WO2004029906A1 - テスト・システム及びその制御方法 - Google Patents
テスト・システム及びその制御方法 Download PDFInfo
- Publication number
- WO2004029906A1 WO2004029906A1 PCT/JP2003/012252 JP0312252W WO2004029906A1 WO 2004029906 A1 WO2004029906 A1 WO 2004029906A1 JP 0312252 W JP0312252 W JP 0312252W WO 2004029906 A1 WO2004029906 A1 WO 2004029906A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- test
- partial score
- computer
- answer
- model
- Prior art date
Links
- 238000012360 testing method Methods 0.000 title claims abstract description 178
- 238000000034 method Methods 0.000 title claims description 46
- 238000006243 chemical reaction Methods 0.000 claims abstract description 22
- 230000004044 response Effects 0.000 claims description 137
- 238000009826 distribution Methods 0.000 claims description 40
- 238000005315 distribution function Methods 0.000 claims description 15
- 238000011156 evaluation Methods 0.000 abstract description 26
- 238000004088 simulation Methods 0.000 description 28
- 238000007476 Maximum Likelihood Methods 0.000 description 22
- 230000006870 function Effects 0.000 description 14
- 238000012545 processing Methods 0.000 description 11
- 230000000694 effects Effects 0.000 description 9
- 230000008569 process Effects 0.000 description 9
- 238000004458 analytical method Methods 0.000 description 7
- 238000002790 cross-validation Methods 0.000 description 6
- 238000011160 research Methods 0.000 description 6
- 238000001276 Kolmogorov–Smirnov test Methods 0.000 description 5
- 238000004891 communication Methods 0.000 description 5
- 239000000203 mixture Substances 0.000 description 5
- 230000008859 change Effects 0.000 description 4
- 230000001186 cumulative effect Effects 0.000 description 3
- 230000007423 decrease Effects 0.000 description 3
- 230000003044 adaptive effect Effects 0.000 description 2
- 238000004590 computer program Methods 0.000 description 2
- 238000012790 confirmation Methods 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 238000003541 multi-stage reaction Methods 0.000 description 2
- 238000010998 test method Methods 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 125000002066 L-histidyl group Chemical group [H]N1C([H])=NC(C([H])([H])[C@](C(=O)[*])([H])N([H])[H])=C1[H] 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000004308 accommodation Effects 0.000 description 1
- 230000002411 adverse Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000000556 factor analysis Methods 0.000 description 1
- 238000009472 formulation Methods 0.000 description 1
- 238000012812 general test Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 230000003340 mental effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000011056 performance test Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 230000003252 repetitive effect Effects 0.000 description 1
- 238000013077 scoring method Methods 0.000 description 1
- 230000006641 stabilisation Effects 0.000 description 1
- 238000011105 stabilization Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G09—EDUCATION; CRYPTOGRAPHY; DISPLAY; ADVERTISING; SEALS
- G09B—EDUCATIONAL OR DEMONSTRATION APPLIANCES; APPLIANCES FOR TEACHING, OR COMMUNICATING WITH, THE BLIND, DEAF OR MUTE; MODELS; PLANETARIA; GLOBES; MAPS; DIAGRAMS
- G09B7/00—Electrically-operated teaching apparatus or devices working with questions and answers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/20—Education
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F15/00—Digital computers in general; Data processing equipment in general
- G06F15/16—Combinations of two or more digital computers each having at least an arithmetic unit, a program unit and a register, e.g. for a simultaneous processing of several programs
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
-
- G—PHYSICS
- G09—EDUCATION; CRYPTOGRAPHY; DISPLAY; ADVERTISING; SEALS
- G09B—EDUCATIONAL OR DEMONSTRATION APPLIANCES; APPLIANCES FOR TEACHING, OR COMMUNICATING WITH, THE BLIND, DEAF OR MUTE; MODELS; PLANETARIA; GLOBES; MAPS; DIAGRAMS
- G09B19/00—Teaching not covered by other main groups of this subclass
- G09B19/06—Foreign languages
Definitions
- the present invention relates to a test system and a control method thereof. More specifically, the present invention relies on a partial scoring model that is a modification of the conventional item response theory, and allows scoring not only a single true and false value but also a partial score with multiple levels when scoring.
- the present invention relates to a test system for designing, executing, and evaluating a test in a more simplified manner than before, and a control method thereof. Background art
- test theory Today, there are two widely known theories used in designing tests and processing the results: classical test theory and item response theory. For a general explanation of these test theories, see Chapters 6 and 7 of “The Techniques of Psychological Statistics”, edited by Hiroshi Watanabe (Fukumura Publishing, 2002), respectively.
- the test in the test theory includes not only academic ability tests but also personality tests and clinical tests in psychology.In this application, such wide application fields are considered. Rather than giving an abstract explanation, I would like to try to give a concrete explanation in order to facilitate understanding, especially with academic ability tests such as foreign language examinations in mind.
- the term “item” in item response theory means a problem in the case of academic ability tests.
- Item response theory is a theory that overcomes the shortcomings of classical test theory, and there are many scholarship tests designed and processed based on this item response theory.
- Hideki Toyoda “Introduction to Item Response Theory” (Asakura Shoten, 2002)
- the well-known language test T 0 EFL... is performed many times a year, and is performed worldwide. It is a collection of the same items because the same subject may be retaken. The same test cannot be used more than once, so the average score and pass rate will differ from test to test, and the distribution of characteristic values will also differ due to differences in English proficiency depending on the region.
- the OEFL points (for example, 500, 650, etc.) Used to determine whether or not to take the exam regardless of where and when the exam was taken, that is, a quality candidate took a different item at a different date and time in a different location Despite this, the subject was given unified treatment A mathematical model that builds a system that continuously and positively performs tests that enable this treatment is an item response model. ”
- the test questions to be set are arranged in a tree beforehand, and the questions are sequentially presented along the route arranged in the tree according to the correctness of the answer by the examinee.
- a test method and system for estimating the examinee's ability in consideration of not only the number of correct answers but also the course of reaching the final point by a route is disclosed. This published patent publication also mentions item reaction theory.
- the use of item response theory is stated, and the questions to be taken are arranged in a tree beforehand.However, if the examinee answers a certain question correctly, the question is located in the lower right corner, and if the wrong answer is answered, the question is located in the lower left corner It is expected that the answer will be one of two values: correct or incorrect. Disclosure of the invention
- the present invention is different from the conventional CAT that predicts a binary answer, and is a test system that allows scoring to give a partial score.
- An object of the present invention is to provide a test, a system, and a control method thereof, which enable processing of partial scores much more easily than a simple model.
- a first computer having an input device and an output device, and a first input device and an output device connected to the first computer via a network including the Internet.
- a second computer a test management server connected to the first and second computers via the network, and a difficulty level and discriminating ability accessible from the test management server.
