A spoken language test method and device based on dynamically adjusting test text

Abstract

The present application provides an oral test method and device based on dynamically adjusting the test text, including: after each test of non-last section is completed, obtain the oral score of each candidate up to the current stage, and determine whether the stage score distribution is a normal distribution: when it is not, determine the first target scoring dimension and the second target scoring dimension based on the skewed distribution type and the density distribution type of the stage score distribution; determine the weight level of the first target scoring dimension and the second target scoring dimension through the scoring dimension level determination rule based on the skewed distribution type and the density distribution type respectively; extract the target test text that meets the weight level of the first target scoring dimension and the weight level of the second target scoring dimension from the test text database for the next oral test until the oral test is completed. In this way, according to the stage oral score of the examinee, the test text of the next stage is dynamically selected, so that the final oral score can be normally distributed.

Description

Technical Field

[0001] The present application relates to the technical field of speech assessment, and in particular to a method and device for oral testing based on dynamically adjusting test text. Background Art

[0002] As the oral assessment standard for language learning, with the popularity of oral English exams, more and more schools need to use oral English training systems in daily teaching to assess and score students' oral English pronunciation to help students improve their oral English. In selective tests, students' oral English test scores are also an important part of the English subject scores of selective tests.

[0003] The mainstream English oral training systems on the market usually only use a single dimension of recognizability to evaluate the oral level of the subjects, and each oral evaluation material is recorded by the corresponding oral personnel through standard reading, and the voice similarity between the subject and the standard recording is calculated for scoring. In daily training, due to repeated practice, the scores of a set of oral materials will eventually be close to accurate, but in selective tests that need to reflect differentiation, too concentrated scores are not conducive to the selection of talents. Therefore, in oral tests, how to ensure that the scores of all candidates are not too concentrated, so as to meet the selection test requirements of talent differentiation, is a technical problem that needs to be solved urgently. Summary of the invention

[0004] In view of this, the purpose of this application is to provide an oral test method and device based on dynamically adjusting the test text, dynamically selecting the test text for the next stage according to the examinee's oral scores at that stage, so that the final examinee's oral scores can be normally distributed, thereby providing better auxiliary role in the differentiation and selection of talents.

[0005] The embodiment of the present application provides a spoken language test method based on dynamically adjusting a test text, the spoken language test method comprising:

[0006] After each test section other than the last section is completed, the oral score of each candidate up to the current stage is obtained; the oral score of the stage includes stage sub-scores under multiple scoring dimensions;

[0007] Determine a stage score distribution based on the stage oral scores of all candidates, and determine whether the stage score distribution is a normal distribution:

[0008] When the answer is no, based on the stage oral scores of all candidates, determine the skewness distribution type of the stage score distribution and the density distribution type of the stage score distribution, and determine the skewness distribution type and density distribution type of the stage sub-scores under each scoring dimension;

[0009] The scoring dimension corresponding to the stage sub-score with the same skewed distribution type as the stage score distribution is determined as the first target scoring dimension, and the scoring dimension corresponding to the stage sub-score with the same density distribution type as the stage score distribution is determined as the second target scoring dimension;

[0010] Based on the skewed distribution type and the density distribution type of the stage score distribution, respectively, and by a predetermined scoring dimension grade determination rule, determine the weight grade of the first target scoring dimension and the weight grade of the second target scoring dimension;

[0011] In the test text database, target test texts that meet the weight level of the first target scoring dimension and the weight level of the second target scoring dimension are extracted for the next oral test until the end condition is met and the oral test is stopped; each test text corresponds to multiple scoring dimensions with weight levels and target weight coefficients.

[0012] Optionally, determine the stage sub-scores under multiple scoring dimensions included in the stage oral score by following the steps below:

[0013] After each test except the last one, obtain the first oral score of each candidate after each test;

[0014] For each first oral score, determining a second oral score under multiple scoring dimensions included in the first oral score according to target weight coefficients corresponding to multiple scoring dimensions corresponding to the test text used to obtain the first oral score;

[0015] Add up all the second oral scores under the same scoring dimension to obtain the stage sub-score under each scoring dimension included in the stage oral score.

[0016] Optionally, determine whether the stage score distribution is a normal distribution by the following steps:

[0017] Determine the similarity between the waveform of the stage score distribution and the normal distribution;

[0018] When the waveform similarity is higher than a preset similarity threshold, it is determined that the stage score distribution is a normal distribution.

[0019] Optionally, when the stage scores are normally distributed, the oral test method includes:

[0020] A test text is randomly selected from the test text database for the next oral test, and the oral test is stopped until the end condition is met.

[0021] Optionally, the skewed distribution type includes positive skewed distribution and negative skewed distribution, and the density distribution type includes discrete distribution and concentrated distribution. Based on the stage oral scores of all examinees, the skewed distribution type of the stage score distribution and the density distribution type of the stage score distribution are determined, and the skewed distribution type and density distribution type of the stage sub-score under each scoring dimension are determined, including:

[0022] Based on the oral scores of all candidates, the dispersion of oral scores and the dispersion of sub-scores of each scoring dimension are calculated, and the number of candidates whose oral scores are lower than the third quartile and the number of candidates whose oral scores are higher than the fifth quartile are calculated.

[0023] When the number of examinees whose oral scores at the stage are lower than the third octave is greater than the preset number threshold, the skewed distribution type of the stage score distribution is determined to be a positive skewed distribution; when the number of examinees whose oral scores at the stage are higher than the fifth octave is greater than the preset number threshold, the skewed distribution type of the stage score distribution is determined to be a negative skewed distribution;

[0024] When the discreteness of the oral scores of the stage is greater than a preset first discreteness threshold, the density distribution type of the stage score distribution is determined to be discrete distribution; when the discreteness of the oral scores of the stage is less than a preset second discreteness threshold, the density distribution type of the stage score distribution is determined to be dense distribution;

[0025] For each scoring dimension, when the number of examinees with scores below the third octave under the scoring dimension is greater than the preset number threshold, the skewed distribution type of the stage sub-score under the scoring dimension is determined to be a positive skewed distribution; when the number of examinees with scores above the fifth octave under the scoring dimension is greater than the preset number threshold, the skewed distribution type of the stage sub-score under the scoring dimension is determined to be a negative skewed distribution;

[0026] For each scoring dimension, when the discreteness of the stage sub-scores under the scoring dimension is greater than the preset first discreteness threshold, the density distribution type of the stage sub-scores under the scoring dimension is determined to be discrete distribution; when the discreteness of the stage sub-scores under the scoring dimension is less than the preset second discreteness threshold, the density distribution type of the stage sub-scores under the scoring dimension is determined to be dense distribution.

[0027] Optionally, the scoring dimension level determination rule specifies a mapping relationship between a skewed distribution type and a weight level, and a mapping relationship between a density distribution type and a weight level, and the weight level of the first target scoring dimension and the weight level of the second target scoring dimension are determined by a predetermined scoring dimension level determination rule based on the skewed distribution type and the density distribution type of the stage score distribution, respectively, including:

[0028] When the skewed distribution type of the stage score distribution is positively skewed distribution or negatively skewed distribution, determining the weight level of the first target scoring dimension to be a low level;

[0029] When the density distribution type of the stage score distribution is discrete distribution or concentrated distribution, the weight level of the second target scoring dimension is determined to be a high level.

