JP2022054636A - Speaking test automatic grading system - Google Patents

Abstract

To provide a speaking test automatic grading system that automatically grades a speaking test for measuring an English ability.SOLUTION: The speaking test automatic grading system according to the present invention includes: a test service server for receiving voice data as an answer to a test from a user terminal; an automatic grading server for receiving the voice data and a request to grade from the test service server and automatically grading the voice data; a grading data storage unit for storing data necessary for automatic grading; and a grading result storage unit for storing the result of automatic grading. The automatic grading server writes the grading data to the grading data storage unit or reads the grading data from the grading data storage unit. The automatic grading server sends the result of automatic grading to the test service server, and the test service server sends the result of grading to the user terminal while stores the result of grading into the grading result storage unit.SELECTED DRAWING: Figure 1

Description

There is an application for application of Article 30, Paragraph 2 of the Patent Act. service / 2020/0617/8865 / "," https: //www.rejob.co.jp/wp-patent/uploads/2020/06/202020617_RareJob PROGOS development_final5.pdf "," html: //info. .Ai / ”and“ https: //info.progos.ai/abouut.html ”are published on the website of RareJob Co., Ltd., which is published on the website.

The present invention relates to an automatic speaking test scoring system, and more particularly to an automatic speaking test scoring system for automatically scoring a speaking test for measuring English proficiency.

In recent years, there has been an increasing number of forms in which English conversation learners and instructors, such as online English conversation, take lessons remotely via a network such as the Internet (registered trademark). Along with this, there is a need for a mechanism for remotely performing tests to measure learners' English proficiency and learning effects and promptly obtaining test results.

In particular, the speaking test for measuring conversational ability is different from the selective test in which the answer is selected from the options and the written test in which the answer is entered or entered in characters, and the respondent's free speech by voice is used. Need to be scored.

On the other hand, many conventional automatic scoring systems are for scoring answers to multiple-choice or one-question-and-answer questions, and are not suitable for scoring free speech by voice.

As the subject of scoring the speech of the respondent, in Patent Document 1, in the method of training and evaluating the foreign language speaking ability by using the voice recognition system, "the content spoken by the user using the voice recognition technology is described. In the process of converting to characters and automatically evaluating the content of foreign language utterances, in order to make up for errors in voice recognition technology, the uttered user is made to input the utterance contents in characters using an input device such as a keyboard. The input character data and the recorded voice data are compared to evaluate the user's foreign language ability, and an accurate evaluation result is provided to the user. "

However, in Patent Document 1, since it is necessary for the user to re-input the utterance content as characters by himself / herself, there is a problem that the user takes time and effort and interrupts the natural flow of conversation. Further, in the case of a long utterance, there is a problem that it is difficult for the user to faithfully reproduce the content of his / her utterance.

Further, in Patent Document 2, the user serves as a predetermined model in a server system that calculates language conversation ability from pronunciation data pronounced based on voice data as a model by utilizing the accuracy of voice recognition. It discloses that it listens to voice data, sends pronunciation data that is pronounced exactly as it is heard, to the server system, and measures the accuracy of pronunciation.

However, Patent Document 2 purely measures the accuracy of pronunciation, and has a problem that not only pronunciation but also various items can be evaluated from various aspects and comprehensive speaking ability cannot be measured. ..

Japanese Unexamined Patent Publication No. 2015-011348 Japanese Unexamined Patent Publication No. 2006-208644

Therefore, the present invention solves the above-mentioned problems, and not only pronunciation but also vocabulary and expression width, grammar / word usage accuracy, fluency, conversation flow control, story composition method and language selection consistency. It provides a speaking test automatic scoring system that can measure comprehensive speaking ability such as intonation and intonation according to the intended content.

In order to solve the above problems, the present invention provides a speaking test automatic scoring system that analyzes the utterances of the respondents of the speaking test and measures the speaking ability.