- a test system for estimating the ability 0 of the examinee from the response of the examinee is provided.
- the test management server (1) responds to the request transmitted from the first computer, and sets 0 ⁇ r. ⁇ 1 with 1 being the perfect score for the problem j in which 1 ⁇ j ⁇ n.
- ⁇ (( ⁇ ) indicates that the partial score r ; is specific to the question j, and the examinee has a potential response of either correct answer 1 or incorrect answer 0. If the candidate is assumed to be an average of the correct and incorrect responses that the candidate can potentially take when the potential problem that can be taken is repeated sj times, the probability that the candidate will correctly answer the potential problem is Yes.
- a '' and b are the discriminating power and the difficulty, which are the inherent characteristics of the problem stored in the problem database, and D is 1.7. If Q j ( ⁇ ) is one and one P j ( ⁇ ),
- the examinee's ability 0 is estimated using the log likelihood ⁇ represented by
- the function form of P j ( ⁇ ) expressed as Equation 1 above is merely an example, and P j ( ⁇ ) need not be limited to this expression form, and may be in various forms. .
- the correct answer It is possible to express the average of the probabilities by Equation 1 and to estimate the examinee's ability 0 using Equation 2.
- the product of the binomial distribution which is the sum of the correctness of the times, and the assumed capability distribution is calculated based on the assumption of the capability distribution of the group on which the test was performed.
- the theoretical distribution function of the partial score is calculated by integrating in the dimension of ability, and the obtained theoretical distribution function and the empirical distribution function of the partial score of the actual data best match.
- the test server It is also possible to include voice data as the transmitted and stored answer. In this case, it becomes possible to set up a listening question on the first computer, and also to set up a speaking problem in which the content of the actual test taker's utterance is targeted.
- the present invention can also be realized as a method for controlling the above-described test system. Further, the present invention may exist as a computer-readable storage medium itself storing a computer program for implementing such a test / system control method. Furthermore, it can also exist as a computer program that executes such a test / system control method.
- FIG. 1 is an outline of an example of a test system according to the present invention.
- FIG. 2 is an outline of a test taker unit constituting the test system according to the present invention.
- FIG. 3 is an outline of a grader unit constituting the test system according to the present invention.
- FIG. 4 is a flowchart outlining a test using the system according to the present invention.In particular, the process of taking an examination and scoring for writing and speaking related to the partial score on which the present invention is based is shown. ing.
- FIG. 5 is a graph showing the results of a score stability confirmation survey conducted by 12 subjects performed to confirm the effectiveness of ability estimation using the test system according to the present invention.
- FIG. 6 is a graph composed of FIGS. 6a to 6g, each showing the scores of 12 subjects in the score stability confirmation survey of FIG.
- Fig. 7 is composed of Fig. 7a to Fig. 7d, and when estimating the number of repetitions s '', when the true s-5, 10, 20, 40, 40, the estimated empirical distribution and Distribution relation with theoretical distribution The maximum value of the number difference (statistic of the Korgomolov-Smirnov test) is plotted for 3 to 10 repetitions.
- Fig. 8 is composed of Fig. 8a and Fig. 8b, each of which is an example of application to the estimation of the number of repetitions s "in the English proficiency test. BEST MODE FOR CARRYING OUT THE INVENTION
- Binary means that the answer takes only two values: correct or incorrect.
- the probability that the examinee answers the question correctly is expressed using parameters that represent the examinee's ability and parameters that characterize the question.
- a two-parameter (parameter) logistic 'model is used in which each problem is characterized by two parameters (discriminating power a and difficulty b).
- a candidate i having ability 0 has The probability of answering question j can be written as
- x is a dummy variable that is 1 if the candidate i answers the question j correctly, and 0 if he answers incorrectly.
- D is a constant.
- L B ( ⁇ ) of the examinee's ability 0 at the end of the n questions can be written as follows.
- P (0) is the probability of a correct answer on the right side of Equation 3
- Q (0) is the probability of a wrong answer, that is, one-one P (0).
- a maximum likelihood estimation method in which a value of 0 that gives the maximum value of the likelihood L B ( ⁇ ) in Equation 4 is an estimated value of the examinee's ability parameter is known and widely used.
- the right side of Equation 4 is written in the form of a product and it is not easy to find the maximum value, take the natural logarithm of both sides to consider in terms of the sum. And determine the maximum value of the log likelihood I n (L B ( ⁇ ) ) are common. This is because the natural logarithm is a monotonically increasing function, and 0, which gives the maximum value of the likelihood L B ( ⁇ ), and 0, which gives the maximum value of its natural logarithm In (L B ( ⁇ )) Because.
- the evaluation of the response (answer) to the problem is not limited to the two values of true and false, It is possible to be evaluated as.
- the dummy variable X in Equation 4 is a binary value of 1 and 0, but also three or more values from 0 to 1 (for example, 0, 0.2, 0.4, 0. 6, 0.8, and 1). If the partial score of test taker i for question j is r, ”, the likelihood corresponding to the partial score can be expressed as follows.
- Equation 5 To interpret the meaning of Equation 5, suppose that s j questions having the same question parameters are set for the same examinee. In the case of the academic test, it is reasonable to assume that the parameters are the same but present different problems. Questionnaires used for personality tests, etc. may present the same problem in terms of content, but this may violate the assumption of local independence, which is the premise of item response theory. Here, it is assumed that question items with the same parameters but different contents are presented.
- Equation 7 Since L B (theta) and L B ( ⁇ ) * gives the maximum value of 0 is the same, the maximum likelihood estimate is the same in. Equation 6 and Equation 7. In Equation 7
- L part ( ⁇ ) and L B * ( ⁇ ) are formally the same.
- the solution of the partial score model L part ( ⁇ ) on which the present invention relies and the L by the general item reaction theory The solution to B ( ⁇ ) is consistent through and L B * ( ⁇ ).
- Equation 8 it is derived that if the number of presentations s is increased, any partial score from 0 to 1 can be expressed. Note that it is actually inconvenient if there is a difference of the s-th root between Equations 6 and 7, so Equation 5 is raised to the s power and its natural logarithm is log likelihood of partial scoring as follows: It is preferred that
- Equation 10 Taking the sum of the problem groups in Equation 10 yields:
- zone is the second term on the right hand side is a sum related factor f 2 common problem group ", from the orthogonality of the assumption of factors, the unique part no correlation with other problems You can see.
- the second term on the right corresponds to this because the item response theory also assumes components specific to the item. In other words, it is not necessary to mention the local independence assumption.
- the factor f is the ability parameter 0 of the item response theory, there is no inconvenience in taking the sum of the interrelated problem groups in the test and processing it as a partial score. Performing the processing proposed by the present invention for a problem that is locally dependent may even be desirable from the assumption of the item response theory.
