JP2017207691A - Correct answer ratio prediction method of multiple-choice question - Google Patents

Abstract

PROBLEM TO BE SOLVED: To maintain a constant level of difficulty of a multiple-choice question.SOLUTION: A correct answer ratio prediction method of a multiple-choice question comprises: an input reception step S1 for receiving an input of plural choices forming a multiple-choice question; a prediction selection ratio acquisition step S2-S8 for, about each of one or more choices which are incorrect choices, out of received plural choices, reading selectivity when a same or similar choice, as an incorrect choice, is set as a question, and when one or more selectivities are read from the database, calculating a prediction selectivity based on the read one or more selectivities, and when selectivities are not read from the database, setting a prescribed value as a prediction selectivity; and a correct answer ratio calculation step S9 for, based on the calculated or set one or more prediction selectivity, calculating the correct answer ratio of the multiple-choice question.SELECTED DRAWING: Figure 4

Description

  The present invention relates to a method for predicting the correct answer rate of an alternative question.

  Various kinds of examination questions are used in various national examinations such as the judicial examination and driver's license examination. Specific examples include a selection problem that allows one or more correct choices to be selected from among multiple choices, a correct or incorrect problem that allows the user to select whether the question sentence is correct or incorrect, and a descriptive problem that allows the answer to be freely described. There is. Of these, the choice question is a choice question that the test taker knows in advance that there is only one correct choice, and there are two or more correct choices or the number of correct choices is not known to the test taker Including multiple choice questions.

  In addition, some of the above-mentioned various national tests and the like are required to be repeatedly performed regularly while keeping the difficulty of the test constant. Among these, the requirement of “keep the difficulty level of the exam constant” is set in order to appropriately determine the ability of the examinees and to maintain fairness among the examinees. It is not so easy to keep the test level, and various studies have been made to keep the test difficulty level at a certain level.

  For example, Patent Document 1 discloses an invention for objectively determining the quality of an alternative problem. In this method, problem evaluation is performed based on the modified ebel index calculated based on the difficulty and necessity of the test questions set by the questioner at the time of writing, and the correct answer rate of the test questions obtained after the test. An index is calculated. According to the description in Patent Document 1, this problem evaluation index reflects both the subjective problem evaluation of the questioner and the objective numerical value obtained after the test is performed. It is said that it is possible to objectively determine the quality of the problem.

  As described above, the same or similar problem is often repeatedly asked in a test that is periodically repeated. According to the method of Patent Document 1, it is possible to brush up the problem determined to be good while reusing the problem determined to be good as it is, so that the difficulty of the test is kept constant as the test is repeated. It becomes possible to adapt to the level of.

  Patent Document 2 discloses an invention for calculating a predicted value of a test question score. Calculating the predicted value of the score (correct answer rate) of the exam questions is an important factor in keeping the difficulty of the exam constant. However, since the invention described in Patent Document 2 calculates a predicted value based on the degree of understanding for each learner, it cannot be used to keep the difficulty level of the test at a certain level.

  In addition, Patent Documents 3 to 5 disclose inventions that automatically generate an alternative problem to be presented to a user who performs learning on a computer. Among these, Patent Document 3 discloses an invention that enables creation of a test question of a desired difficulty level by designating a creation condition when creating the test question. Patent Document 4 discloses an invention in which a test question is created by effectively using the previous answer results of many users, particularly erroneous answers. Patent Document 5 discloses an invention in which only a problem that a learner has answered incorrectly can be repeatedly presented.

  Patent Document 6 does not relate to the test problem, but relates to predicting the cause of the failure of the apparatus, the purchase / non-purchase of recommended products, the repayment / non-payment in the case of financing based on the past data. The invention is disclosed.

JP 2012-088494 A JP 05-281899 A JP 2010-128305 A JP 2009-086203 A JP 2005-134484 A JP 2012-194894 A

  By the way, the inventor of the present application has newly discovered a problem that it becomes difficult to maintain the difficulty level at a certain level when an alternative problem is included in the test problem. Hereinafter, this point will be described in detail.

