KR101812750B1 - System for generating examination of korean spelling - Google Patents

Abstract

The present invention relates to a Hangul spelling test paper generating system, and more particularly, it can improve the reliability of a test by generating a spelling check paper having a constant degree of difficulty, and when an examiner sets a goal, extracts items from the item database The present invention relates to a system and method for generating a Korean spelling test paper so that an optimal test paper can be generated by extracting items suitable for a target level of a student set by an examiner from an article database.

Description

{SYSTEM FOR GENERATING EXAMINATION OF KOREAN SPELLING}

The present invention relates to a Hangul spelling test paper generating system, and more particularly, it can improve the reliability of a test by generating a spelling check paper having a constant degree of difficulty, and when an examiner sets a goal, extracts items from the item database The present invention relates to a system and method for generating a Korean spelling test paper so that an optimal test paper can be generated by extracting items suitable for a target level of a student set by an examiner from an article database.

The Korean Spelling Developmental Test (Yang Min-hwa, Bae Jong-min, 2014) and KOLRA (Korean Reading Test, Bae Soo-young, Kim Mi Bae, Yoon Hyo-jin, Chang Sang Min) are standardized test tools to measure spelling ability.

The Korean spelling development test corresponding to the former is standardized until the first to third grades of elementary school. There are two types of the spelling test, one is phonological type and the other is morpheme type. This test is similar to the traditional dictation method in which the examinee writes words and example sentences. The test result is verified with the naked eye, the score is calculated by direct scoring, and the standard score It is inconvenient to use and it is difficult to secure objectivity in the evaluation result. In addition, the score calculation process is not easy because it is composed of whole words and target phonemes. The test paper consists of only four types of phonemic type (type and bell type) and type small type (type and bell type) It was insufficient.

  The latter KOLRA test is standardized for elementary school grades 1-6, and there is a dictation test in the subregion of the writing evaluation area. This test consists of writing words and sentences. The test results are calculated by the self-scoring by the examiner, and the standard score is calculated by inputting the number of origin in the computer. However, since this test is also composed of one type, it is difficult to check the progress of learning, and since the test result provides a standard score based on the total score of dictation, it is difficult to analyze the spoken ability level of the testee in detail .

All of the two tests described above can only provide a spell-level test of the testee's age, and they can be tested for spelling (eg, difficulty in beginning, difficulty in phonological change, etc.) And there is a problem that it is difficult to provide information for linking the evaluation result with the arbitration because various inspection sites are not provided for conducting the progress inspection.

On the other hand, in Korean Patent Laid-Open Publication No. 2001-0079233 (published on August 22, 2001), which is related to a test paper automatic generation system by extracting general test items, a learner is given a mathematical problem consisting of a plurality of terms A learning problem generation system, comprising: a problem generation rule in which a distribution of a number of terms and a relationship between numbers of the terms are set; a difficulty level is automatically set by analyzing a learning ability of the learner; The problem generation rule is searched according to the learning contents of the question creation rule, and a problem generation is generated in real time by generating a random number using the retrieved problem generation rule, and is provided to the learner. , You can find the questionnaire including the question It is not possible to provide a means for securing the reliability of the test because the scores obtained from the respective tests have the same meaning, and furthermore, the test results are continuously accumulated and updated, And does not provide a structure for facilitating the understanding of the level and the degree of progress (degree of progress), and furthermore, it does not provide a structure for verifying additional preliminary questions to be added to the question bank.

As another related prior art, a method for examining and scoring test questions having different selection arrangements on the Internet is disclosed in Korean Patent Laid-Open Publication No. 10-2010-0111819 (published on Oct. 18, 2010) The candidate accessing a test question and question system on the Internet and requesting a test question for learning; Generating a plurality of individual test papers comprising test questions having different selection arrangements according to a request of a candidate; Sending a test sheet to a test taker's terminal requesting a test question together with individual test papers randomly called out among a plurality of individual test papers; The candidate who receives the individual test sheets and the answer sheet through the terminal creates an answer for the individual test sheet and transmits the answer sheet to the test question drafting and scoring system; And the step of scoring the candidate's answer sheet, while the test score questioning and scoring system scored the candidate's answer sheet. However, this technique also has various points of difficulty, The test result is continuously accumulated and updated so that it is easy to grasp the current level and the improvement degree of the testee (the progress of the spelling ability) And not to present a structure for the verification of additional preliminary questions to be added to the question bank.