[0030] Optionally, the weight levels and target weight coefficients of the multiple scoring dimensions corresponding to each test text are determined by the following steps:

[0031] Acquire multiple test texts; wherein each test text corresponds to multiple scoring dimensions, and each scoring dimension is pre-set with a corresponding initial weight coefficient;

[0032] For each test text, extract scoring features of the test text and determine scoring parameters of multiple scoring features included in the test text;

[0033] According to a predetermined weight coefficient adjustment rule, based on the scoring parameters of the various scoring features included in the test text, the initial weight coefficients of the corresponding scoring dimensions are adjusted to determine the target weight coefficient of each scoring dimension of the test text; the weight coefficient adjustment rule specifies the mapping relationship between the scoring parameters of the scoring features and the weight coefficient adjustment amount;

[0034] According to a predetermined weight level rule, based on the target weight coefficient of each scoring dimension corresponding to the test text, the weight level of each scoring dimension corresponding to the test text is determined; the weight level rule stipulates the mapping relationship between the weight coefficient and the weight level, and the weight level includes high, medium and low.

[0035] Optionally, after stopping the oral test, the oral test method further includes:

[0036] Get the total oral test score of each candidate;

[0037] Under the condition that the ranking remains unchanged, the total oral test scores of all candidates are slightly adjusted so that the score distribution of the total oral test scores of all candidates after adjustment is normal distribution;

[0038] The fine-tuned score will be used as the candidate's final oral test score.

[0039] The embodiment of the present application provides a spoken language test device based on dynamically adjusting a test text, the spoken language test device comprising:

[0040] An acquisition module is used to acquire the oral scores of each candidate up to the current stage after each test except the last stage is completed; the oral scores of the stage include stage sub-scores under multiple scoring dimensions;

[0041] The normal distribution determination module is used to determine the stage score distribution based on the stage oral scores of all candidates, and determine whether the stage score distribution is a normal distribution:

[0042] A distribution type determination module is used to determine the skewed distribution type and the density distribution type of the stage score distribution based on the stage oral scores of all examinees when the stage score distribution is not a normal distribution, and to determine the skewed distribution type and the density distribution type of the stage sub-score under each scoring dimension;

[0043] A target dimension determination module, used to determine the scoring dimension corresponding to the stage sub-score of the same skewed distribution type as the stage score distribution as the first target scoring dimension, and determine the scoring dimension corresponding to the stage sub-score of the same density distribution type as the stage score distribution as the second target scoring dimension;

[0044] A weight level determination module, for determining the weight level of the first target scoring dimension and the weight level of the second target scoring dimension based on the skewed distribution type and the density distribution type of the stage score distribution, respectively, by using a predetermined scoring dimension level determination rule;

[0045] The first selection module is used to extract target test texts that meet the weight level of the first target scoring dimension and the weight level of the second target scoring dimension from the test text database for the next oral test until the end condition is met and the oral test is stopped; each test text corresponds to multiple scoring dimensions with weight levels and target weight coefficients.

[0046] Optionally, the oral test device further comprises a sub-score determination module, wherein the sub-score determination module is used to:

[0047] After each test except the last one, obtain the first oral score of each candidate after each test;

[0048] For each first oral score, determining a second oral score under the multiple scoring dimensions included in the first oral score according to the target weight coefficients of the multiple scoring dimensions corresponding to the test text used to obtain the first oral score;

[0049] Add up all the second oral scores under the same scoring dimension to obtain the stage sub-score under each scoring dimension included in the stage oral score.

[0050] Optionally, when the normal distribution determination module is used to determine whether the stage score distribution is a normal distribution, the normal distribution determination module is used to:

[0051] Determine the similarity between the waveform of the stage score distribution and the normal distribution;

[0052] When the waveform similarity is higher than a preset similarity threshold, it is determined that the stage score distribution is a normal distribution.

[0053] Optionally, the oral testing device further includes a second selection module, wherein the second selection module is used to:

[0054] When the stage score distribution is a normal distribution, a test text is randomly selected from the test text database for the next oral test, and the oral test is stopped until the end condition is met.

[0055] Optionally, the skewed distribution type includes positive skewed distribution and negative skewed distribution, and the density distribution type includes discrete distribution and concentrated distribution. When the distribution type determination module is used to determine the skewed distribution type and density distribution type of the stage score distribution based on the stage oral scores of all examinees, and to determine the skewed distribution type and density distribution type of the stage sub-scores under each scoring dimension, the distribution type determination module is used to:

[0056] Based on the oral scores of all candidates, the dispersion of oral scores and the dispersion of sub-scores of each scoring dimension are calculated, and the number of candidates whose oral scores are lower than the third quartile and the number of candidates whose oral scores are higher than the fifth quartile are calculated.

[0057] When the number of examinees whose oral scores at the stage are lower than the third octave is greater than the preset number threshold, the skewed distribution type of the stage score distribution is determined to be a positive skewed distribution; when the number of examinees whose oral scores at the stage are higher than the fifth octave is greater than the preset number threshold, the skewed distribution type of the stage score distribution is determined to be a negative skewed distribution;

[0058] When the discreteness of the oral scores of the stage is greater than a preset first discreteness threshold, the density distribution type of the stage score distribution is determined to be discrete distribution; when the discreteness of the oral scores of the stage is less than a preset second discreteness threshold, the density distribution type of the stage score distribution is determined to be dense distribution;

[0059] For each scoring dimension, when the number of examinees with scores below the third octave under the scoring dimension is greater than the preset number threshold, the skewed distribution type of the stage sub-score under the scoring dimension is determined to be a positive skewed distribution; when the number of examinees with scores above the fifth octave under the scoring dimension is greater than the preset number threshold, the skewed distribution type of the stage sub-score under the scoring dimension is determined to be a negative skewed distribution;

[0060] For each scoring dimension, when the discreteness of the stage sub-scores under the scoring dimension is greater than the preset first discreteness threshold, the density distribution type of the stage sub-scores under the scoring dimension is determined to be discrete distribution; when the discreteness of the stage sub-scores under the scoring dimension is less than the preset second discreteness threshold, the density distribution type of the stage sub-scores under the scoring dimension is determined to be dense distribution.

[0061] Optionally, when the weight level determination module is used to determine the weight level of the first target scoring dimension and the weight level of the second target scoring dimension based on the skewed distribution type and the density distribution type of the score distribution respectively, by a predetermined scoring dimension level determination rule, the weight level determination module is used to:

[0062] When the skewed distribution type of the stage score distribution is positively skewed distribution or negatively skewed distribution, determining the weight level of the first target scoring dimension to be a low level;

[0063] When the density distribution type of the stage score distribution is discrete distribution or concentrated distribution, the weight level of the second target scoring dimension is determined to be a high level.