- Equation 14 [-P) ln (lF) -AP k - ⁇ , ⁇ ( ⁇ - ⁇ ) holds. Substituting Equations 17 and 18 into Equation 14
- Equation 2 5 Holds. If Equation 25 is satisfied for all the similar items constituting the likelihood, it is considered that the maximum likelihood solution of the partial score and the solution by the binary data approximately match.
- the partial score model in the present invention has already been shown to have a correspondence with a normal binary item response model if the number of times of repetitive definition is the same for a problem with the same parameter or similar parameter.
- the number of repetitions of all questions is the same, that is, the number of stages of partial scores is not always the same for all problems.
- Equation 9 it is necessary to extend Equation 9 as follows.
- the partial score is a graded grade such as a questionnaire.
- the number of steps is m + 1
- the number of repetitions s can be estimated to be m.
- the question item is “1.
- Well-applied” ⁇ 2.
- Slightly applicable. ⁇ 3.
- Not applicable at all ⁇ ⁇ ⁇ 4 ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ . Therefore, when each answer is c, it is converted to (1-1) / 3, and it is analyzed as partial score data of four stages of "0,1 / 3,2 / 3,1".
- the partial score is not the average value of multiple correct / false binary questions, but the scoring result such as the grader's rating.
- the number of stages is large.
- the partial score r when the partial score r is in the m + l stage, it is necessary to repeat The number of times s must be m. If the number of steps is small, the estimation in the previous section is considered to be good, but if the number of steps is large, the possibility of problems will increase. For example, if a scorer gives a score of 100 on a scale of 100 on a scale of 100 out of 100, and the answer is 1 on a scale of 1 to 10, the repetition rate is 10.
- the scorer gives a score of 69 to the answer of the remaining one person, it becomes difficult to make a partial score unless the number of repetitions is 101 as soon as possible.
- the score is not limited to an integer but also a real number, it is difficult to estimate the number of repetitions by such a concept.
- ⁇ (r) is the relative cumulative frequency of the distribution function up to stage r in the theoretical distribution
- F '(r) is the relative cumulative frequency up to stage r in the empirical distribution
- Equation 29 is obtained, and as a result, Equation 30 is obtained: R is 0, l Z m, 2 m.
- the minimum number of stages m that gives the Kolmogorov statistic of the Smirnov test can be used as an estimate of the number of repetitions, and by applying the following simulation and applying it to actual data, The effectiveness of this method has been confirmed.
- the method for estimating the number of repetitions in the previous section requires that an item parameter be given. Therefore, based on the partial score data created, the item parameter and ability parameter were estimated simultaneously (the number of repetitions at this stage is 1).
- the partial score model used in the computer adaptive test design and processing system according to the present invention has been described above.
- This partial score model was obtained by modifying the binary item response model. Therefore, in this partial score model, the number of parameters to be estimated is the same as in the case of the binary model, except when there is a special interest in the characteristics of the problem, when designing the academic test and processing the results. It can be said that there is little need to use a complex multi-valued model that has been attempted in the past.
- the partial score model used in the present invention is compared with a conventionally known step response model or the like, the following becomes clear.
- Item response theory usually requires that data be binary and one-dimensional, but partial score models can be applied to multi-valued and multi-dimensional data.
- the partial score model is a simple model (compared to the step response model ⁇ other multi-value models) and is easy for the user to understand.
- the partial score model is seamless with the commonly used two-parameter logistic model (unlike the step response model and other multi-value models), so the results can be easily interpreted, It is also useful for analyzing mixed data.
- the partial score model has fewer parameters than the other models (such as the step response model) and does not have a problem in estimation.
- the partial score model has a wide range of applications because any answer (answer) result can be applied by converting it to a partial score from 0 to 1.
- the partial score model can be easily applied to questionnaire data as well as tests,
- one of the inventors of the present invention performed a simulation using a partial score model. According to the results, (1) the step response model identified that the test was a small number of items. Bias is applied to force estimation, but this phenomenon does not occur in the partial score model. (2) The rank correlation between the score of the number of correct answers and the ability estimation value is higher in the partial score model than in the step response model (the correlation with the true value is almost the same in both models).
- the total score is divided by the sum of the number of items to obtain input data as a partial score between 0 and 1, that is, whether the likelihood of the binary model is correct or not.
- the dummy variable used for is treated as a substantive variable representing a partial score or as a weight of the true / false probability.
- Equation 5 when scoring the examinee's answer to an essay question, a partial score of 25% from zero (0) to full score (1) can be given.
- r is treated as a likelihood function that can take five values: 0, 0 ⁇ 25, 0.50, 0.75, 1 .
- the item parameters a (discriminating power) and b (difficulty) included in Equation 3 that defines P have already been estimated using data in pretests performed in advance using the same problem.
- the item parameters a and b included in the logistic curve P are estimated in advance for each problem.
- the ability 0 of the examinee is estimated by the maximum likelihood estimation method and the Bayes estimation method using the log likelihood of Equation 32. These methods themselves are generally known statistical techniques and are not features of the present invention. However, no matter which estimation method is used, the necessary log likelihood can be obtained for the first time by the partial score model which is the core of the present invention.
- the partial score model is different from such a simple example, by setting multiple evaluation criteria and adopting an analytical evaluation method that evaluates one question from multiple viewpoints. Capacity estimation Accuracy can be improved. For example, there is a possibility that it is possible to clarify the test subject's ability difference that does not become apparent in the basic application example described above. For example, in a writing problem, (a) Goal Achievement, (b) Grammar (Grammar), (c)
- the purpose of "communication of intention to arrange accommodation to the other party” is set for each problem, and the set purpose is "achieved or achieved. No "(1 or 0).
- a score between 0 and 1 that allows partial scores in 25% increments shall be given.
- five values of 0, 0.25, 0.5, 0.75, and 1 are assigned to in Expression 5.
- the item parameters a (discriminating power) and b (difficulty) included in Equation 3 that defines the oralistic curve P included in the likelihood function are the same as those in the above-described example.
- the evaluation items listed here are merely examples. In the test system according to the present invention, it is also possible to evaluate from another viewpoint.
- the above is a description of how the item response theory including the partial score model is applied to the English proficiency test and the ability of the examinee is estimated.
- the present invention is based on the item response theory including the partial score model.
- This is a test system and a test method for realizing capability estimation by using a general personal computer in an Internet connection environment.
- the operation of the test system according to the present invention will be outlined with reference to the accompanying drawings.
- FIG. 1 shows an outline of a first embodiment of a test system according to the present invention.
- Candidates are personal computers in an Internet-connected environment, such as at a language school that conducts tests (eg, English proficiency tests) designed, performed, and processed by the system according to the present invention. Take the test using 101. If the candidates are properly authenticated, they can take the test at home.