The speaking test automatic scoring system according to a certain aspect of the present invention receives voice data and a scoring request from a test service server that receives voice data as a test answer from a user terminal, and automatically scores the voice data. It is equipped with an automatic scoring server, a scoring data storage unit that stores data required for automatic scoring, and a scoring result storage unit that stores the scoring results of automatic scoring. The scoring data is written and read, the voice data is automatically graded using the scoring data, the automatic scoring server sends the scoring result of the automatic scoring to the test service server, and the test service server performs the scoring result. Is stored in the scoring result storage unit, and the scoring result is transmitted to the user terminal.

In the speaking test automatic scoring system according to a certain aspect of the present invention, the automatic scoring server includes a voice recognition unit, and the voice recognition unit receives voice input from the test service server, recognizes the input voice, and results in voice recognition. Is stored in the scoring data storage unit.

In the speaking test automatic scoring system according to a certain aspect of the present invention, the automatic scoring server includes a natural language processing unit, and the natural language processing unit receives a voice recognition result from the voice recognition unit and receives the voice recognition result. It is characterized by performing natural language processing and transmitting the natural language processing result to a scoring data storage unit.

In the speaking test automatic scoring system according to an aspect of the present invention, the automatic scoring server includes a scoring data processing unit, and the scoring data processing unit receives a scoring processing request from the test service server and receives a scoring data. Therefore, it is characterized in that processing is performed according to each of a plurality of predetermined items.

In the speaking test automatic scoring system according to a certain aspect of the present invention, the automatic scoring server includes an item evaluation scoring unit, and the item evaluation scoring unit automatically scores each of a plurality of predetermined items.

In the speaking test automatic scoring system according to a certain aspect of the present invention, the automatic scoring server is provided with an overall evaluation scoring unit, and the overall evaluation scoring unit is based on the scoring results for each of a plurality of items automatically graded by the item evaluation scoring unit. And score the overall evaluation.

In the speaking test automatic scoring system according to an aspect of the present invention, the automatic scoring server includes a scoring result processing unit, and the scoring result processing unit transmits the scoring result to the test service server.

In the speaking test automatic scoring system according to a certain aspect of the present invention, the test service server includes a test answer receiving unit, and when the test answer receiving unit receives voice data as a test answer from a user terminal, the voice data is stored in the automatic scoring server. Is transmitted, and a request for scoring processing is transmitted.

In the speaking test automatic scoring system according to a certain aspect of the present invention, the test service server includes a scoring result processing unit, and the scoring result processing unit receives the scoring result from the automatic scoring server and stores the scoring result in the scoring result storage unit. It is characterized by doing.

In the automatic scoring method using the speaking test automatic scoring system according to a certain aspect of the present invention, the speaking test automatic scoring system includes a test service server that receives voice data as a test answer from a user terminal, and voice data and scoring from a test service server. It is equipped with an automatic scoring server that receives requests and automatically grades voice data, a scoring data storage unit that stores data required for automatic scoring, and a scoring result storage unit that stores the scoring results of automatic scoring. The automatic scoring server has a step of writing and reading scoring data to the scoring data storage unit, the automatic scoring server has a step of automatically scoring voice data using the scoring data, and the automatic scoring server has an automatic scoring server. The test service server includes a step of transmitting the scoring result of the automatic scoring to the test service server, and a step of storing the scoring result in the scoring result storage unit and transmitting the scoring result to the user terminal.

An automated speaking test scoring system according to an aspect of the present invention may be provided with a program comprising performing each step of the above method.

In the present invention, "CEFR" refers to the Common European Framework of Reference for Languages. The CEFR (Common European Framework of Reference for Languages) is a framework for language proficiency common to the world and languages, and is established as a common standard in Europe to show the acquisition status and linguistic performance of foreign language learners.

In the present invention, the "width of expression" refers to the amount of grammar, syntax, vocabulary and phrases available to the user. In the present invention, "accuracy" refers to the extent to which the user can use the grammar / wording correctly. In the present invention, "fluency" refers to how smoothly the user can speak English. In the present invention, "interaction" refers to the extent to which a conversation can be started, ended, maintained, and expressed in an intended direction. In the present invention, "consistency" refers to whether or not a coherent order of speaking and language selection can be made. In the present invention, "phonology" refers to the degree to which pronunciation, intonation, and intonation can be expressed correctly and without misunderstanding the other party.