Korean Patent Publication No. 2001-0079233 Korean Patent Publication No. 10-2010-0111819

The present invention has been made to solve the limitations and problems of the prior arts, and it is an object of the present invention to automatically extract some selection items from a plurality of spelling test selection items stored in a test item database having various items, It is possible to analyze the spelling ability level of the testee in detail and to provide a means of generating a spell check sheet having a difficulty level so as to secure the objectivity and reliability of the spell check through each generated spell check sheet, By providing a means for extracting selection items suitable for the target level (eg, upper 30% level or upper 10% level) of the testee (student) set by the examiner and generating a spelling test sheet for each level, Make it possible to measure spell level and progress (degree of progress), and The reliability and validity of the test subjects and the results of individual tests are compared and compared with the test results of the upper level testees set at the top 30% or the top 10% level to easily ascertain the current level and the improvement level The present invention also provides a system for generating a Hangul spelling test report.

According to another aspect of the present invention, there is provided a system for generating a Hangul spelling test page,

An input unit for inputting unique information (name, grade, sex, date of birth) of the examinee and a test target selected by the examiner;

A test subject data DB storing test subject unique information and test related data;

A questionnaire DB for spell checking which stores a selection item for spell checking and a spare item for spell checking;

A selection item (for example, 25-30 items) matching the test subject's selected test target with the unique information of the testee inputted by the input unit from the spelling test selection items stored in the test item DB for the spelling test, An item extraction unit for extracting additional preliminary items from the preliminary items for spell checking stored in the item DB; And

In addition to the evaluation items that are scored with the scores of the testee, the test items are generated by generating a set of test items of the entire test by applying the statistical models and algorithms using the selected items extracted from the test item DB for spelling test. , A basic item for comparison that is used as an index that relatively objectifies the final score of the testee is not directly included in the final score but the new test item includes a new additional item to be added as a preliminary item A spell check sheet generating and storing unit for storing the generated spell check sheet in a form of a file in the subject data DB as the test related data in connection with the unique information of the test subject;

A spelling checker output unit outputting a spelling checker generated by the spelling checker generating and storing unit; And

From the results of the separate scoring of the additional preliminary questions included in the generated spelling test, it is determined that the adequacy of the additional preliminary items as a selection item is appropriate to be used as a selection item (i.e., evaluation item) in terms of degree of difficulty and discrimination And a preliminary question analyzing section for allowing the additional preliminary question to be changed and stored as a selection item for spelling inspection in the spelling examination question DB.

Here, it is possible to set an evaluation target, which includes a leading character, a vowel, a vowel, and a phonological variation, as inspection targets set by the examiner, and the set inspection target includes, for example, a selectable upper 30% The level of difficulty level can be set to the target level of evaluation.

Here, the preliminary questions are stored in the question database for the spelling test in the form of a question bank, and when the spelling test is generated, some of the preliminary questions that are not scored for evaluation of the spelling ability of the examinees The additional preliminary question is included in a plurality of spelling checklists, and when it is judged that it is suitable to be used as a selection item (that is, an evaluation item) in terms of degree of difficulty and discrimination through the verification procedure, The change item is stored in the DB as a selection item for spell checking.

In the present invention, the unique information of the testee includes details of date of birth, school age, region, and sex.

The system for generating a Hangul spelling test paper according to the present invention as described above has the following effects.

According to the present invention, some selected items are automatically extracted from a plurality of spelling test items stored in a test item database having various items to generate a new spelling test sheet, and the spelling ability level of the testee is precisely Thereby enabling the user to generate a spell check sheet having a difficulty level.

In addition, among the selected items, items that reflect the content of each grade level and have moderate difficulty and discrimination are used as common items commonly used in many spelling test sites, and the scoring results of the answer to this question By not being reflected in the final score, it is possible to equalize the meanings of the scores derived from the different examiners so that homogeneity can be secured in the score interpreting, thereby ensuring the objectivity and reliability of the spelling test through the respective spelling checklists (Eg, upper 30% level or upper 10% level) of the testee (student) set by the examiner and extracts a selection item suitable for the level, Make it possible to measure the spelling level and progress (progress) of the student, The reliability and validity of the test subjects and the results of individual tests are compared and compared with the test results of the upper level testees set at the top 30% or the top 10% level to easily ascertain the current level and the improvement level And the like.