- the results of the answers entered by the examinee into the examinee unit 101 which is a personal computer via a keyboard, mouse, microphone, etc., are sent to the grader unit 1 via a network 103 such as the Internet. It is sent to 0 2 and, for example, a grader whose native language is English to be tested performs scoring while allowing partial scores.
- the test management server 104 has a problem database 105.
- the problem database 105 stores a group of problems which are implemented as a pretest and in which the item parameters (the discriminating power a and the difficulty b in Equation 3) are estimated in advance.
- the test management server 104 selects a group of questions from the question database 105 and transmits the selected group of questions to the examinee unit 101.
- FIG. 2 illustrates the outline of the test taker unit 101.
- Candidate 101 is usually a general personal computer with an Internet connection. One night.
- the input device 207 is a mechanical input device such as a keyboard, a mouse, and a touch panel
- the voice input / output device 209 is a microphone, a speaker, or the like.
- the examinee first inputs his / her own ID manually from the input device 207 or by voice from the voice input / output device 209, and instructs the start of the test.
- the examinee's ID that is uniquely issued to each examinee from the test management server 104 when the examinee registers to apply for an examinee is used.
- a password is issued along with an ID for security management.
- the test management server 104 recognizes the fact and gives an appropriate question. In response to the instruction, it is transmitted from the test management server 104 and displayed on the display 208, or selected according to its own level output from the audio input / output device 209 including the speaker.
- the examinee inputs the answer to the question via the input device 207 or the microphone (voice input / output device 209).
- the answer especially the answer to writing / speaking questions that require scoring that allows partial score, is sent to the grader unit via the communication interface 202 and the network 103 such as the Internet. Sent to 102.
- the answers are not sent directly from the examinee's unit 101 to the grader's unit 102, but are scored in real time. After being sent to the test management server 104 to be evaluated once, it is generally sent to the grader unit 102 judged to be appropriate from among a plurality of testers. This is not surprising from the scoring economy, where it is efficient to score after a certain number of answers have been collected.
- FIG 3 illustrates the outline of the grader unit 102.
- the grader unit 102 like the examinee unit 101, is usually a general personal computer having an Internet connection environment. Answer results sent from the examinee's unit 101 or the test management server 104 via the network 103 such as the Internet are displayed on the display 108, or a speaker (sound input / output device) is displayed. 309) and scored using an input device 407 such as a keyboard or mouse. The scoring result is returned to the test management server 104 via the network 103 such as the Internet.
- the examinee's unit, the grader's unit, and the test management server communicate over a communication network using a communication line such as the Internet. Was configured as a terminal.
- the test system according to the present invention can be realized as a second embodiment using a stand-alone personal computer having no communication function.
- a database storing a number of questions whose difficulty and discrimination power have been estimated in advance is built in a storage device such as a hard disk of the personal convenience, and the examinee can use, for example, a CD or the like.
- Answering via a keyboard and a microphone the questions of writing and speaking that are set according to instructions included in a program for performing the test of the present invention provided in a form stored in a DVD or the like.
- the answer results are temporarily stored in a hard disk or the like, and the grader reads the answer results from the hard disk and performs scoring allowing partial scores.
- the method of processing partial scores in the case of the second embodiment is the same as in the case of the first embodiment.
- the candidate's ability is estimated using a likelihood function based on the partial score model.
- FIG. 4 is a flow chart showing the outline of the test execution using the test system according to the present invention.
- the lighting using the partial score model on which the present invention is based is used.
- the implementation and treatment process of the test on speaking In the writing problem, test takers generally use a keyboard (input device 207 in Fig. 2) to answer in the form of typing in sentences such as English sentences.
- the examinee uses a microphone (a voice input / output device 309 in FIG. 3) provided at the personal convenience room to respond to the presented question or to make free speech. Is input, and the content spoken as the voice is the target of the evaluation.
- the grader waits in front of the grader unit, another terminal connected to the personal computer currently used by the examinee via a network such as the Internet, and scores in real time. Although it is possible to do so, in practice, it is common practice that the examinee's answers are stored in the test management server and then sent to the grader unit to collectively score a large number of answers. It is. First, the examinee accesses the designated web page on the Internet at the examinee unit 101. On that web page,
- the test management server 104 selects a writing or speaking problem from the problem database 105 (step 401).
- select the question that includes the evaluation item that has the most appropriate discriminating ability and difficulty in relation to the ability 0 estimated from the results of scoring the multiple-choice questions by the candidate. can do.
- writing skills have a correlation with reading abilities, and speaking abilities have a correlation with listening abilities.
- problem selection is merely an example, and is not an essential part of the test system based on the item response theory including the partial score model according to the present invention.
- the item parameters included in the oral sticky curve corresponding to the selected question are determined in advance for each evaluation item from the data in the pretest conducted earlier.
- the test system according to the present invention does not exclude the possibility of simultaneous maximum likelihood estimation.
- the item parameters of the problem stored in the problem database 105 coexist both when they are already estimated and when they are not.
- Estimation of item parameters is performed based on the partial score model used by the likelihood function of Equation 5 as in the estimation of ability 0.
- a process called equalization that standardizes the discriminating power and difficulty of each problem is also performed. This equalization process allows for absolute evaluation that is independent of the population of the candidate.
- the equalization itself is valid for the item reaction theory in general, and is not a feature of the present invention.
- the selected question is sent to the examinee unit 1 via a network 103 such as the Internet. 0 1 is transmitted (step 4 02). If the question is sent in text format, the question is given to the examinee on the display 208, and if the question is audio format, the speaker (speech input / output device 209) gives the subject a question (step 40). 3). The examinee gives an answer to the question in the form of typing in a sentence or in the form of a spoken voice (step 404). The document or voice file that constitutes the answer is transmitted to the test management server 104 via the network 103 such as the Internet, and is temporarily stored (step 405).
- the above process is repeated for a fixed number of examinees, and a fixed number of answer files are stored in the test management server 104.
- the questions that are given to these multiple candidates are not necessarily the same. This is because, from the general theory of item response theory, ability 0 can be appropriately estimated even if the questions to be asked are different.
- the grader unit 102 accesses the web page opened by the test management server 104, and sends the answers stored in the test management server 104 for scoring. When a request is made, a certain number of answer files are sent to the grader unit 102 (step 406).
- the grader grades the answer in a manner that allows the partial score already described (step 407), and returns the grading result to the test management server 104 (step 408).
- whether a plurality of answers are scored collectively or in real time is not related to the feature of the present invention.
- the test management server 104 substitutes the partial score received from the grader unit 102 into the likelihood function of Expression 5, and estimates the ability 0 (Step 409).
- item parameters may be estimated at the same time. Estimation methods include maximum likelihood estimation and Bayesian estimation.
- the estimation is completed, if necessary, the estimated value of 0 is converted to a score suitable for comparison with another test (step 410).