  First, an example of a method for creating an alternative question in a national examination will be described. To create an alternative question, a database that stores a plurality of question sentence candidates and a plurality of option candidates prepared for each question sentence candidate is used. The test creator first selects one or more from a plurality of question sentence candidates, and selects a predetermined number (for example, four for a four-choice question) from a plurality of corresponding option candidates for each. At this time, it is also possible to add a new question sentence candidate or a new option candidate. Then, after making minor corrections to the question sentence or each option as necessary, the question is presented as a new question.

  The number of option candidates prepared for each question sentence candidate generally exceeds 20, for example. If 4 are selected from among them, the number of combinations of selection methods is enormous (for example, the number of combinations when selecting 4 from 20 is 4,845). As a result, since there are almost no combinations that have not been asked in the past, for example, as in Patent Document 1, it is not possible to align the difficulty of the test by determining the quality of each problem based on the past question results. ,Have difficulty.

  As another method, it is conceivable to predict the correct answer rate based on the past selection rate of correct answer options, and adjust the problem content so as to maintain a certain difficulty level using the result. Regarding the above-mentioned multiple selection problem and correct / incorrect problem, a certain degree of difficulty is often obtained by this method. That is, the variation in the selection rate of the correct answer choices is small in the multiple selection problem and the correct / incorrect question, and it is effective to obtain the predicted correct answer rate by a linear function of the integrated value of these selection rates. However, for alternative questions, the difficulty level other than the correct answer choices greatly affects the difficulty level of the entire problem, and it is difficult to obtain a certain level of difficulty even with this method.

  In addition, it is possible to use a general data analysis method as described in Patent Document 6, but it is suitable for the analysis of the alternative problem in the same manner as Patent Document 1 in that a large amount of learning data is required. In addition, it is not always clear whether an appropriate result can be obtained.

  Therefore, one of the objects of the present invention is to provide a method for predicting the correct answer rate of an alternative question that can keep the difficulty of the alternative problem at a certain level.

  The method for predicting the correct answer rate of an alternative question according to the present invention includes, for each of a plurality of option candidates, type information indicating whether the option is a correct answer option or an incorrect answer option when the same or similar option is presented in the past, A method for predicting the correct answer rate of an alternative question, which is executed by an information processing apparatus connected to a database that stores a selection rate when the same or similar option is presented in the past, and includes a plurality of alternative questions An input receiving step for accepting an input of an option, and a selection rate when an identical or similar option is given as an incorrect answer option for each of one or more options that are incorrect answer options among the plurality of options. When one or more selection rates are read from the database, a predicted selection rate is calculated based on the read one or more selection rates. On the other hand, when the selection rate is not read from the database, a prediction selection rate acquisition step of setting a predetermined value as the prediction selection rate, and one or more calculated or set in the prediction selection rate acquisition step A correct answer rate calculating step of calculating a correct answer rate of the alternative question based on the predicted selection rate.

  According to the present invention, the selection rate of correct answer options that may fluctuate greatly due to the influence of incorrect answer options, both when calculating the predicted selection rate and when calculating the correct answer rate of alternative questions Since the correct answer rate is calculated based only on the selection rate of erroneous answer options that are not used and have little fluctuation in the past data, it is possible to predict the correct answer rate of the alternative question to some extent correctly. Therefore, it becomes possible to keep the difficulty level of the alternative problem at a certain level.

It is a figure which shows the structure of the alternative problem correct answer rate prediction system 1 which implements the correct answer rate prediction method of the alternative problem by embodiment of this invention. It is a figure which shows an example of the relevant information of the alternative problem contained in the implemented problem data D1 shown in FIG. It is a figure which shows an example of the relevant information of the alternative problem contained in the newly created problem data D2 shown in FIG. It is a flowchart which shows the flow of the process which the correct answer rate prediction calculating device 4 shown in FIG. 1 performs. It is a figure which shows an example of the data (past data) memorize | stored in the past test data storage device 3 shown in FIG. 1, and an example of the new problem input into the correct answer rate prediction calculating device 4 shown in FIG. . The figure which shows an example of the data (past data) memorize | stored in the past test data memory | storage device 3 shown in FIG. 1, and the other example of the new problem input into the correct answer rate prediction calculating device 4 shown in FIG. It is. It is a figure which shows the example of calculation of the correct answer rate estimated value P about the same past data as what was shown in FIG.5 and FIG.6. It is a figure which shows the example of calculation of the correct answer rate predicted value P about the past data different from what was shown in FIG.5 and FIG.6.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. The configuration of the present invention is not limited to the embodiment described below.

