if the average test accuracy corresponding to the tested persons exceeding the preset number threshold is different from the first preset threshold or the second preset threshold according to the plurality of test accuracy among the preset number of tested persons, changing the first preset threshold and/or the second preset threshold corresponding to the knowledge point entity according to the average test accuracy.

4. The method according to claim 3, wherein if it is determined that, among the predetermined number of testees, according to the test accuracy rates, an average test accuracy rate corresponding to the testees exceeding a predetermined number of thresholds is different from the first predetermined threshold or the second predetermined threshold, changing the first predetermined threshold and/or the second predetermined threshold corresponding to the knowledge point entity according to the average test accuracy rate comprises:

if the average test accuracy corresponding to the tested persons exceeding the preset number threshold is determined to be different from the first preset threshold or the second preset threshold according to the plurality of test accuracy among the preset number of tested persons, and the difference between the average test accuracy and the first preset threshold or the second preset threshold is larger than or equal to a preset value, the first preset threshold and/or the second preset threshold corresponding to the knowledge point entity is changed according to the average test accuracy;

and if the difference between the average test accuracy and the first preset threshold or the second preset threshold is smaller than a preset value, retaining the first preset threshold and/or the second preset threshold corresponding to the knowledge point entity.

5. The method of claim 1, further comprising:

according to the test result of each test question and the knowledge point label associated with each test question, counting the test accuracy of the natural spelling test question corresponding to the knowledge point label;

analyzing the ability mastering information of the testee for the knowledge points corresponding to the knowledge point labels according to the test accuracy of the knowledge point labels and the natural spelling test questions corresponding to the knowledge point labels;

and generating and displaying a test report according to the test accuracy of the natural spelling test questions corresponding to the knowledge point labels, the capability mastering information and the user portrait.

6. The method of claim 1, wherein the test question type entities comprise a reading question entity and a dictation question entity.

7. The method of claim 6, further comprising:

acquiring a test data sample of a tested person; the test data sample comprises a test result of each test question, the accuracy of each knowledge point and a user portrait;

analyzing whether the corresponding relation between the knowledge point entities in the user portrait and/or the corresponding relation between the knowledge block entities is accurate or not according to the test result of each test question, the accuracy of each knowledge point and the user portrait;

and if not, adding correction marks to the corresponding relation between inaccurate knowledge point entities in the user portrait and/or the corresponding relation between knowledge block entities, and inputting the user portrait added with the correction marks, the test result of each test question and the accuracy of each knowledge point into the atlas correction neural network model to be trained for model training to obtain the trained atlas correction neural network model.

8. The method of claim 7, further comprising:

acquiring historical test data of a tested person in a preset time period;

inputting the historical test data into the trained atlas correction neural network model to obtain a corrected natural spelling knowledge atlas;

and replacing the natural spelling knowledge map by the corrected natural spelling knowledge map.

9. The method of claim 1, further comprising:

responding to a pronunciation test question generation request, and analyzing a target investigation point contained in the pronunciation test question generation request; the target investigation point is a natural spelling knowledge point and comprises a target phoneme;

searching a word combination condition matched with the target investigation point, and reading a phoneme to be combined according to the word combination condition;

combining the target investigation point and the phoneme to be combined according to a preset word combination condition to obtain a plurality of combined words;

screening target words which meet word screening conditions in the plurality of combined words; the word screening condition comprises discarding words with actual meanings;

associating the target word with a label corresponding to the pronunciation test question to obtain the pronunciation test question;

alternatively, the first and second electrodes may be,

responding to a spelling test question generation request, and analyzing target investigation points contained in the spelling test question generation request; the target investigation point is a natural spelling knowledge point and comprises a target phoneme;

acquiring pronunciation data of each phoneme contained in the target word, and combining the pronunciation data of each phoneme to obtain pronunciation information of the target word;

and associating the pronunciation information with the label corresponding to the pronunciation test question to obtain the spelling test question.

10. A method of entity tagging, the method comprising:

11. The method of claim 10, wherein the natural spelling questions comprise pronunciation questions and spelling questions; adding color identification to the matched natural spelling knowledge entity according to the test result of each test question, comprising:

if the test accuracy is smaller than or equal to the second preset threshold, adding a third color identifier to the natural spelling knowledge entity matched with the first label; and the third color identification is used for representing that the natural spelling knowledge corresponding to the natural spelling knowledge entity is not mastered.

12. A natural spelling test apparatus, the apparatus comprising:

the matching module is used for selecting a natural spelling knowledge entity matched with the label from a preset natural spelling knowledge map according to the label associated with the natural spelling test question; the natural spelling knowledge entity comprises a knowledge block entity, a level entity, a test type entity and a test question type entity, wherein the knowledge block entity comprises at least one knowledge point entity, and the test type entity comprises a pronunciation entity and a spelling entity;

the mark adding module is used for adding a color mark to the matched natural spelling knowledge entity according to the test result of each test question, determining a natural spelling knowledge map containing the color mark as a user portrait of the tested person, and displaying the user portrait through the tested person terminal;

the map building module is used for dividing the initial knowledge map into a plurality of levels according to the level entities; setting a plurality of knowledge block entities in the level corresponding to each level entity, and selecting connecting lines representing the corresponding relations among the knowledge block entities to connect the knowledge block entities according to the corresponding relations among the knowledge block entities; the corresponding relation comprises an inclusion relation, a progressive relation and an extension relation; setting a plurality of knowledge point entities in each knowledge block entity, and establishing an inclusion relation of the knowledge block entities including the plurality of knowledge point entities; in each knowledge point entity, a pronunciation entity and a spelling entity are set, and an inclusion relationship that the knowledge point entity includes the pronunciation entity and the spelling entity is established.

13. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor realizes the steps of the method of any one of claims 1 to 11 when executing the computer program.