In addition, according to an embodiment of the present invention, a preliminary item analysis unit for analyzing a result of separate scoring of additional preliminary items included in the generated spelling matter test sheet is included, so that a newly added additional preliminary item is included in a plurality of spelling test papers (Ie, evaluation items) in terms of degree of difficulty and discrimination, it is possible to continuously determine the selection items for spell checking in the item DB for spell checking And the like.

Brief Description of the Drawings Fig. 1 is an overall configuration diagram of a Hangul spelling test paper generating system according to an embodiment of the present invention.
FIG. 2A is a flowchart illustrating a process of generating a Hangul spelling checker according to an embodiment of the present invention; FIG.
FIG. 2B is a flowchart illustrating a process of marking a spelling test sheet and analyzing the result according to an embodiment of the present invention. FIG.
3 is a view for explaining an iterative algorithm used at the time of item generation in the process of generating Hangul spelling test paper according to an embodiment of the present invention;

In the following, the present invention will be described with reference to the accompanying drawings. In the following description of the present invention, a detailed description of known functions and configurations incorporated herein will be omitted when it may make the subject matter of the present invention rather unclear.

FIG. 1 is an overall configuration diagram of a Hangul spelling test paper generating system according to an embodiment of the present invention.

As shown in FIG. 1, the Hangul spelling check page generating system according to an embodiment of the present invention includes:

An input unit 110 for receiving unique information (name, grade, sex, date of birth) of the examinee and an inspection target selected by the inspector;

A subject data base 120 in which the subject data and examination related data are stored;

A test item database 130 for storing spelling test items and spelling test items;

(For example, 25-30 questions) matching the test target selected by the examiner and the unique information of the subject inputted by the input unit 110 from the spelling test selection items stored in the spelling test item DB 130. [ An item extraction unit 141 for extracting an additional preliminary item from the spelling test preliminary items stored in the item test database 130;

A set of test items of a whole test is generated by applying a statistical model and an algorithm using the selected items extracted from the test item DB 130 for spelling test to generate a spelling test sheet, In addition to the evaluation items, there are basic items for comparison, which are used as indicators to relatively objectify the final score of the testee, though the scores are not directly included in the final score, and a new additional reserve A spelling test paper generation and storage unit 142 for storing the spelling test paper in the form of a file in the subject data DB 120 as the test related data in association with the unique information of the subject;

A spelling checker output unit 150 for outputting a spelling checker generated by the spelling checker generating and storing unit 142; And

From the results of the separate scoring of the additional preliminary questions included in the generated spelling test, it is determined that the adequacy of the additional preliminary items as a selection item is appropriate to be used as a selection item (i.e., evaluation item) in terms of degree of difficulty and discrimination And a preliminary question analyzing unit 174 for storing the additional preliminary question in the spelling examination questionnaire database 130 as a selection item for spelling examination.

Meanwhile, as shown in FIG. 1, the question extracting unit 141 and the spelling check place generating and storing unit 142 may be integrally configured as an examination place control unit 140. FIG.

Here, the preliminary questions are stored in the questionnaire DB 120 in the form of a question bank, and when the spelling checker is generated, some questions among the preliminary questions that are not scored for evaluation of the spelling ability of a testee, , The additional preliminary questions added to the spelling checklist are included in the spelling checklists, and when it is judged that they are suitable to be used as selection items (that is, evaluation items) in terms of degree of difficulty and discrimination, It is saved as a selection item for spell checking in the item DB for inspection.

In the present invention, the unique information of the testee includes details of date of birth, school age, region, and sex.

Hereinafter, a process of generating Hangul spelling test paper according to an embodiment of the present invention will be described with reference to the flowchart of FIG. 2A.