- the problem of allowing partial scores which was difficult to process with conventional general item reaction theory, has been solved. Even for tests that include the same, it is possible to perform the same capacity estimation as in the conventional item response theory.
- the inventors set the score stabilization by 12 subjects on May 21 to 28, 2003. A gender identity check was conducted. As a method, the same subject was asked to take an English proficiency test using the test system according to the present invention three times in succession, and it was confirmed whether or not a large fluctuation occurred in the score.
- the English proficiency test conducted here consisted of four skills tests: listening, reading, writing, and speaking.
- the subjects were 12 university students from a certain university in Tokyo who are relatively good at English. If the test according to the present invention ⁇ The English proficiency evaluation by the system is appropriate and the parameter estimation of each problem including equalization is properly performed in the test system according to the present invention. For example, if the same candidate took the test three times a day, their English proficiency would not change during that time, and the resulting score should not change significantly.
- the absolute evaluation is not affected by the ability level of the population. Turned out to be possible. Assuming that the partial score is expressed as an average of correctness when multiple items having the same parameter are repeated, theoretically, rij in Equation 5 is considered as a substantial variable.
- the estimation result is the same as that of the likelihood function used for the binary evaluation in the conventional item reaction theory.
- the experimental results shown in Figs. 5 and 6 show that this theoretical result was confirmed experimentally.
- test system of the present invention it is possible to achieve higher-precision capability estimation than before, while maintaining consistency with the conventional item response theory. This is a remarkable effect of the present invention.
- Equation 3 5 p u (e)> o is the step response model.
- Equation 3 3 is the boundary response curve that determines between the steps
- Equation 3 4 is the step response A step response curve expressing the reaction probability is obtained. As long as the above condition is satisfied, any function in Equation 33 can be used freely.
- a is a parameter common to all curves in Equation 38 and is called discriminative power.
- b u is a parameter related to the threshold value of each stage and is called difficulty.
- the step response model has one discriminative power and m ⁇ 1 difficulty parameter corresponding to the threshold value of each step for each item.
- i is the subject
- 0 is the parameter representing its characteristic value
- j is the item
- s is the number of repetitions of binary item conversion
- P is the two-parameter logistic model
- Q 1-P.
- the partial score model is based on the two-parameter mouth dystic model. Also, it is assumed that items having the same or similar item parameters are potentially repeatedly performed on subjects. In this case, r can be considered as a true / false average of repeated execution. It can be proved that the maximum likelihood solution of such a partial score model and the two-parameter mouth-distic model considered for repetition are the same (approximate for similar parameters) (Fujimori, 2002a).
- the characteristics of the step response model are as follows. (1) It is famous as a model corresponding to multi-valued data in item response theory. (2) It has been more than 30 years since its publication, and applied research has been reported (for example, Noguchi (1) 9 9 9) etc.). (3) There is a publicly available analysis software MULTI LOG.
- the step response model has a better fit to the data than the partial score model because the number of model parameters is large.
- the number of model parameters is large, there is a risk that problems arise, such as the need for a large amount of data for accurate estimation of the parameters.
- the partial score model is only simple, and the fit is expected to be inferior, but the stability of the estimated values and the like is considered to be good. I will need it.
- the step response model (and most of the multivalued models proposed so far) has the problem of inflexibility in changing the steps because of the model parameter at each step.
- the teacher assessed the items that were scored out of 20 at the time of the proficiency test in five coarse scales of 0, 5, 10, 15, and 20, the analysis using the five-step response model Become.
- you decide to deduct one point for an answer that is a typographical error or the like there is a problem that not only the value of the parameter of the model but also the number of the parameter itself changes immediately. If a questionnaire was assigned a rating of 5 on a 4-point scale, the graded response model would not be able to use the previous parameter values as it was.
- the simulation data based on the partial score model was created as follows. First, a 2-parameter mouthstick model is assumed as a component of the partial score model.
- the t- discriminating parameter that determines the distribution type of the parameter of this two-parameter mouth dystic model as follows is an average of 0.65, a standard deviation of 0.25, a lower limit of 0.3, and an upper limit of 2.0. It is assumed that the cut normal distribution and the difficulty parameter follow a normal distribution with mean 0 and standard deviation 0.5. Assume that the capacity parameter 0 follows a normal distribution with mean 0 and standard deviation 1.0.
- the ability parameter 0 is created according to the standard normal distribution, and the probability of correct answer expected from the two-parameter mouth dystic model is compared with a uniform random number in the range 0 to 0. False answer 0
- This binary data pattern according to the two-parameter logistic model was repeatedly created 10 times for each of 50,000 subjects and 200 items (data 1 to 10). However, the same parameter is used for each of the five items. Next, the average of the sum of the correctness and error for each of the 5 items of the same parameter of this data is taken, and the values are taken in five stages of 0, 0.2, 0, 4, 0.6, 0.8, and 1.0. Partial score data was used.
- the number of subjects is 500 as in the case of the binary data, but the number of items is 40.
- data for cross-validation was created by adding a new capacity parameter 0 of the subject for 500 people.
- the data based on the step response model was created as follows.
- the step response model also assumes a two-parameter logistic model as a component.
- the distribution form of the parameter is the same as in Section 0.
- the number of data steps is assumed to be 5 from 1 to 5. Therefore, four boundary response curves between stages are required from the model.
- one discriminating parameter is first generated according to the distribution, and this is used as the discriminating power common to each boundary reaction curve.
- four difficulty parameters are created, and the smallest one is selected as the difficulty of the boundary reaction curves of stages 1 and 2. In the same way, the difficulty level of each boundary response curve is determined in order from the one with the smallest degree of difficulty. The difference between these boundary response curves is taken as each step response curve.
- One capacity parameter 0 according to the standard normal distribution is created, and this value is fixed, and the interval of the reaction probability expected in each step response curve (the sum of the magnitudes of all the step response curves when 0 is fixed is 1 ), It is assumed that the reaction occurred when a uniform random number from 0 to 1 was entered.
- the above process was used as data for repetition parameter estimation for 500 people.
- data for cross validation was created for 500 people using the item parameters determined above.
- the estimation of the parameters was based on a home-built FORTRAN program based on the alternate simultaneous maximum likelihood estimation of the item parameter and the latent characteristic value 0.
- the maximum likelihood estimation of the number is possible, since the estimation program of the step response model supports only the alternating simultaneous maximum likelihood estimation, both models use the maximum likelihood estimation in consideration of the convenience of comparison. (The results are omitted, but there is no large difference between the marginal likelihood estimation and the alternate simultaneous estimation for the partial score model).
- the estimated value of 0 and the degree of difficulty are set in the range of -3.5 to 3.5, and the range of the estimated value of discriminative power is set to 0.02 to 2.0.