[0064] Optionally, the oral language testing device further comprises a text information determination module, wherein the text information determination module is used to:

[0065] Acquire multiple test texts; wherein each test text corresponds to multiple scoring dimensions, and each scoring dimension is pre-set with a corresponding initial weight coefficient;

[0066] For each test text, extract scoring features of the test text and determine scoring parameters of multiple scoring features included in the test text;

[0067] According to a predetermined weight coefficient adjustment rule, based on the scoring parameters of the various scoring features included in the test text, the initial weight coefficients of the corresponding scoring dimensions are adjusted to determine the target weight coefficient of each scoring dimension of the test text; the weight coefficient adjustment rule specifies the mapping relationship between the scoring parameters of the scoring features and the weight coefficient adjustment amount;

[0068] According to a predetermined weight level rule, based on the target weight coefficient of each scoring dimension corresponding to the test text, the weight level of each scoring dimension corresponding to the test text is determined; the weight level rule stipulates the mapping relationship between the weight coefficient and the weight level, and the weight level includes high, medium and low.

[0069] Optionally, the oral language testing device further comprises a fine-tuning module, wherein the fine-tuning module is used to:

[0070] Get the total oral test score of each candidate;

[0071] Under the condition that the ranking remains unchanged, the total oral test scores of all candidates are slightly adjusted so that the score distribution of the total oral test scores of all candidates after adjustment is normal distribution;

[0072] The fine-tuned score will be used as the candidate's final oral test score.

[0073] An embodiment of the present application also provides an electronic device, comprising: a processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor and the memory communicate via the bus, and when the machine-readable instructions are executed by the processor, the steps of the above-mentioned oral test method are performed.

[0074] An embodiment of the present application further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned oral test method are executed.

[0075] An oral test method and device based on dynamically adjusting test text provided by an embodiment of the present application include: after each test section other than the last section is completed, obtaining the oral score of each candidate up to the current stage; the oral score of the stage includes stage sub-scores under multiple scoring dimensions; determining the stage score distribution based on the stage oral scores of all candidates, and determining whether the stage score distribution is a normal distribution: when it is not, determining the skewed distribution type of the stage score distribution and the density distribution type of the stage score distribution based on the stage oral scores of all candidates, and determining the skewed distribution type and density distribution type of the stage sub-score under each scoring dimension; and dividing the stage sub-scores with the same skewed distribution type as the stage score distribution into a normal distribution; The scoring dimension corresponding to the stage score is determined as the first target scoring dimension, and the scoring dimension corresponding to the stage sub-score with the same density distribution type as the stage score distribution is determined as the second target scoring dimension; based on the skewed distribution type and the density distribution type of the stage score distribution, the weight level of the first target scoring dimension and the weight level of the second target scoring dimension are determined by a predetermined scoring dimension level determination rule; in the test text database, the target test text that meets the weight level of the first target scoring dimension and the weight level of the second target scoring dimension is extracted to conduct the next oral test until the end condition is met and the oral test is stopped; each test text corresponds to multiple scoring dimensions with weight levels and target weight coefficients.

[0076] In this way, this application dynamically selects the test text for the next stage according to the examinee's oral scores at that stage, so that the final examinee's oral scores can be normally distributed, thereby providing better auxiliary role in the differentiation and selection of talents.

[0077] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, preferred embodiments are specifically cited below and described in detail with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0079] Figure 1 A flowchart of a spoken language test method based on dynamically adjusting test text provided in an embodiment of the present application;

[0080] Figure 2 One of the structural schematic diagrams of a spoken language testing device based on dynamically adjusting test text provided in an embodiment of the present application;

[0081] Figure 3 A second structural diagram of a spoken language testing device based on dynamically adjusting test text provided in an embodiment of the present application;

[0082] Figure 4 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0083] To make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. The components of the embodiments of the present application usually described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, each other embodiment obtained by those skilled in the art without making creative work belongs to the scope of protection of the present application.

[0084] Research has found that the mainstream English oral training systems on the market usually only use a single dimension of recognizability to evaluate the oral proficiency of the subjects, and each oral evaluation material is recorded by the corresponding oral personnel through standard reading, and the voice similarity between the subject and the standard recording is calculated for scoring. In daily training, due to repeated practice, the scores of a set of oral materials will eventually be close to accurate, but in selective tests that need to reflect differentiation, too concentrated scores are not conducive to talent selection. Therefore, in oral tests, how to ensure that the scores of all candidates are not too concentrated, so as to meet the selection test requirements of talent differentiation, is a technical problem that needs to be solved urgently.

[0085] Based on this, the embodiment of the present application provides an oral test method based on dynamically adjusting the test text. According to the examinee's oral scores at that stage, the test text for the next stage is dynamically selected, so that the final oral scores of the examinee can be normally distributed, thereby providing better auxiliary role in the differentiation and selection of talents.

[0086] See also Figure 1 , Figure 1 The flowchart of a spoken language test method based on dynamically adjusting the test text provided in the embodiment of the present application is as follows. Figure 1 As shown in , the oral test method based on dynamically adjusting the test text provided in the embodiment of the present application includes:

[0087] S101. After each test section other than the last section is completed, the oral score of each candidate up to the current stage is obtained; the oral score of the stage includes stage sub-scores under multiple scoring dimensions.

[0088] Here, an oral test process includes multiple test sections, each of which uses a test text for oral test, and after each test section is completed, the test score of each candidate in the target group is counted; the oral score of the stage is the sum of all the scores from the beginning of the first test section to the end of this test section.

[0089] For example, when the test ends at the end of the 5th section, the oral score obtained is the sum of the scores of the previous 5 sections.

[0090] The scoring dimensions are used to examine the user's oral expression ability from different aspects and to score the examinee's oral performance from different aspects. The scoring dimensions include four scoring dimensions: recognizability, fluency, tone and intonation. Among them, the stage oral performance is the sum of the four stage sub-performances under the four scoring dimensions.

[0091] In one implementation mode provided in the present application, the stage sub-scores under multiple scoring dimensions included in the stage oral score are determined by the following steps: after each test section other than the last section is completed, the first oral score of each candidate after each test section is obtained; for each first oral score, the second oral score under the multiple scoring dimensions included in the first oral score is determined according to the target weight coefficients corresponding to the multiple scoring dimensions corresponding to the test text used to obtain the first oral score; all the second oral scores belonging to the same scoring dimension are added together to obtain the stage sub-score under each scoring dimension included in the stage oral score.

[0092] Here, the test texts used in each test are different, and the target weight coefficients of the multiple scoring dimensions corresponding to each test text are also different.

[0093] When determining the second oral score under the multiple scoring dimensions included in the first oral score, the first oral score can be multiplied by the target weight coefficients of the multiple scoring dimensions corresponding to the corresponding test text to obtain multiple scores with the same number of scoring dimensions, which are the second oral scores.

[0094] When determining the first oral score of the examinee after completing each test, specifically: collect the examinee's reading voice of the test text used in the test, input the reading voice into a pre-trained scoring model, and then obtain the examinee's first oral score.