First, the function of a main constituent part of the Hangul spelling check paper generating system according to an embodiment of the present invention can be implemented through a computer program such as a Hangul spelling check paper generating program. S101), the inspector inputs unique information of the examinee and the inspection target through the input unit 110 (steps S102 and S103). As an inspection target, a preliminary, vowel, longitudinal, phonological fluctuation , And an evaluation level target such as the upper 30% level or the upper 10% level may be set as the target criterion of the testee's spelling ability. Next, the item extraction unit 141 of the test sheet control unit 140 extracts a selection item and an additional preliminary item from the spelling test item database 130 (S104) (Sejong corpus, etc: https://ithub.korean.go.kr/user/noticeView.do) can be constructed as a question bank. In the Sejong corpus, words related to the curriculum of each grade are extracted, , And 10-fold of the extracted items are stored in the question bank and stored and managed as preliminary items. The item extraction unit 141 selects the selected items as the items constituting the spelling examiner among the preliminary questions. On the other hand, among the selected items, items that reflect the content of each grade level and have moderate degree of difficulty and differentiation are used as common items commonly used in many spelling test sites, and the scoring results of the answers It is not reflected in the final score. It is used to equalize the meanings of scores derived from different examiners and to ensure homogeneity in score interpretation.

 According to the present invention, in the spell checking sheet generating and storing unit 142, a spell checking sheet is generated and stored using a selection item extracted by the item extracting unit 141. First, The modeling and algorithm is applied to generate the items of the whole test in a single dimension. The items are selected by selecting one of the following parameters or non-parametric models according to the scale of the data pool or the purpose of the test.

1. Parametric and nonparametric models:

① Nonparametric model: The nonparametric model is used in the following two cases. First, in the early stage, the number of testees is small, so it is often not appropriate to use a statistical model in which the ability, difficulty, and discrimination of the testees are distributed. In the beginning, it is appropriate to use a nonparametric model because the distribution of data pools (data pools stored as selection items in the spelling check item database 130) is not a normal distribution. Second, we set the target level (eg, upper 30% of the spelling level by age) and use the nonparametric model when we want to check whether the testee has reached. If you set the goal level, if you set the top 30% of the spelling level by age or grade, questions with high difficulty level at the grade or age level will be extracted through the nonparametric model.

② Parametric model: It is a model that generates questions when the number of test cases accumulated in the data pool accumulates, and the difficulty and discrimination of the testees and the items form a specific distribution, that is, a normal distribution.

2. Using EM algorithm and MMLE

① EM algorithm:

The EM algorithm is an effective iterative algorithm used when estimating the most probable model when no information is determined. That is, as a method for estimating a model parameter when it is not known, the probability that observation data X is generated from this model (lambda_0) is calculated at the initial stage in which the model variable is set at random or by some other method do.

  P (X | lambda_0)

Now, we define a new model with a very small modification of the variables inside the model as lambda_1, then again find the posterior probability of the observed data and determine what is larger P (X | lambda_1). This process is repeated to stop P (X | lambda) for the new model lambda if it is less than the previous model (if the likelihood is smaller). E and M steps (see FIG. 3).

 - Step E (E Step): Assuming that the testee's ability level is fixed, estimate the degree of difficulty and discrimination of the item, and then estimate the degree of difficulty and discrimination of the item. Because there is no information about difficulty, discrimination, and testee level, both estimates are made.

 - M Step: At this stage, it is a step to find out the possibility of representing the difficulty of the item, the degree of discrimination and the level of the testee best.

3. Using the two-factor, multi-item question model considering item difficulty and discrimination, construct a question by choosing one of the two questions or multi-item questions to construct the item that fits the purpose of question generation.

① This item is a question that scored the type (eg, basic beginner, basic vowel, basic vowel, etc.) or words that are expected to be seen in one item as zero and one point. There are 1 parameter model, 2 parameter model, and 3 parameter model. The model is decided according to the purpose of inspection.

  - 1 Parametric Model: When determining the test taker's spelling level

  - 2 Parametric Model: When trying to distinguish difficulty level of item, attainment level of testee's spelling ability

  - 3 parameter model: It is used to estimate difficulty, discrimination, and guessing,

② Multi-item Question: The multi-item item estimates one discrimination and two difficulty levels for each item using the generalized partial credit model. That is, each difficulty corresponding to each partial score can be calculated, and it can be interpreted as difficulty to receive more than zero point, difficulty to receive more than one point. When calculating the score of each item, the generalized subscore model can be used to calculate the scores of the testee.

As described above, the process of generating the spelling checker is performed, and the spelling checker, which has been created, is stored in the form of a file in the subject data base DB 120 as inspection related data associated with the subject information of the subject (S105).

Then, the completed spelling checker is outputted through the spelling checker output unit 150 (S106).