According to the present invention, not only pronunciation but also vocabulary and expression width, grammar / word usage accuracy, fluency, conversation flow control, story composition method and language selection consistency, and intended content are met. It has the effect of being able to measure overall speaking ability such as intonation and intonation.

Further, according to the present invention, the English ability of the user can be determined from various angles by using a plurality of machine learning models at the same time.
Other objects, features and advantages of the invention will become apparent from the following embodiments of the invention with respect to the accompanying drawings.

FIG. 1 is a diagram showing the entire speaking test automatic scoring system according to the present invention. FIG. 2 is a diagram showing a configuration example of a test service server according to the present invention. FIG. 3 is a diagram showing a configuration example of an automatic scoring server according to the present invention. FIG. 4 is a diagram showing an outline of processing between the test service server and the automatic scoring server according to the present invention. FIG. 5 is a diagram showing an outline of the scoring process of the automatic scoring server according to the present invention. FIG. 6 is a diagram showing a processing flow of the entire speaking test automatic scoring system according to the present invention. FIG. 7 is a diagram showing an example of a test service screen according to the present invention.

FIG. 1 is a diagram showing the entire speaking test automatic scoring system according to the present invention. The speaking test automatic scoring system 1 according to the present invention receives voice data and scoring requests from the test service server 10 that receives voice data as a test answer from the user terminal 2 and the test service server 10, and automatically scores the voice data. The automatic scoring server 20 includes an automatic scoring server 20, a scoring data storage unit 30 that stores data necessary for automatic scoring, and a scoring result storage unit 40 that stores the scoring results of automatic scoring. The storage unit 30 writes and reads the scoring data, automatically grades the voice data using the scoring data, and the automatic scoring server 20 transmits the scoring result of the automatic scoring to the test service server 10. The test service server 10 stores the scoring result in the scoring result storage unit 40, and transmits the scoring result to the user terminal 2.

FIG. 2 is a diagram showing a configuration example of a test service server according to the present invention.
The test service server 10 communicates with the user terminal 2 via a network such as the Internet (registered trademark), and provides a test service to the user terminal 2. The test service server 10 receives voice data as a test answer from the user terminal 2 and returns the scoring result to the user terminal 2. The test service server 10 includes a test response receiving unit 11 and a scoring result notification unit 12.

When the test response receiving unit 11 of the test service server 10 receives the voice data as the test answer from the user terminal 2, the test response receiving unit 11 transmits the voice data to the automatic scoring server 20 and also transmits a request for scoring processing.

The scoring result notification unit 12 of the test service server 10 receives the scoring result from the automatic scoring server 20, and stores the scoring result in the scoring result storage unit 40.

FIG. 3 is a diagram showing a configuration example of an automatic scoring server according to the present invention.
The automatic scoring server 20 receives voice data as a test answer from the test service server 10 and returns the scoring result to the test service server 10. The automatic scoring server 20 includes a voice recognition unit 21, a natural language processing unit 22, a scoring data processing unit 23, an item evaluation scoring unit 24, an overall evaluation scoring unit 25, and a scoring result processing unit 26.

The voice recognition unit 21 of the automatic scoring server 20 receives voice input from the test service server 10, recognizes the input voice, and stores the voice recognition result in the scoring data storage unit. Voice recognition is performed by the voice recognition API. The voice recognition API may be an external AIP for voice recognition. The recognized voice data is converted into text data.

The natural language processing unit 22 of the automatic scoring server 20 receives the voice recognition result from the voice recognition unit 21, performs natural language processing on the received voice recognition result, and stores the natural language processing result in the scoring data storage unit 30. Send. As an example of natural language processing, tokenization, Lemmaization, dependency analysis, part-of-speech analysis, and the like may be performed. The natural language processing unit 22 may use an external natural language processing server to generate features as preprocessing.