[0095] S102. Determine a stage score distribution based on the stage oral scores of all examinees, and determine whether the stage score distribution is a normal distribution.

[0096] In one implementation provided in the present application, whether the stage score distribution is a normal distribution is determined by the following steps: judging the waveform similarity between the stage score distribution and the normal distribution; when the waveform similarity is higher than a preset similarity threshold, determining that the stage score distribution is a normal distribution.

[0097] It should be noted that the reason for determining whether the distribution of stage scores is a normal distribution is that for selective tests that need to reflect the degree of differentiation, a primary condition is that the score distribution of all candidates after the test is completed is a normal distribution, so as to select talents.

[0098] S103: When the answer is no, based on the stage oral scores of all examinees, determine the skewed distribution type and the density distribution type of the stage score distribution, and determine the skewed distribution type and the density distribution type of the stage sub-scores under each scoring dimension.

[0099] In this step, when it is determined that the stage score distribution is not a normal distribution, it means that the test texts selected in the first few sections of the oral test do not meet the requirements of the final test results, and the test texts need to be adjusted. Here, first, it is necessary to determine the skewed distribution type and density distribution type of the stage score distribution and the skewed distribution type and density distribution type of the stage sub-score under each scoring dimension.

[0100] In one embodiment provided in the present application, the skewed distribution type includes positive skewed distribution and negative skewed distribution, the density distribution type includes discrete distribution and concentrated distribution, and the step of determining the skewed distribution type of the stage score distribution and the density distribution type of the stage score distribution based on the stage oral scores of all examinees, and determining the skewed distribution type and the density distribution type of the stage sub-score under each scoring dimension, includes: based on the stage oral scores of all examinees, counting the discreteness of the stage oral scores and the discreteness of the stage sub-scores under each scoring dimension, counting the number of examinees whose stage oral scores are lower than the third octant and the number of examinees whose stage oral scores are higher than the fifth octant; when the number of examinees whose stage oral scores are lower than the third octant is greater than a preset number threshold, determining the skewed distribution type of the stage score distribution to be positively skewed distribution, and when the number of examinees whose stage oral scores are higher than the fifth octant is greater than the preset number threshold, determining the skewed distribution type of the stage score distribution to be negatively skewed distribution; when the stage oral scores are lower than the fifth octant, determining the skewed distribution type of the stage score distribution to be negatively skewed distribution. When the discreteness of the oral score of the stage is greater than the preset first discreteness threshold, the density distribution type of the stage score distribution is determined to be discrete distribution; when the discreteness of the oral score of the stage is less than the preset second discreteness threshold, the density distribution type of the stage score distribution is determined to be dense distribution; for each scoring dimension, when the number of examinees with scores lower than the third octave under the scoring dimension is greater than the preset number threshold, the skewed distribution type of the stage sub-score under the scoring dimension is determined to be positively skewed distribution; when the number of examinees with scores higher than the fifth octave under the scoring dimension is greater than the preset number threshold, the skewed distribution type of the stage sub-score under the scoring dimension is determined to be negatively skewed distribution; for each scoring dimension, when the discreteness of the stage sub-score under the scoring dimension is greater than the preset first discreteness threshold, the density distribution type of the stage sub-score under the scoring dimension is determined to be discrete distribution; when the discreteness of the stage sub-score under the scoring dimension is less than the preset second discreteness threshold, the density distribution type of the stage sub-score under the scoring dimension is determined to be dense distribution.

[0101] The third octave is explained by the following example. Assuming that the full score of the oral English test is 100, the octaves are 12.5, 25, 37.5, 50, 62.5, 75, 87.5, and 100, and the third octave is 37.5. Similarly, the fifth octave is 62.5.

[0102] In addition, the skewed distribution type of the stage score distribution and the skewed distribution type of the stage sub-score can also be determined by the following steps: when the number of examinees whose stage oral scores are lower than the first reference threshold score is greater than the preset number threshold, the skewed distribution type of the stage score distribution is determined to be a positive skewed distribution; when the number of examinees whose stage oral scores are higher than the second reference threshold score is greater than the preset number threshold, the skewed distribution type of the stage score distribution is determined to be a negative skewed distribution; for each scoring dimension, when the number of examinees whose scores under the scoring dimension are lower than the first reference threshold score is greater than the preset number threshold, the skewed distribution type of the stage sub-score under the scoring dimension is determined to be a positive skewed distribution; when the number of examinees whose scores are higher than the second reference threshold score under the scoring dimension is greater than the preset number threshold, the skewed distribution type of the stage sub-score under the scoring dimension is determined to be a negative skewed distribution.

[0103] Among them, when the discreteness of the oral score of the stage is between the first discreteness threshold and the second discreteness threshold, the density distribution type of the oral score of the stage is determined to be the standard distribution; when the discreteness of the stage sub-score is between the first discreteness threshold and the second discreteness threshold, the density distribution type of the stage sub-score is determined to be the standard distribution.

[0104] Among them, when the skewed distribution type of the stage score distribution is positively skewed distribution, it means that a large number of users have obtained low scores; when the skewed distribution type of the stage score distribution is negatively skewed distribution, it means that a large number of users have obtained high scores; when the density distribution type of the stage score distribution is discrete distribution, it means that there are a lot of people with high and low scores in this group, but few people with medium scores; when the density distribution type of the stage score distribution is dense distribution, it means that this group generally gets the same scores.

[0105] In another embodiment provided in the present application, when the stage score distribution is a normal distribution, the oral test method includes: randomly selecting a test text from the test text database for the next oral test, until the end condition is reached, and stopping the oral test.

[0106] Here, when the stage score distribution is a normal distribution, it means that there is no need to make a targeted selection for the test, and a test text can be randomly selected from the test text database as the next test question. The test text database stores a plurality of different test texts.

[0107] S104. Determine the scoring dimension corresponding to the stage sub-score with the same skewed distribution type as the stage score distribution as the first target scoring dimension, and determine the scoring dimension corresponding to the stage sub-score with the same density distribution type as the stage score distribution as the second target scoring dimension.

[0108] Here, after determining the skewed distribution type and density distribution type of the stage score distribution, and the skewed distribution type and density distribution type of the stage sub-scores under each scoring dimension, it is necessary to determine the weak dimensions or strong dimensions among multiple scoring dimensions, so as to select the test text in a targeted manner in the next test.

[0109] Among them, the first target scoring dimension is determined by the skewed distribution type of the stage score distribution, and the second target scoring dimension is determined by the density distribution type of the stage score distribution.

[0110] S105. Based on the skewed distribution type and the density distribution type of the stage score distribution, respectively, and using a predetermined scoring dimension level determination rule, determine the weight levels of the first target scoring dimension and the second target scoring dimension.