In addition, in the automatic scoring extension category of the present invention as described above, when the subject's answer to the output spelling test sheet is completed, that is, when the spelling test is completed, a process of marking the completed spelling test sheet and analyzing the result may be further performed , This specific process will be described with reference to the flow chart of FIG. 2B attached hereto.

First, among the functions of the main constituent part of the Hangul spelling check paper generating system according to the embodiment of the present invention, the scoring process of the spelling checker and the analysis of the examination result can also be implemented through a computer program such as a Hangul spelling checker inspection result analysis program, The inspector performs the Hangul spelling checker examination result analysis program (S201) while scanning or photographing the spelling checker imaged by the spelling checker imaging unit 160 using an apparatus such as a camera or a scanner in the form of an image file and storing S202). The spelling checker scoring unit 143 extracts a character from the stored image file, compares the extracted character with the spelling checker file stored in the subject DB 120, and performs scoring (S203). In addition to the evaluation items included in the question set, the score items between different spelling checkers have the same meaning (without affecting the number or timing of scoring) And the additional additional preliminary item added to the spelling checklist is used as an optional item for the additional preliminary item to be the evaluation item. It has goodness and is used as an appraisal.

The scoring result analyzing unit 171 of the result analyzing unit 170 analyzes the scoring result (score, current level, progress degree) of the examinee by using the result analyzing unit 170 The error analysis unit 172 derives the error analysis result, and the development level analysis unit 173 derives the phoneme prediction development level analysis result and displays it on the display unit 180 (S204a).

In addition, according to the embodiment in which the preliminary item analysis unit 174 is additionally configured in the result analysis unit 170 as in the embodiment of FIG. 1 showing the preferred embodiment of the present invention, the preliminary item analysis unit 174 additionally (S204b1) as a selection item from the result of the separate scoring of the additional preliminary item contained in the examination sheet, and when the result is determined to be an appropriate item, the selection item for spelling examination is added to the spelling examination item DB 130 The change is stored (S204b2).

While the invention has been shown and described with reference to certain preferred embodiments thereof, it is to be understood that the invention is not limited to the disclosed embodiments. Those skilled in the art will appreciate that various modifications, additions and substitutions are possible, without departing from the scope of the appended claims, It would be too obvious to belong to. Accordingly, the true scope of the present invention should be determined by the technical idea of the appended claims.

110: input unit
120: Test subject data DB
130: Questionnaire for spelling test
140:
150: Spelling checker output unit
160: spelling checker imaging unit
170: Result Analysis Unit
171: Scoring result analysis section
172: Error result analysis unit
173: Development Level Analysis Department
174: Preliminary Item Analysis Department
180:

Claims (3)

An input unit 110 for inputting the unique information of the examinee and the inspection target selected by the inspector;
A subject data base 120 in which the subject data and examination related data are stored;
A test item database 130 for storing spelling test items and spelling test items;
A selection item corresponding to a test subject selected by the examiner, and a selection item for the spelling check DB (130), which are inputted by the input unit (110) from the spelling check items stored in the spelling check item DB (130) 130) for extracting additional preliminary questions from the preliminary questions for spell checking;
A set of test items of a whole test is generated by applying a statistical model and an algorithm using the selected items extracted from the test item DB 130 for spelling test to generate a spelling test sheet, In addition to the evaluation items, there are basic items for comparison, which are used as indicators to relatively objectify the final score of the testee, though the scores are not directly included in the final score, and a new supplementary reserve A spelling test paper generation and storage unit 142 for storing the spelling test paper in the form of a file in the subject data DB 120 as the test related data in association with the unique information of the subject;
A spelling checker output unit 150 for outputting a spelling checker generated by the spelling checker generating and storing unit 142; And
If the result of the separate spelling test for the additional spelling question included in the generated spelling test is judged to be suitable as a selection item in terms of degree of difficulty and discrimination, the additional preliminary question And a preliminary item analysis unit (174) for storing the selected item as a selection item for spelling inspection in the spelling examination item DB (130). 2. The system of claim 1, wherein the test target is set as an evaluation target target including a leading character, a vowel, a vowel, and a phonological variation. 2. The method of claim 1, wherein the inspection target is set to an evaluation level target as a higher difficulty level including a selectable upper 30% level or a higher 10% level, and
The result analyzer 170 is configured to perform a step of comparing the score of the testee's score with the score of the top 30% or the cumulative score of the top 10% students to determine the current level and progress of the testee Korean spelling checker creation system.

Images