Upon receiving the scoring processing request from the test service server 10, the scoring data processing unit 23 of the automatic scoring server 20 processes the scoring data according to each of a plurality of predetermined items. The content of processing differs depending on each item. For example, the speed at which the answer starts, the length of the answer, the end time of the answer, and the like may be extracted, calculated, or aggregated from the scoring data. Further, the number of words may be totaled from the scoring data, or the number of unique features may be calculated or totaled from the scoring data. Further, an index that characterizes the sentence, such as WPM or Gillow index, may be extracted or calculated from the scoring data. Further, the frequency of appearance of words and phrases may be counted by using a predetermined dictionary from the voice data converted into text data. Further, the similarity and the distance may be calculated by using the embedded expression obtained by inputting the text data converted from the voice data to the model learned by the present invention or the externally learned model. It should be noted that these processes are examples, and other necessary pre-processes may be performed on the scoring data.

The item evaluation scoring unit 24 of the automatic scoring server 20 automatically scores each of a plurality of predetermined items. When the item evaluation scoring unit 24 receives the request for scoring processing, it executes additional required feature quantity generation according to each skill, and combines them to create one vector. The vector is passed to the trained model arranged in the item evaluation scoring unit 24, and the estimation result output by the trained model is acquired. The item evaluation scoring unit 24 returns the estimation result output by the trained model, the trained model, and the certainty (that is, the prediction probability). Examples of the data input to the item evaluation scoring unit 24 are binary data of voice, text data converted from voice data, data processed in natural language, and the like. Examples of data output from the item evaluation scoring unit 24 are the estimated English level, the name of the model used, the degree of certainty, and the like.

The overall evaluation scoring unit 25 of the automatic scoring server 20 grades the overall evaluation based on the scoring results for each of the plurality of items automatically graded by the item evaluation scoring unit 24. The data input to the overall evaluation scoring unit 25 is the scoring result for a plurality of items in the item evaluation scoring unit 24, that is, the estimation result, and the data output from the overall evaluation scoring unit 25 is the estimated English. Level, name of model used, certainty, etc.

The scoring result processing unit 26 of the automatic scoring server 20 transmits the scoring result to the test service server 10.

FIG. 4 is a diagram showing an outline of processing between the test service server and the automatic scoring server in the present invention.
Hereinafter, the processing of each part will be described along with the processing flow. First, the test response receiving unit 11 of the test service server 10 receives voice data as a test answer from the user terminal 2, stores the voice data in the scoring result storage unit 40, and stores the voice data in the voice of the automatic scoring server 20. It is transmitted to the recognition unit 21. The test response receiving unit 11 further transmits a scoring processing request to the scoring data processing unit 23 of the automatic scoring server 20.

The voice recognition unit 21 of the automatic scoring server 20 receives the voice data transmitted from the test answer reception unit 11, stores the voice recognition result in the scoring data storage unit 30, and stores the voice recognition result in the natural language processing unit 22. Send to. The natural language processing unit 22 that has received the voice recognition result from the voice recognition unit 21 performs natural language processing on the voice recognition result, and stores the result of the natural language processing in the scoring data storage unit 30.

The scoring data processing unit 23, which has received the scoring processing request from the test response receiving unit 11, processes the scoring data according to each of a plurality of predetermined items, and the processing result is used for scoring. It is stored in the data storage unit 30. Further, the scoring data processing unit 23 transmits a request for item scoring to the item evaluation scoring unit 24. Upon receiving the item scoring request from the scoring data processing unit 23, the item evaluation scoring unit 24 automatically scores each of the plurality of predetermined items, and transmits the result of the automatic scoring to the scoring result processing unit 26. The scoring result processing unit 26, which has received the automatic scoring results for a plurality of items from the item evaluation scoring unit 24, stores the automatic scoring results in the scoring data storage unit 30.