[0111] In one embodiment provided in the present application, the scoring dimension level determination rule stipulates a mapping relationship between the skewed distribution type and the weight level, and a mapping relationship between the density distribution type and the weight level, and the weight level of the first target scoring dimension and the weight level of the second target scoring dimension are determined based on the skewed distribution type and the density distribution type of the stage score distribution, respectively, through a predetermined scoring dimension level determination rule, including: when the skewed distribution type of the stage score distribution is a positively skewed distribution or a negatively skewed distribution, determining that the weight level of the first target scoring dimension is a low level; when the density distribution type of the stage score distribution is a discrete distribution or a concentrated distribution, determining that the weight level of the second target scoring dimension is a high level.

[0112] Here, a low weight level means that the target weight coefficient value of the scoring dimension is smaller than the target weight coefficient values ​​of other scoring dimensions; a high weight level means that the target weight coefficient value of the scoring dimension is larger than the target weight coefficient values ​​of other scoring dimensions.

[0113] When the skewed distribution type of the score distribution in the stage is positive skewed distribution, the first target scoring dimension determined is the weak dimension of the examinee group, so the next test needs to select a test text with a low weight level of the weak dimension as a test question, even if most people get low scores under this scoring dimension, it will not have much impact on the result, so that the total score is close to normal distribution;

[0114] When the skewed distribution type of the score distribution in the stage is negatively skewed, the first target scoring dimension determined is the dimension that the examinee group is good at, so the next test needs to select a test text with a low weight level for the good dimension as the test question, so that even if most people get high scores in this scoring dimension, it will not have much impact on the result, so that the total score approaches a normal distribution;

[0115] When the density distribution type of the stage score distribution is discrete distribution, the second target scoring dimension determined is a dimension in the user's general score set, so the next test needs to select a test text with a high weight level of the scoring dimension in the user's general score set as a test question;

[0116] When the density distribution type of the score distribution in the stage is dense distribution, the second target scoring dimension determined is a dimension in which user scores are more dispersed. Therefore, the next test needs to select a test text with a high weight level in the scoring dimension in which user scores are more dispersed as a test question.

[0117] S106. From the test text database, extract the target test text that meets the weight level of the first target scoring dimension and the weight level of the second target scoring dimension to conduct the next oral test until the end condition is met and the oral test is stopped; each test text corresponds to multiple scoring dimensions with weight levels and target weight coefficients.

[0118] Here, after determining the weight level of each scoring dimension corresponding to the target test text required for the next test, a test text that meets the requirements is selected from the test text database for the next test. When multiple test texts that meet the requirements are found, one test text can be selected from the multiple test texts that meet the requirements as the target test text.

[0119] In one embodiment provided in the present application, the weight levels and target weight coefficients of the multiple scoring dimensions corresponding to each test text are determined by the following steps: obtaining multiple test texts; wherein each test text corresponds to multiple scoring dimensions, and each scoring dimension is pre-set with a corresponding initial weight coefficient; for each test text, extracting scoring features of the test text, and determining the scoring parameters of the multiple scoring features included in the test text; according to a predetermined weight coefficient adjustment rule, adjusting the initial weight coefficients of the corresponding scoring dimensions based on the scoring parameters of the multiple scoring features included in the test text, and determining the target weight coefficient of each scoring dimension corresponding to the test text; the weight coefficient adjustment rule specifies the mapping relationship between the scoring parameters of the scoring features and the weight coefficient adjustment amount; according to a predetermined weight level rule, determining the weight level of each scoring dimension corresponding to the test text based on the target weight coefficient of each scoring dimension corresponding to the test text; the weight level rule specifies the mapping relationship between the weight coefficient and the weight level, and the weight level includes high, medium and low.

[0120] Here, the weight levels and target weight coefficients of the multiple scoring dimensions corresponding to each test text in the test text database are determined before the oral test. The scoring features include the number of text events, the correlation between the topic vector and the word vector, the number of word vectors, and the number of lexical stressed syllables.

[0121] The number of text events included in each test text is extracted by a text event extraction model, and the test text is converted into a text vector by a word vector extraction model to obtain a word vector corresponding to the test text, and the obtained word vector is extracted and analyzed to determine the correlation between the topic vector and the word vector, the number of word vectors, and the number of lexical stress syllables included in the test text. The word vector extraction model preferably selects the word2vec model.

[0122] The method of adjusting the initial weight coefficients of the corresponding scoring dimensions according to the predetermined weight coefficient adjustment rules and based on the scoring parameters of the various scoring features included in the test text, and determining the target weight coefficient of each scoring dimension of the test text, includes:

[0123] For each test text, the initial weight coefficient of recognizability is adjusted according to the number of text events included in the test text, and the target weight coefficient of recognizability is determined according to the specific number of text events;

[0124] Adjusting the initial weight coefficient of the tone according to the correlation between the topic vector and the word vector included in the test text, and determining the target weight coefficient of the tone according to the correlation value between the specific topic vector and the word vector;

[0125] Adjusting the initial weight coefficient of fluency according to the number of word vectors included in the test text, and determining the target weight coefficient of fluency according to the specific number of word vectors;

[0126] The initial weight coefficient of intonation is adjusted according to the number of lexical stressed syllables included in the test text, and the target weight coefficient of intonation is determined according to the specific number of lexical stressed syllables.

[0127] It should be noted that if there are many text events in the test text, it means that the test text has a large amount of information, and the user's voice needs to have a high degree of recognizability to avoid missing information; if there are sentences or paragraphs in the test text that are highly relevant to the topic (the correlation value between the topic vector and the word vector is high), the user's voice needs to have a corresponding tone to reflect the core content of the text; if there are too many word vectors in the test text, the user's voice needs to remain fluent and coherent; if the test text has many stressed syllables, the user needs to read it out, especially in English voice, different stressed syllables may represent different words, and the evaluation of stressed syllables is of great significance to the correct communication of voice information.

[0128] For example, taking a test text as an example, the following steps are used to explain the weight system adjustment process: Assume that the initial weight coefficients of the test text corresponding to multiple scoring dimensions are 35% for recognizability, 30% for fluency, 20% for tone, and 15% for pitch;

[0129] The recognizability weight coefficient adjustment rule can be set as follows: when the number of text events included in the test text is greater than 7, the scoring weight is adjusted to make the recognizability +5%; when the number of text events included in the test text is 7 ≥ the number of text events > 6, the scoring weight is adjusted to make the recognizability +3%; when the number of text events included in the test text is 6 ≥ the number of text events > 4, the scoring weight is adjusted to make the recognizability +1%; when the number of text events included in the test text is 4 ≥ the number of text events > 2, the scoring weight is not adjusted; when the number of text events included in the test text is 2 ≥ the number of text events, the scoring weight is adjusted to make the recognizability -1%;

[0130] The fluency weight coefficient adjustment rule can be set as follows: when the number of word vectors in the test text is greater than 100, the score weight is adjusted to make the fluency +5%; when the number of word vectors in the test text is 100 ≥ the number of word vectors > 80, the score weight is adjusted to make the fluency +3%; when the number of word vectors in the test text is 80 ≥ the number of word vectors > 50, the score weight is adjusted to make the fluency +1%; when the number of word vectors in the test text is 50 ≥ the number of word vectors > 20, the score weight is not adjusted; when the number of word vectors in the test text is 20 ≥ the number of word vectors, the score weight is adjusted to make the fluency -1%;