- Table 1 shows the correlation between the true value of the ability parameter of the simulation data created by the partial score model (hereinafter referred to as partial score data), the score of the number of correct answers, and the estimated value of 0 estimated by both models.
- partial score data For correlation, Kendall's rank correlation coefficient is calculated (hereinafter, unless otherwise specified, correlation refers to Kendall's rank correlation).
- rank correlation was calculated instead of Pearson's product moment correlation, which is commonly used, is that in many cases where the actual number of correct answers is scored and the order inversion of estimated values becomes a problem in the actual operation of item response theory.
- the correlation with true 0 is higher in the partial score model, but there is little difference from the step response model.
- the correlation with the number of correct answers slightly increases, and the estimated value by the partial score model gives a high correlation.
- the mean square error (MSE) of the discriminating power is 0.014 for Dataset 1 and 0.017 for the difficulty level.
- MSE mean square error
- Table 3 shows the results of the simulation data created by the step response model (hereinafter referred to as “step reaction data”).
- Table 4 shows the results of applying the item parameters estimated based on the data to the cross-validation data.
- the step response model has a slightly higher correlation with true 0 than the partial score model, but it is not a large difference and the case is reversed as in datasets 1 and 5.
- the overall correlation with the true value is slightly lower than in Tables 1 and 2
- the step response model is a model that is more difficult to reproduce than the partial score model. It can be said that the correlation with the number of correct answers is higher in the partial score model than in the step response model, even though it is step response data.
- the difference is larger than in the case of partial score data.
- the reversal of the order relationship with the total score is more prevalent in the step-response model.
- the reason for this reversal is that in the step-response model, the step-response curve in the middle part of the rating is relatively low depending on the item. The reason is that, depending on the response results of other items, the response of the item can hardly have an influence on the estimation of 0 (regardless of the estimation error, but on the determination of the point estimation value). Since this phenomenon is common not only to the step response model but also to other multi-valued models that model the response probability curve for each option or category, care must be taken when using these models. Conceivable. Of course, since the partial score model is also based on the 2-parameter mouth dystick model, there is some inversion of the total score and 0, but the degree is kept low.
- the mean square error (MSE) of the discriminating power is 0.2993, and for the difficulty level, it is 0.0636, which is slightly higher than the partial score model.
- MSE mean square error
- Estimation accuracy is poor You can see that Items 8 and 15 shown in Table 5 are items with poor estimation results in Dataset 1, but both cannot be estimated when the true boundary response curves are too close to each other. I understand. In such a case, it is necessary to take measures such as treating the two boundary response curves as one. However, there are problems such as the criteria to be used and this is not performed in this study. In a way, it can be said that there is a problem with the step response model in the need for such measures.
- the MSE of the discrimination power of the partial score model is 0.0136, which gives a better estimate than the step response model.
- marginal maximum likelihood estimation still has the potential to improve performance, so we will not discuss it further here and will report it in another study.
- the discriminative power of the step response model is higher overall than the partial score model (Table 7).
- Table 7 A similar phenomenon occurs when the data is binarized and analyzed using the usual two-parameter logistic model. For example, in the case of a four-point rating, binarization is essentially a phenomenon that occurs because one replaces step two with four and step three.
- the “binary model” in Table 7 is the result of such data conversion. It can be seen that the discrimination power is slightly higher than the partial score model. The occurrence of this phenomenon can be better understood by comparing the average response results for each group of item 4 in Fig. 1 in the case of quaternary and binary values.
- the MSE of the alternate and simultaneous estimation has a discrimination power of 0.0094, a difficulty level of 0.0007, and a marginal In the likelihood estimation, the discriminative power was 0.003 and the difficulty was 0.0014 (Table 9).
- Table 9 shows the estimated values obtained by analyzing the questionnaire data using both models and using marginal maximum likelihood estimation. Similarly, Table 10 shows that the discriminative power is increased in the step response model even when the marginal maximum likelihood estimation is used, as in the simulation.
- the value of the correlation is close to the correlation between the two models obtained from the simulation results of the step response data, but this alone will not make it possible to determine that this data is occurring according to the step response model .
- the reason for this is that if the effects of multidimensionality are taken into account when creating simulation data, for example, the correlation between the true value and the estimated value as well as the number of correct answers (in the case of a questionnaire, the total score) decreases. This is because there are things that can be easily predicted.
- a simple partial score model may be more appropriate than a step response model that must be done. Of course, this is not the case if you are interested in the reactions of the individual steps.
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Educational Administration (AREA)
- Educational Technology (AREA)
- Tourism & Hospitality (AREA)
- Economics (AREA)