Further, the item evaluation scoring unit 24 transmits a request for overall scoring to the overall evaluation scoring unit 25. Upon receiving the request for overall scoring from the item evaluation scoring unit 24, the overall evaluation scoring unit 25 grades the overall evaluation based on the scoring results for each of the plurality of items automatically graded by the item evaluation scoring unit 24, and overall evaluation. The scoring result of is transmitted to the scoring result processing unit 26. The scoring result processing unit 26, which has received the scoring result of the overall evaluation from the overall evaluation scoring unit 25, stores the scoring result of the overall evaluation in the scoring data storage unit 30.

The scoring result processing unit 26 sends the automatic scoring results for a plurality of items received from the item evaluation scoring unit 24 and the scoring results of the overall evaluation received from the overall evaluation scoring unit 25 to the scoring result notification unit 12 of the test service server 10. Send. The scoring result processing unit 26 stores the automatic scoring results for a plurality of items received from the scoring result processing unit 26 and the scoring results of the overall evaluation in the scoring result storage unit 40, and transmits them to the user terminal 2.

FIG. 5 is a diagram showing an outline of the scoring process of the automatic scoring server according to the present invention.
In the present invention, automatic scoring using a skill model is performed. In the present invention, the skill model is a model relating to an item for measuring the quality of utterance, among English proficiency or English conversation ability. As an example, the present invention creates a skill model for six items defined by the CEFR (Common European Framework of Reference for Languages). The six items defined by the CEFR are "range", "accuracy", "fluency", "interaction", "coherence", and And "phonology".

Here, in the present invention, the "width of expression" refers to the amount of grammar, syntax, vocabulary and phrases that can be used by the user. In addition, "accuracy" refers to the extent to which the user can use grammar and wording correctly. In addition, "fluency" refers to how smoothly the user can speak English. In addition, "interaction" refers to the extent to which a conversation can be started, ended, maintained, and expressed in the intended direction. In addition, "consistency" refers to the ability to make a coherent order of speaking and language selection. In addition, "phonology" refers to the degree to which pronunciation, intonation, and intonation can be expressed correctly and without misunderstanding the other party.

In the present invention, the skill model relating to "width of expression", which is one of the items scored by the item evaluation scoring unit 24, is a skill model that measures the amount of grammar, syntax, vocabulary, and phrases that can be used by the user. Is.

In the present invention, the skill model related to "accuracy", which is one of the items scored by the item evaluation scoring unit 24, is a skill model that measures how well the user can use the grammar / wording. Is.

In the present invention, the skill model relating to "fluency", which is one of the items to be scored by the item evaluation scoring unit 24, is a skill model for measuring how smoothly the user can speak English.

In the present invention, the skill model related to "interaction", which is one of the items to be scored by the item evaluation scoring unit 24, starts, ends, maintains, and indicates the degree of intention in the intended direction. It is a skill model that measures whether or not you can do it.

In the present invention, the skill model regarding "consistency", which is one of the items to be scored by the item evaluation scoring unit 24, is a skill model for measuring whether or not a coherent order of speaking and language selection can be performed.

In the present invention, to what extent can the skill model related to "phonology", which is one of the items scored by the item evaluation scoring unit 24, be able to express pronunciation, intonation, and intonation without misunderstanding the other party? It is a skill model to measure.

The item evaluation scoring unit 24 creates a learned skill model for each of these plurality of items. The item evaluation scoring unit 24 automatically scores each item using the learned skill model for each created item, and returns the scoring result to the scoring result processing unit 26.

The overall evaluation scoring unit 25 uses the learned skill models for a plurality of items in the item evaluation scoring unit 24 as inputs, and creates a learned skill model for the overall evaluation. That is, the overall evaluation scoring unit 25 integrates the learned skill models for a plurality of items in the item evaluation scoring unit 24 to create a learned skill model for the overall evaluation. The overall evaluation scoring unit 25 grades the overall evaluation using the created skill model for the overall evaluation, and returns the scoring result to the scoring result processing unit 26.