[0131] The tone weight coefficient adjustment rule can be set as follows: when the word vector whose correlation between the topic vector and the word vector in the test text is greater than 0.4 is set as the relevant word vector, when the number of relevant word vectors in the test text is greater than 7, the score weight is adjusted to make the tone +5%; when 7≥the number of relevant word vectors in the test text is greater than 6, the score weight is adjusted to make the tone +3%; when 6≥the number of relevant word vectors in the test text is greater than 4, the score weight is adjusted to make the tone +1%; when 4≥the number of relevant word vectors in the test text is greater than 2, the score weight is not adjusted; when 2≥the number of relevant word vectors in the paragraph of the test text, the score weight is adjusted to make the tone -1%;

[0132] The adjustment rule of the intonation weight coefficient can be set as follows: when the number of stressed syllables in the test text is greater than 7, the scoring weight is adjusted to make the intonation +5%; when the number of 7≥stressed syllables in the test text is greater than 6, the scoring weight is adjusted to make the intonation +3%; when the number of 6≥stressed syllables in the test text is greater than 4, the scoring weight is adjusted to make the intonation +1%; when the number of 4≥stressed syllables in the test text is greater than 2, the scoring weight is not adjusted; when the number of 2≥stressed syllables in the test text is adjusted to make the intonation -1%;

[0133] Finally, the weight coefficient of a single scoring dimension is divided by the sum of the weight coefficients of all scoring dimensions to obtain the target weight coefficient of a single dimension.

[0134] Here, the weight level rule may be to set the weight level of the scoring dimension that has been deducted during the weight adjustment process to a low level, set the weight level of the scoring dimension that has been added during the weight adjustment process to a high level, and set the weight level of the rest to medium. Alternatively, multiple scoring dimensions are sorted from large to small according to the target weight coefficient, and the weight level of the first scoring dimension is set to high, the weight level of the last scoring dimension is set to low, and the rest are set to medium.

[0135] In another embodiment provided in the present application, after stopping the oral test, the oral test method further includes: obtaining the total oral test score of each candidate; under the condition that the ranking remains unchanged, fine-tuning the total oral test scores of all candidates so that the score distribution of the adjusted total oral test scores of all candidates is a normal distribution; and using the fine-tuned scores as the final total oral test scores of the candidates.

[0136] Here, through fine-tuning, the final oral scores of all candidates are made normally distributed, thus meeting the requirements of differentiated selection.

[0137] The fine-tuning process is illustrated through the following examples: for example, on the basis of setting the recognizability weight to be always the highest, if there is score concentration, then the weight corresponding to the dimension with the most concentrated scores is reduced to make the total score more flat; if there is score dispersion, then the weight corresponding to the dimension with the most concentrated scores is increased to make the total score more concentrated; if a dimension has more high scores or more low scores, then the weight corresponding to the dimension is reduced to reduce its influence so that the total score has the characteristics of a normal distribution with fewer high and low scores and more middle scores. One thing that needs to be noted is to ensure that the ranking of candidates remains unchanged.

[0138] An oral test method based on dynamically adjusting a test text provided in an embodiment of the present application includes: after each test section other than the last section is completed, obtaining the oral score of each candidate up to the current stage; the oral score of the stage includes stage sub-scores under multiple scoring dimensions; determining the stage score distribution based on the stage oral scores of all candidates, and determining whether the stage score distribution is a normal distribution: when it is not, determining the skewed distribution type of the stage score distribution and the density distribution type of the stage score distribution based on the stage oral scores of all candidates, and determining the skewed distribution type and density distribution type of the stage sub-score under each scoring dimension; comparing the stage sub-scores with the same skewed distribution type as the stage score distribution The scoring dimension corresponding to the stage sub-score is determined as the first target scoring dimension, and the scoring dimension corresponding to the stage sub-score with the same density distribution type as the stage score distribution is determined as the second target scoring dimension; based on the skewed distribution type and the density distribution type of the stage score distribution, the weight level of the first target scoring dimension and the weight level of the second target scoring dimension are determined by a predetermined scoring dimension level determination rule; in the test text database, the target test text that meets the weight level of the first target scoring dimension and the weight level of the second target scoring dimension is extracted to conduct the next oral test until the end condition is met and the oral test is stopped; each test text corresponds to a plurality of scoring dimensions with weight levels and target weight coefficients.

[0139] In this way, this application dynamically selects the test text for the next stage according to the examinee's oral scores at that stage, so that the final examinee's oral scores can be normally distributed, thereby providing better auxiliary role in the differentiation and selection of talents.

[0140] See also Figure 2 , Figure 3 , Figure 2 This is one of the structural schematic diagrams of a spoken language testing device based on dynamically adjusting the test text provided in an embodiment of the present application, Figure 3 The second structural diagram of a spoken language test device based on dynamically adjusting the test text provided in the embodiment of the present application. Figure 2 As shown in , the oral testing device 200 comprises:

[0141] The acquisition module 201 is used to acquire the oral English score of each examinee up to the current stage after each test of the non-last stage is completed; the oral English score of the stage includes the stage sub-scores under multiple scoring dimensions;

[0142] The normal distribution determination module 202 is used to determine the stage score distribution based on the stage oral scores of all examinees, and determine whether the stage score distribution is a normal distribution:

[0143] The distribution type determination module 203 is used to determine the skewed distribution type and the density distribution type of the stage score distribution based on the stage oral scores of all examinees when the stage score distribution is not a normal distribution, and determine the skewed distribution type and the density distribution type of the stage sub-score under each scoring dimension;

[0144] A target dimension determination module 204 is used to determine the scoring dimension corresponding to the stage sub-score of the same skewed distribution type as the stage score distribution as the first target scoring dimension, and determine the scoring dimension corresponding to the stage sub-score of the same density distribution type as the stage score distribution as the second target scoring dimension;

[0145] A weight level determination module 205 is used to determine the weight level of the first target scoring dimension and the weight level of the second target scoring dimension based on the skewed distribution type and the density distribution type of the stage score distribution respectively, according to a predetermined scoring dimension level determination rule;

[0146] The first selection module 206 is used to extract target test texts that meet the weight level of the first target scoring dimension and the weight level of the second target scoring dimension from the test text database to conduct the next oral test until the end condition is met and the oral test is stopped; each test text corresponds to multiple scoring dimensions with weight levels and target weight coefficients.

[0147] Optional, such as Figure 3 As shown, the oral test device 200 further includes a sub-score determination module 207, and the sub-score determination module 207 is used to:

[0148] After each test except the last one, obtain the first oral score of each candidate after each test;

[0149] For each first oral score, determining a second oral score under the multiple scoring dimensions included in the first oral score according to the target weight coefficients of the multiple scoring dimensions corresponding to the test text used to obtain the first oral score;

[0150] Add up all the second oral scores under the same scoring dimension to obtain the stage sub-score under each scoring dimension included in the stage oral score.