  FIG. 1 is a diagram showing a configuration of an alternative question correct answer rate prediction system 1 that implements an alternative answer correct rate prediction method according to an embodiment of the present invention. As shown in the figure, the alternative question correct answer rate prediction system 1 includes a past test data addition device 2, a past test data storage device 3, and a correct answer rate prediction calculation device 4 (information processing device). The

  Each of the past test data addition device 2 and the correct answer rate prediction calculation device 4 is a functional unit realized by the central processing unit executing a computer program loaded in a cache memory in a computer such as a personal computer. The past test data storage device 3 is configured by a database configured using a database language such as SQL, for example. As hardware, these devices can be mixed in one hardware, and these devices can be mounted on different hardware.

  The past test data adding device 2 is a device that accepts input of the completed question data D1 via an input unit (not shown) (keyboard, mouse, communication port, etc.) and writes it in the past test data storage device 3. The executed problem data D1 is data relating to a problem that has actually been executed as a test, and includes one or more alternative questions and related information.

  FIG. 2 is a diagram illustrating an example of the related information of the alternative problem included in the completed problem data D1. As shown in the figure, the related information included in the completed problem data D1 includes a problem number, an execution time, and information for each option. The information for each option includes a classification number, type information, and a selection rate. Although FIG. 2 shows the case where there are four options, the choice problem that is the subject of the present invention is not limited to the one having four options. It can be suitably applied to one problem. In the following, the explanation will be continued focusing on the alternative problem having four options.

  The problem number is information indicating a group of options. Choice options for the correct answer rate prediction in the alternative question accuracy prediction system 1 are grouped in advance, and the question number belongs to which group the options included in the alternative question belong. Is shown. Multiple options included in a single choice problem always belong to the same group.

  The implementation date is date information indicating the date when the test including the alternative question was conducted. Here, the date information is used, but it is sufficient if the order of tests is known as will be described later. For example, it may be replaced by a serial number indicating the order of tests.

  The classification number is information indicating an option number in the group. Even if the choices included in the exam questions that are actually given are of the same content as the choices that were given in the past (including those that differ only in the numerical values in the text), (Including whether the choices are correct or incorrect) and the individuality of the exam question creator, etc. In the present invention, “similar options” refer to options that are substantially the same in content but different in expression. The classification number is information for collectively managing these similar options, and is determined by the test question creator when the test question is created.

  Although a category number is used here, similar options can be managed by a method that does not use a category number. For example, when the user of the past test data adding device 2 inputs the already-executed question data D1, any of the options already stored in the past test data storage device 3 is the option to be newly input. It is good also as designating whether it corresponds to.

  In addition, there are similarities between options expressed so as to be correct in content and options expressed so as to be erroneous in content even if they are “similar options” according to the above-mentioned criteria. It may be preferable to treat it as not (ie assign a different classification number). This is because there is a large difference in the selection rate even if the options expressed so as to be correct in content and the options expressed as incorrect in content are substantially the same content. Because there is. For example, in light of the provisions of the current Labor Standards Act (Article 34, Paragraph 1), the option “If you work over 8 hours, you must give a break for 1 hour” is correct in content, If the working hour exceeds 8 hours, you must give a break for 45 minutes. Since these options differ only in the numerical values in the text, it can be said that they are substantially the same content, and therefore, according to the above-mentioned criteria, they are “similar options”. Since there is a big difference, it is preferable to treat them as not similar.

  The type information is information indicating whether the option is the correct answer option or the incorrect answer option in the selected question, and is configured by one of information indicating “correct answer” and information indicating “incorrect answer”. The The selection rate is information indicating what percentage of the examinees have selected the option as a result of the actual selection question being asked.

  Returning to FIG. The correct answer rate prediction calculation device 4 accepts input of newly created question data D2 via input means (keyboard, mouse, communication port, etc.) not shown, and receives the received information and the stored contents of the past test data storage device 3. This is a device that predicts (calculates) the correct answer rate for each alternative question. The newly created question data D2 is data relating to a problem that has been newly created and has not yet been implemented as an actual test, and includes one or more alternative questions and related information, similar to the conducted question data D1. Composed. The correct answer rate calculated by the correct answer rate prediction calculation device 4 is provided as a correct answer rate predicted value P to the user via output means (display, communication port, etc.) not shown.