The scoring result processing unit 26 transmits to the test service server 10 the scoring result for each item transmitted from the item evaluation scoring unit 24 and the scoring result for the overall evaluation transmitted from the overall evaluation scoring unit 25 by the processing program.

FIG. 6 is a diagram showing a processing flow of the entire speaking test automatic scoring system according to the present invention.
In the automatic scoring method using the speaking test automatic scoring system according to the present invention, the automatic scoring server 20 has a step of writing and reading scoring data to the scoring data storage unit 30, and the automatic scoring server 20 has scoring data. The step of automatically scoring the voice data using It is provided with a step of storing and transmitting the scoring result to the user terminal 2.

The automatic scoring method using the speaking test automatic scoring system according to the present invention is realized by a program that causes the speaking test automatic scoring system of the present invention to execute each step of the automatic scoring method of the present invention. The program may be stored in the server and executed in the server or stored in a recording medium.

Specifically, in step S1 in FIG. 6, the speaking test is started by the test service server 10. In step S2, the user answers the speaking test, and the answer is transmitted to the test answer receiving unit 11 of the test service server 10. In step S3, the test answer receiving unit 11 stores the voice data as an answer in the scoring result storage unit 40. In step S4, the voice recognition unit 21 of the automatic scoring server 20 performs voice recognition on the voice data transmitted from the test response reception unit 11. In step S5, the natural language processing unit 22 performs natural language processing on the recognition result transmitted from the voice recognition unit 21. In step S6, it is determined whether or not the test answers have been accepted for all the prepared test questions, and the processes from step S2 to step S5 are repeated until the test answers are accepted for all the questions. In step S7, an automatic scoring request is received from the test response reception unit 11, and in response to the reception of the automatic scoring request, the scoring data processing unit 23 of the automatic scoring server 20 automatically informs the item evaluation scoring unit 24 about each item. Request a score. In step S8, the item evaluation scoring unit 24 automatically scores each item. In step S9, the overall evaluation scoring unit 25 scores the overall evaluation. In step S10, the scoring result processing unit 26 transmits the scoring results received from the item evaluation scoring unit 24 and the overall evaluation scoring unit 25 to the scoring result notification unit 12 of the test service server 10, and the scoring result notification unit 12 is the user terminal. The scoring result is sent to 2.

FIG. 7 is a diagram showing an example of a test service screen according to the present invention.
When the scoring result notification unit 12 of the test service server 10 transmits the scoring result to the user terminal 2, the scoring result may be displayed on the test service screen as shown in FIG. Preferably, as a comprehensive evaluation, the CEFR compliant level corresponding to the scoring result of the overall evaluation is displayed. In addition, the level determined independently by the language learning service provider corresponding to the CEFR compliant level corresponding to the scoring result of the overall evaluation or the level according to other criteria may be displayed. Further, as the evaluation by index, the CEFR compliant level corresponding to the scoring result for each item scored by the item evaluation scoring unit 24 is displayed. In addition, the level determined independently by the language learning service provider corresponding to the CEFR compliant level corresponding to the scoring result for each item or the level according to other standards may be displayed.

The test service screen may display a textual explanation of the evaluation and feedback to the user. In addition, as feedback to the user, learning advice for leveling up may be displayed. Further, the test service screen is not limited to the example of FIG. 7, and the scoring result may be displayed by using a graph, a chart, or a chart.

The test service screen shown in FIG. 7 is an example, and the displayed items and layout are not limited to the example shown in FIG. 7, and may be arbitrary. In FIG. 7, as an example, among the items related to a plurality of skill models, the scoring results regarding "width of expression" and "accuracy" are displayed, but "width of expression", "accuracy", and "accuracy" are preferable. The scoring results are displayed for all six items of "fluency", "interaction", "consistency" and "phonology". In addition, the item to be displayed is any one or any combination of "width of expression", "accuracy", "fluency", "interaction", "consistency" and "phonology". You may.
Although the above description has been made for Examples, it will be apparent to those skilled in the art that the present invention is not limited thereto and various modifications and modifications can be made within the scope of the spirit of the present invention and the appended claims.