[0151] Optionally, when the normal distribution determination module 202 is used to determine whether the stage score distribution is a normal distribution, the normal distribution determination module 202 is used to:

[0152] Determine the similarity between the waveform of the stage score distribution and the normal distribution;

[0153] When the waveform similarity is higher than a preset similarity threshold, it is determined that the stage score distribution is a normal distribution.

[0154] Optionally, the spoken language testing device 200 further includes a second selection module 208, and the second selection module 208 is used to:

[0155] When the stage score distribution is a normal distribution, a test text is randomly selected from the test text database for the next oral test, and the oral test is stopped until the end condition is met.

[0156] Optionally, the skewed distribution type includes positive skewed distribution and negative skewed distribution, and the density distribution type includes discrete distribution and concentrated distribution. When the distribution type determination module 203 is used to determine the skewed distribution type and density distribution type of the stage score distribution based on the stage oral scores of all examinees, and to determine the skewed distribution type and density distribution type of the stage sub-scores under each scoring dimension, the distribution type determination module 203 is used to:

[0157] Based on the oral scores of all candidates, the dispersion of oral scores and the dispersion of sub-scores of each scoring dimension are calculated, and the number of candidates whose oral scores are lower than the third quartile and the number of candidates whose oral scores are higher than the fifth quartile are calculated.

[0158] When the number of examinees whose oral scores at the stage are lower than the third octave is greater than the preset number threshold, the skewed distribution type of the stage score distribution is determined to be a positive skewed distribution; when the number of examinees whose oral scores at the stage are higher than the fifth octave is greater than the preset number threshold, the skewed distribution type of the stage score distribution is determined to be a negative skewed distribution;

[0159] When the discreteness of the oral scores of the stage is greater than a preset first discreteness threshold, the density distribution type of the stage score distribution is determined to be discrete distribution; when the discreteness of the oral scores of the stage is less than a preset second discreteness threshold, the density distribution type of the stage score distribution is determined to be dense distribution;

[0160] For each scoring dimension, when the number of examinees with scores below the third octave under the scoring dimension is greater than the preset number threshold, the skewed distribution type of the stage sub-score under the scoring dimension is determined to be a positive skewed distribution; when the number of examinees with scores above the fifth octave under the scoring dimension is greater than the preset number threshold, the skewed distribution type of the stage sub-score under the scoring dimension is determined to be a negative skewed distribution;

[0161] For each scoring dimension, when the discreteness of the stage sub-scores under the scoring dimension is greater than the preset first discreteness threshold, the density distribution type of the stage sub-scores under the scoring dimension is determined to be discrete distribution; when the discreteness of the stage sub-scores under the scoring dimension is less than the preset second discreteness threshold, the density distribution type of the stage sub-scores under the scoring dimension is determined to be dense distribution.

[0162] Optionally, when the weight level determination module 205 is used to determine the weight level of the first target scoring dimension and the weight level of the second target scoring dimension based on the skewed distribution type and the density distribution type of the score distribution respectively by a predetermined scoring dimension level determination rule, the weight level determination module 205 is used to:

[0163] When the skewed distribution type of the stage score distribution is positively skewed distribution or negatively skewed distribution, determining the weight level of the first target scoring dimension to be a low level;

[0164] When the density distribution type of the stage score distribution is discrete distribution or concentrated distribution, the weight level of the second target scoring dimension is determined to be a high level.

[0165] Optionally, the spoken language testing device 200 further includes a text information determination module 209, and the text information determination module 209 is used to:

[0166] Acquire multiple test texts; wherein each test text corresponds to multiple scoring dimensions, and each scoring dimension is pre-set with a corresponding initial weight coefficient;

[0167] For each test text, score feature extraction is performed on the test text to determine score parameters of multiple score features included in the test text;

[0168] According to a predetermined weight coefficient adjustment rule, based on the scoring parameters of the various scoring features included in the test text, the initial weight coefficients of the corresponding scoring dimensions are adjusted to determine the target weight coefficient of each scoring dimension of the test text; the weight coefficient adjustment rule specifies the mapping relationship between the scoring parameters of the scoring features and the weight coefficient adjustment amount;

[0169] According to a predetermined weight level rule, based on the target weight coefficient of each scoring dimension corresponding to the test text, the weight level of each scoring dimension corresponding to the test text is determined; the weight level rule stipulates the mapping relationship between the weight coefficient and the weight level, and the weight level includes high, medium and low.

[0170] Optionally, the spoken language testing device 200 further includes a fine-tuning module 210, and the fine-tuning module 210 is used to:

[0171] Get the total oral test score of each candidate;

[0172] Under the condition that the ranking remains unchanged, the total oral test scores of all candidates are slightly adjusted so that the score distribution of the total oral test scores of all candidates after adjustment is normal distribution;

[0173] The fine-tuned score will be used as the candidate's final oral test score.

[0174] See also Figure 4 , Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Figure 4 As shown in , the electronic device 400 includes a processor 410 , a memory 420 and a bus 430 .

[0175] The memory 420 stores machine-readable instructions executable by the processor 410. When the electronic device 400 is running, the processor 410 communicates with the memory 420 via the bus 430. When the machine-readable instructions are executed by the processor 410, the above-mentioned Figure 1 The specific implementation of the steps of the oral test method in the method embodiment shown can be found in the method embodiment, and will not be repeated here.

[0176] The present application also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the computer program can execute the above-mentioned Figure 1 The specific implementation of the steps of the oral test method in the method embodiment shown can be found in the method embodiment, and will not be repeated here.

[0177] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0178] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of devices or units can be electrical, mechanical or other forms.

[0179] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0180] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0181] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium that is executable by a processor. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, and other media that can store program codes.

[0182] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present application, which are used to illustrate the technical solutions of the present application, rather than to limit them. The protection scope of the present application is not limited thereto. Although the present application is described in detail with reference to the above-mentioned embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-mentioned embodiments within the technical scope disclosed in the present application, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A spoken language test method based on dynamically adjusting the test text, characterized in that: The oral test method comprises: After each test section other than the last section is completed, the oral score of each candidate up to the current stage is obtained; the oral score of the stage includes stage sub-scores under multiple scoring dimensions; Determine a stage score distribution based on the stage oral scores of all candidates, and determine whether the stage score distribution is a normal distribution: When the answer is no, based on the stage oral scores of all candidates, determine the skewness distribution type of the stage score distribution and the density distribution type of the stage score distribution, and determine the skewness distribution type and density distribution type of the stage sub-scores under each scoring dimension; The scoring dimension corresponding to the stage sub-score with the same skewed distribution type as the stage score distribution is determined as the first target scoring dimension, and the scoring dimension corresponding to the stage sub-score with the same density distribution type as the stage score distribution is determined as the second target scoring dimension; Based on the skewed distribution type and the density distribution type of the stage score distribution, respectively, and by a predetermined scoring dimension grade determination rule, determine the weight grade of the first target scoring dimension and the weight grade of the second target scoring dimension; In the test text database, target test texts that meet the weight level of the first target scoring dimension and the weight level of the second target scoring dimension are extracted for the next oral test until the end condition is met and the oral test is stopped; each test text corresponds to multiple scoring dimensions with weight levels and target weight coefficients.