- Software Systems (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- General Business, Economics & Management (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computer Hardware Design (AREA)
- Marketing (AREA)
- General Engineering & Computer Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Electrically Operated Instructional Devices (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
- Emergency Protection Circuit Devices (AREA)
- Maintenance And Management Of Digital Transmission (AREA)
- Complex Calculations (AREA)
Abstract
Description
Claims
Priority Applications (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
AU2003266616A AU2003266616A1 (en) | 2002-09-25 | 2003-09-25 | Test system and control method thereof |
JP2004539533A JP3645901B2 (ja) | 2002-09-25 | 2003-09-25 | テスト・システム及びその制御方法 |
GB0507837A GB2409314A (en) | 2002-09-25 | 2003-09-25 | Test system and control method thereof |
US10/529,263 US7103508B2 (en) | 2002-09-25 | 2003-09-25 | Test system and control method |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2002279513 | 2002-09-25 | ||
JP2002-279513 | 2002-09-25 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2004029906A1 true WO2004029906A1 (ja) | 2004-04-08 |
Family
ID=32040459
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2003/012252 WO2004029906A1 (ja) | 2002-09-25 | 2003-09-25 | テスト・システム及びその制御方法 |
Country Status (7)
Country | Link |
---|---|
US (1) | US7103508B2 (ja) |
JP (1) | JP3645901B2 (ja) |
KR (1) | KR100747141B1 (ja) |
CN (1) | CN1578972A (ja) |
AU (1) | AU2003266616A1 (ja) |
GB (1) | GB2409314A (ja) |
WO (1) | WO2004029906A1 (ja) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2006215207A (ja) * | 2005-02-02 | 2006-08-17 | Wao Corporation | 情報処理装置およびプログラム |
JP2007279306A (ja) * | 2006-04-05 | 2007-10-25 | Nagaoka Univ Of Technology | 項目応答理論におけるパラメータ推定方法 |
JP2008242637A (ja) * | 2007-03-26 | 2008-10-09 | Kddi Corp | 能力推定システムおよび方法ならびにプログラムおよび記録媒体 |
TWI485668B (zh) * | 2013-01-24 | 2015-05-21 | Univ Fooyin | 拼字學習電腦程式、內儲程式之電腦可讀取記錄媒體及內儲程式之可攜式電子裝置 |
JPWO2015008501A1 (ja) * | 2013-07-16 | 2017-03-02 | 株式会社ベネッセコーポレーション | 携帯型情報処理装置、テスト支援システム及びテスト支援方法 |
JP2022117805A (ja) * | 2021-02-01 | 2022-08-12 | カシオ計算機株式会社 | 発音トレーニングプログラム及び端末装置 |
Families Citing this family (28)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP4872214B2 (ja) * | 2005-01-19 | 2012-02-08 | 富士ゼロックス株式会社 | 自動採点装置 |
US20070111182A1 (en) * | 2005-10-26 | 2007-05-17 | International Business Machines Corporation | Method and system for distributing answers |
US20080046232A1 (en) * | 2006-08-18 | 2008-02-21 | Jan Groppe | Method and System for E-tol English language test online |
US8376755B2 (en) * | 2008-05-09 | 2013-02-19 | Location Inc. Group Corporation | System for the normalization of school performance statistics |
JP5609193B2 (ja) * | 2010-03-19 | 2014-10-22 | 富士通株式会社 | 試験プログラム、試験装置、および試験方法 |
US8761658B2 (en) * | 2011-01-31 | 2014-06-24 | FastTrack Technologies Inc. | System and method for a computerized learning system |
US20120329029A1 (en) * | 2011-06-23 | 2012-12-27 | Rauta Mihai Catalin | Computer implemented teaching method and apparatus |
US20130157245A1 (en) * | 2011-12-15 | 2013-06-20 | Microsoft Corporation | Adaptively presenting content based on user knowledge |
KR101311036B1 (ko) * | 2012-02-09 | 2013-09-24 | 이언주 | 라쉬 모형과 순차적 확률비 검증법을 이용한 유비티 시스템 |
KR101616909B1 (ko) * | 2012-10-31 | 2016-04-29 | 에스케이텔레콤 주식회사 | 자동 채점 장치 및 방법 |
US20140295400A1 (en) * | 2013-03-27 | 2014-10-02 | Educational Testing Service | Systems and Methods for Assessing Conversation Aptitude |
JP6408467B2 (ja) * | 2013-07-19 | 2018-10-17 | 株式会社ベネッセコーポレーション | 情報処理装置、情報処理方法及びプログラム |
US20150161900A1 (en) * | 2013-12-06 | 2015-06-11 | Act, Inc. | Methods for improving test efficiency and accuracy in a computer adaptive test (cat) |
CN103942993B (zh) * | 2014-03-17 | 2016-05-18 | 深圳市承儒科技有限公司 | 一种基于irt的自适应在线测评系统及其方法 |
US20150279226A1 (en) * | 2014-03-27 | 2015-10-01 | MyCognition Limited | Adaptive cognitive skills assessment and training |
JP6247628B2 (ja) * | 2014-12-09 | 2017-12-13 | 株式会社日立製作所 | 学習管理システムおよび学習管理方法 |
CN104505089B (zh) * | 2014-12-17 | 2018-05-18 | 福建网龙计算机网络信息技术有限公司 | 口语纠错方法及设备 |
JP5904651B1 (ja) * | 2014-12-25 | 2016-04-13 | 学校法人産業能率大学 | 維持管理指標算出装置及び維持管理指標算出方法 |
US20160225278A1 (en) * | 2015-01-31 | 2016-08-04 | Usa Life Nutrition Llc | Method and apparatus for incentivization of learning |
CN107851398A (zh) * | 2015-04-03 | 2018-03-27 | 卡普兰股份有限公司 | 用于自适应评估和训练的系统及方法 |
EP3324254A1 (de) * | 2016-11-17 | 2018-05-23 | Siemens Aktiengesellschaft | Einrichtung und verfahren zur bestimmung der parameter einer regeleinrichtung |
CN106682768B (zh) * | 2016-12-08 | 2018-05-08 | 北京粉笔蓝天科技有限公司 | 一种答题分数的预测方法、系统、终端及服务器 |
CN108921434B (zh) * | 2018-07-04 | 2020-08-14 | 北京希子教育科技有限公司 | 一种通过人机交互完成用户能力预测的方法 |
WO2020065663A1 (en) * | 2018-09-25 | 2020-04-02 | Merittrac Services Pvt. Ltd | Methods and systems for partial credit model (pcm) scoring in classical test theory (ctt) |
KR102015075B1 (ko) * | 2018-10-16 | 2019-08-27 | (주)뤼이드 | 학습 효율을 기반으로 개인 맞춤형 교육 컨텐츠를 제공하기 위한 기계학습 방법, 장치 및 컴퓨터 프로그램 |
US11102530B2 (en) | 2019-08-26 | 2021-08-24 | Pluralsight Llc | Adaptive processing and content control system |
US11295059B2 (en) | 2019-08-26 | 2022-04-05 | Pluralsight Llc | Adaptive processing and content control system |
US20210343175A1 (en) * | 2020-05-04 | 2021-11-04 | Pearson Education, Inc. | Systems and methods for adaptive assessment |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2001056634A (ja) * | 1999-08-20 | 2001-02-27 | Toshiba Corp | 自動採点システム |
JP2002006734A (ja) * | 2000-06-26 | 2002-01-11 | Society For Testing English Proficiency Inc | コンピュータ適応型検定試験の方法及びシステム |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP3687785B2 (ja) * | 2001-08-15 | 2005-08-24 | 株式会社日本統計事務センター | 採点処理方法および採点処理システム |
-
2003
- 2003-09-25 KR KR1020047004513A patent/KR100747141B1/ko active IP Right Grant
- 2003-09-25 AU AU2003266616A patent/AU2003266616A1/en not_active Abandoned
- 2003-09-25 US US10/529,263 patent/US7103508B2/en not_active Expired - Fee Related
- 2003-09-25 GB GB0507837A patent/GB2409314A/en not_active Withdrawn
- 2003-09-25 WO PCT/JP2003/012252 patent/WO2004029906A1/ja active Application Filing
- 2003-09-25 JP JP2004539533A patent/JP3645901B2/ja not_active Expired - Lifetime