1 Speaking test automatic scoring system 2 User terminal 10 Test service server 11 Test response reception unit 12 Scoring result notification unit 21 Speech recognition unit 22 Natural language processing unit 23 Scoring data processing unit 24 Item evaluation scoring unit 25 Overall evaluation scoring unit 26 Scoring Result processing unit 20 Automatic scoring server 30 Scoring data storage unit 40 Scoring result storage unit

Claims (11)

Speaking test automatic scoring system
A test service server that accepts voice data as test answers from user terminals,
An automatic scoring server that receives the voice data and a scoring request from the test service server and automatically scores the voice data.
A scoring data storage unit that stores the data required for automatic scoring, and
A scoring result storage unit for storing the scoring result of the automatic scoring is provided.
The automatic scoring server writes and reads scoring data to the scoring data storage unit, and automatically grades the voice data using the scoring data.
The automatic scoring server transmits the scoring result of the automatic scoring to the test service server.
The test service server is a speaking test automatic scoring system, characterized in that the scoring result is stored in the scoring result storage unit and the scoring result is transmitted to the user terminal. The automatic scoring server includes a voice recognition unit.
The speaking according to claim 1, wherein the voice recognition unit receives voice input from the test service server, recognizes the input voice, and stores the voice recognition result in the scoring data storage unit. Automatic test scoring system. The automatic scoring server is provided with a natural language processing unit.
The natural language processing unit receives the voice recognition result from the voice recognition unit, performs natural language processing on the received voice recognition result, and transmits the natural language processing result to the scoring data storage unit. 2. The speaking test automatic scoring system according to claim 2. The automatic scoring server includes a scoring data processing unit.
When the scoring data processing unit receives a scoring processing request from the test service server, the scoring data processing unit performs processing according to each of a plurality of predetermined items on the scoring data. The speaking test automatic scoring system according to any one of claims 1 to 3. The automatic scoring server is provided with an item evaluation scoring unit, and the item evaluation scoring unit automatically scores each of a plurality of predetermined items, according to any one of claims 1 to 4. Speaking test automatic scoring system described in. The automatic scoring server includes an overall evaluation scoring unit, and the overall evaluation scoring unit grades the overall evaluation based on the scoring results for each of the plurality of items automatically graded by the item evaluation scoring unit. The speaking test automatic scoring system according to any one of claims 5, which is characterized by this. The speaking according to any one of claims 1 to 6, wherein the automatic scoring server includes a scoring result processing unit, and the scoring result processing unit transmits the scoring result to the test service server. Automatic test scoring system. The test service server includes a test response reception unit, and when the test response reception unit receives the voice data as the test response from the user terminal, the test response server transmits the voice data to the automatic scoring server and scores the voice data. The speaking test automatic scoring system according to any one of claims 1 to 7, wherein a processing request is transmitted. The test service server includes a scoring result processing unit, and the scoring result processing unit receives a scoring result from the automatic scoring server and stores the scoring result in the scoring result storage unit. The speaking test automatic scoring system according to any one of Items 1 to 8. It is an automatic scoring method using a speaking test automatic scoring system, and the speaking test automatic scoring system includes a test service server that receives voice data as a test answer from a user terminal, and the voice data and a scoring request from the test service server. An automatic scoring server that receives and automatically scores the voice data, a scoring data storage unit that stores data necessary for the automatic scoring, and a scoring result storage unit that stores the scoring results of the automatic scoring. Prepare,
A step in which the automatic scoring server writes and reads scoring data in the scoring data storage unit, and
A step in which the automatic scoring server automatically scores the voice data using the scoring data.
A step in which the automatic scoring server transmits the scoring result of the automatic scoring to the test service server, and
A step in which the test service server stores the scoring result in the scoring result storage unit and transmits the scoring result to the user terminal.
An automatic scoring method characterized by being equipped with. A program according to any one of claims 1 to 9, wherein the automatic scoring system for a speaking test is made to execute each step of the method according to claim 10.

Images