2. The oral test method according to claim 1, characterized in that: The following steps are used to determine the stage sub-scores under multiple scoring dimensions included in the stage oral score: After each test except the last one, obtain the first oral score of each candidate after each test; For each first oral score, determining a second oral score under multiple scoring dimensions included in the first oral score according to target weight coefficients corresponding to multiple scoring dimensions corresponding to the test text used to obtain the first oral score; Add up all the second oral scores under the same scoring dimension to obtain the stage sub-score under each scoring dimension included in the stage oral score.

3. The oral test method according to claim 1, characterized in that: Determine whether the distribution of grades in the stage is normal by following the steps below: Determine the similarity between the waveform of the stage score distribution and the normal distribution; When the waveform similarity is higher than a preset similarity threshold, it is determined that the stage score distribution is a normal distribution.

4. The oral test method according to claim 1, characterized in that: When the stage scores are normally distributed, the oral test method includes: A test text is randomly selected from the test text database for the next oral test, and the oral test is stopped until the end condition is met.

5. The oral test method according to claim 1, characterized in that: The skewed distribution type includes positive skewed distribution and negative skewed distribution, the density distribution type includes discrete distribution and concentrated distribution, and the step of determining the skewed distribution type of the stage score distribution and the density distribution type of the stage score distribution based on the stage oral scores of all examinees, and determining the skewed distribution type and density distribution type of the stage sub-scores under each scoring dimension includes: Based on the oral scores of all candidates, the dispersion of oral scores and the dispersion of sub-scores of each scoring dimension are calculated, and the number of candidates whose oral scores are lower than the third quartile and the number of candidates whose oral scores are higher than the fifth quartile are calculated. When the number of examinees whose oral scores at the stage are lower than the third octave is greater than the preset number threshold, the skewed distribution type of the stage score distribution is determined to be a positive skewed distribution; when the number of examinees whose oral scores at the stage are higher than the fifth octave is greater than the preset number threshold, the skewed distribution type of the stage score distribution is determined to be a negative skewed distribution; When the discreteness of the oral scores of the stage is greater than a preset first discreteness threshold, the density distribution type of the stage score distribution is determined to be discrete distribution; when the discreteness of the oral scores of the stage is less than a preset second discreteness threshold, the density distribution type of the stage score distribution is determined to be dense distribution; For each scoring dimension, when the number of examinees with scores below the third octave under the scoring dimension is greater than the preset number threshold, the skewed distribution type of the stage sub-score under the scoring dimension is determined to be a positive skewed distribution; when the number of examinees with scores above the fifth octave under the scoring dimension is greater than the preset number threshold, the skewed distribution type of the stage sub-score under the scoring dimension is determined to be a negative skewed distribution; For each scoring dimension, when the discreteness of the stage sub-scores under the scoring dimension is greater than the preset first discreteness threshold, the density distribution type of the stage sub-scores under the scoring dimension is determined to be discrete distribution; when the discreteness of the stage sub-scores under the scoring dimension is less than the preset second discreteness threshold, the density distribution type of the stage sub-scores under the scoring dimension is determined to be dense distribution.

6. The oral test method according to claim 5, characterized in that: The scoring dimension level determination rule specifies a mapping relationship between the skewed distribution type and the weight level, and a mapping relationship between the density distribution type and the weight level. The weight level of the first target scoring dimension and the weight level of the second target scoring dimension are determined by the predetermined scoring dimension level determination rule based on the skewed distribution type and the density distribution type of the stage score distribution, respectively, including: When the skewed distribution type of the stage score distribution is positively skewed distribution or negatively skewed distribution, determining the weight level of the first target scoring dimension to be a low level; When the density distribution type of the stage score distribution is discrete distribution or concentrated distribution, the weight level of the second target scoring dimension is determined to be a high level.

7. The oral test method according to claim 1, characterized in that: The following steps are used to determine the weight levels and target weight coefficients of the multiple scoring dimensions corresponding to each test text: Acquire multiple test texts; wherein each test text corresponds to multiple scoring dimensions, and each scoring dimension is pre-set with a corresponding initial weight coefficient; For each test text, extract scoring features of the test text and determine scoring parameters of multiple scoring features included in the test text; According to a predetermined weight coefficient adjustment rule, based on the scoring parameters of the various scoring features included in the test text, the initial weight coefficients of the corresponding scoring dimensions are adjusted to determine the target weight coefficient of each scoring dimension of the test text; the weight coefficient adjustment rule specifies the mapping relationship between the scoring parameters of the scoring features and the weight coefficient adjustment amount; According to a predetermined weight level rule, based on the target weight coefficient of each scoring dimension corresponding to the test text, the weight level of each scoring dimension corresponding to the test text is determined; the weight level rule stipulates the mapping relationship between the weight coefficient and the weight level, and the weight level includes high, medium and low.

8. The oral test method according to claim 1, characterized in that: After stopping the oral test, the oral test method further includes: Get the total oral test score of each candidate; Under the condition that the ranking remains unchanged, the total oral test scores of all candidates are slightly adjusted so that the score distribution of the total oral test scores of all candidates after adjustment is normal distribution; The fine-tuned score will be used as the candidate's final oral test score.

9. A spoken language test device based on dynamically adjusting test text, characterized in that: The oral testing device comprises: An acquisition module is used to acquire the oral scores of each candidate up to the current stage after each test except the last stage is completed; the oral scores of the stage include stage sub-scores under multiple scoring dimensions; The normal distribution determination module is used to determine the stage score distribution based on the stage oral scores of all candidates, and determine whether the stage score distribution is a normal distribution: A distribution type determination module is used to determine the skewed distribution type and the density distribution type of the stage score distribution based on the stage oral scores of all examinees when the stage score distribution is not a normal distribution, and to determine the skewed distribution type and the density distribution type of the stage sub-score under each scoring dimension; A target dimension determination module, used to determine the scoring dimension corresponding to the stage sub-score of the same skewed distribution type as the stage score distribution as the first target scoring dimension, and determine the scoring dimension corresponding to the stage sub-score of the same density distribution type as the stage score distribution as the second target scoring dimension; A weight level determination module, for determining the weight level of the first target scoring dimension and the weight level of the second target scoring dimension based on the skewed distribution type and the density distribution type of the stage score distribution, respectively, by using a predetermined scoring dimension level determination rule; The first selection module is used to extract target test texts that meet the weight level of the first target scoring dimension and the weight level of the second target scoring dimension from the test text database for the next oral test until the end condition is met and the oral test is stopped; each test text corresponds to multiple scoring dimensions with weight levels and target weight coefficients.

10. An electronic device, characterized in that: include: A processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor and the memory communicate through the bus, and the machine-readable instructions are executed by the processor to execute the steps of the oral test method as described in any one of claims 1 to 8.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the oral testing method according to any one of claims 1 to 8 are executed.

Images