- 2003-09-25 CN CNA038013770A patent/CN1578972A/zh active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2001056634A (ja) * | 1999-08-20 | 2001-02-27 | Toshiba Corp | 自動採点システム |
JP2002006734A (ja) * | 2000-06-26 | 2002-01-11 | Society For Testing English Proficiency Inc | コンピュータ適応型検定試験の方法及びシステム |
Non-Patent Citations (1)
Title |
---|
GIICHIRO SUZUKI: "graphic tokeigaku 1st edition", 28 November 1997, JIKKYO SHUPPAN CO., LTD., ISBN: 4-407-02390-2, pages: 122 - 124, XP002986705 * |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2006215207A (ja) * | 2005-02-02 | 2006-08-17 | Wao Corporation | 情報処理装置およびプログラム |
JP4628121B2 (ja) * | 2005-02-02 | 2011-02-09 | 株式会社ワオ・コーポレーション | 情報処理装置およびプログラム |
JP2007279306A (ja) * | 2006-04-05 | 2007-10-25 | Nagaoka Univ Of Technology | 項目応答理論におけるパラメータ推定方法 |
JP2008242637A (ja) * | 2007-03-26 | 2008-10-09 | Kddi Corp | 能力推定システムおよび方法ならびにプログラムおよび記録媒体 |
TWI485668B (zh) * | 2013-01-24 | 2015-05-21 | Univ Fooyin | 拼字學習電腦程式、內儲程式之電腦可讀取記錄媒體及內儲程式之可攜式電子裝置 |
JPWO2015008501A1 (ja) * | 2013-07-16 | 2017-03-02 | 株式会社ベネッセコーポレーション | 携帯型情報処理装置、テスト支援システム及びテスト支援方法 |
JP2018124583A (ja) * | 2013-07-16 | 2018-08-09 | 株式会社ベネッセコーポレーション | 携帯型情報処理装置、テスト支援システム及びテスト支援方法 |
JP2019200439A (ja) * | 2013-07-16 | 2019-11-21 | 株式会社ベネッセコーポレーション | 携帯型情報処理装置、テスト支援システム及びテスト支援方法 |
JP2022117805A (ja) * | 2021-02-01 | 2022-08-12 | カシオ計算機株式会社 | 発音トレーニングプログラム及び端末装置 |
JP7371644B2 (ja) | 2021-02-01 | 2023-10-31 | カシオ計算機株式会社 | 発音トレーニングプログラム及び端末装置 |
Also Published As
Publication number | Publication date |
---|---|
JPWO2004029906A1 (ja) | 2006-01-26 |
US7103508B2 (en) | 2006-09-05 |
GB0507837D0 (en) | 2005-05-25 |
KR100747141B1 (ko) | 2007-08-07 |
GB2409314A (en) | 2005-06-22 |
KR20050042743A (ko) | 2005-05-10 |
AU2003266616A1 (en) | 2004-04-19 |
JP3645901B2 (ja) | 2005-05-11 |
CN1578972A (zh) | 2005-02-09 |
US20050256663A1 (en) | 2005-11-17 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP3645901B2 (ja) | テスト・システム及びその制御方法 | |
Beischel et al. | The gender/sex 3× 3: Measuring and categorizing gender/sex beyond binaries. | |
Atabek et al. | Pre-school and primary school pre-service teachers’ attitudes towards using technology in music education | |
Paus et al. | Learning through online peer discourse: Structural equation modeling points to the role of discourse activities in individual understanding | |
Ünaldı | Self and teacher assessment as predictors of proficiency levels of Turkish EFL learners | |
Holster et al. | Measuring and predicting graded reader difficulty | |
Shin et al. | Evaluating different standard-setting methods in an ESL placement testing context | |
Yang | Modeling the relationships between test-taking strategies and test performance on a graph-writing task: Implications for EAP | |
Gogolin et al. | Diagnosing students’ understanding of the nature of models | |
Gonzalez | The intricate relationship between measures of vocabulary size and lexical diversity as evidenced in non-native and native speaker academic compositions | |
Foorman et al. | Florida Assessments for Instruction in Reading, Aligned to the Language Arts Florida Standards, FAIR-FS, Grades 3 through 12. Technical Manual. | |
Grenier et al. | Learning literacy in Canada: Evidence from the international survey of reading skills | |
Fitori | Difficulties in teaching and learning English grammar in Libyan universities | |
Kim et al. | Learning without awareness by academic and nonacademic samples: an individual differences study | |
Estaji et al. | A study of Test-Taking strategies of Iranian IELTS repeaters: Any change in the strategy use? | |
Cartwright | Linking the British Columbia English examination to the OECD combined reading scale | |
Zaghlool et al. | Interrogating The Influence Of Automated Writing Evaluation Tools In Language Learning: E-Learners’ Perspectives | |
CN111915227B (zh) | 一种教学评价数据分析方法 | |
Duong Thi et al. | Cognitive Diagnostic Analyses of the Progress in International Reading Literacy Study (PIRLS) 2011 Results | |
Varughese | Language difficulties in mathematics courses for students from non-English speaking backgrounds in the transition from secondary to tertiary education | |
Golubski | Reassessing assessment: using Bayesian methods to examine the efficacy of multiple-select multiple-choice items | |
Gu | Eliciting and Deciphering Mathematics Teachers’ Knowledge in Statistical Thinking, Statistical Teaching, and Statistical Technology | |
Peter | Effects of English Language on Students’ Performance in Teaching and Learning of Mathematical Modelling at Junior Secondary School Level in Bauchi State, Nigeria | |
Liu | Investigating the relationships between a reading test and can-do statements of performance on reading tasks | |
Close | Developing a Technology-Enhanced Solution to Language Inequality in English-Based Mathematics Tests |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
WWE | Wipo information: entry into national phase |
Ref document number: 2004539533 Country of ref document: JP |
|
WWE | Wipo information: entry into national phase |
Ref document number: 1020047004513 Country of ref document: KR |
|
AK | Designated states |
Kind code of ref document: A1 Designated state(s): AE AG AL AM AT AU AZ BA BB BG BR BY BZ CA CH CN CO CR CU CZ DE DK DM DZ EC EE EG ES FI GB GD GE GH GM HR HU ID IL IN IS JP KE KG KP KR KZ LC LK LR LS LT LU LV MA MD MG MK MN MW MX MZ NI NO NZ OM PG PH PL PT RO RU SC SD SE SG SK SL SY TJ TM TN TR TT TZ UA UG US UZ VC VN YU ZA ZM ZW |
|
AL | Designated countries for regional patents |
Kind code of ref document: A1 Designated state(s): GH GM KE LS MW MZ SD SL SZ TZ UG ZM ZW AM AZ BY KG KZ MD RU TJ TM AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HU IE IT LU MC NL PT RO SE SI SK TR BF BJ CF CG CI CM GA GN GQ GW ML MR NE SN TD TG |
|
WWE | Wipo information: entry into national phase |
Ref document number: 20038013770 Country of ref document: CN |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
WWE | Wipo information: entry into national phase |
Ref document number: 10529263 Country of ref document: US |
|
ENP | Entry into the national phase |
Ref document number: 0507837 Country of ref document: GB Kind code of ref document: A Free format text: PCT FILING DATE = 20030925 |
|
WWE | Wipo information: entry into national phase |
Ref document number: 0507837.3 Country of ref document: GB |
|
122 | Ep: pct application non-entry in european phase |