  FIG. 3 is a diagram illustrating an example of the related information of the alternative problem included in the newly created problem data D2. As shown in the figure, the related information included in the newly created problem data D2 includes a problem number and information for each option. The information for each option includes a classification number and type information. The detailed meanings of these numbers and information are as described above with respect to the executed problem data D1. In the following, when the term “new problem” is used, it refers to each alternative problem included in the newly created problem data D2 and related information.

  Hereinafter, the processing performed by the correct answer rate prediction calculation device 4 will be described in detail with reference to a flow chart and actual data examples.

  FIG. 4 is a flowchart showing a flow of processing performed by the correct answer rate prediction calculation device 4. FIG. 5 shows an example of data (past data) stored in the past test data storage device 3 and an example of a new question input to the correct answer rate prediction calculation device 4 for the option group of question number 1. FIG.

  In FIG. 5, the numbers 1 to 14 shown in the header of past data indicate that the larger the number is, the more the past data has been implemented. In the past data, the option with the numerical value of the selection rate is the option actually used in the corresponding test. A thick frame indicates an option used as a correct answer option.

  When the test question creator creates the question of question number 1, one correct answer option and three wrong answer options are selected from the options 1 to 22 shown in FIG. FIG. 5 shows an example in which the option 12 is selected as the correct answer option and the options 3, 8, and 15 are selected as the incorrect answer options. The test question writer then corrects the question texts and choices as necessary to complete the new question.

  Referring now to FIG. 4, the correct answer rate prediction calculation device 4 first takes in a new question completed by the test question creator as described above (step S1, input acceptance step). In addition to the question text and the contents of the options, the new problem includes a problem number indicating an option group, and a category number and a type number of each option, as described with reference to FIG.

  Next, the correct answer rate predicting / calculating device 4 has each of all wrong answer options (that is, n-1 choices excluding one correct answer choice among n choices (n = 4 in this case)) in each new question. , A loop process (predictive selectivity acquisition step) for acquiring the predictive selectivity is executed (step S2).

  In the loop process in step S2, first, a past question having an option that is the same as or similar to the attention incorrect answer option is extracted (step S3). Referring to the example of FIG. 5, the past problem extracted when the erroneous answer option of attention corresponds to option 3 is the past problem corresponding to the test execution order 1, 3, 5, 6, 10 It becomes. Similarly, when the attention incorrect answer option corresponds to option 8, past questions corresponding to the test execution orders 2, 3, and 13 are extracted, and the attention incorrect answer option corresponds to option 15 The past questions corresponding to the test execution orders 1 and 12 are extracted.

  Subsequently, the correct answer rate prediction calculation device 4 determines whether or not one or more past questions have been extracted in step S3 (step S4). As a result, if it is determined that it has been extracted, the process proceeds to step S5. If it is determined that it has not been extracted, the process proceeds to step S8.

  In step S5, the correct answer rate prediction calculation device 4 selects, from the extracted past questions, ones that have the same or similar options as the focused error answer options as the error answer options (step S5). Referring to the example of FIG. 5 again, first, regarding the case where the erroneous answer option of interest corresponds to option 3, the extracted past questions (corresponding to the test execution order 1, 3, 5, 6, 10). In all of (5), an option that is the same as or similar to option 3 is an erroneous answer option. Therefore, all the extracted five past problems are selected. Similarly, even when the erroneous answer option of interest corresponds to option 15, the same or similar option as option 15 is wrong in all of the extracted past questions (two corresponding to the test execution order 1 and 12). Since it is an answer option, all the extracted two past questions are selected. On the other hand, in the case where the erroneous answer option of interest corresponds to option 8, the same or similar option as option 8 among the extracted past questions (three corresponding to test execution order 2, 3, 13) Are the two answer options corresponding to the test execution order 2 and 3. Therefore, only the past questions corresponding to the test execution orders 2 and 3 are selected.

  It is one of the features of the present invention that only the wrong answer option is selected in step S5. That is, the selection rate of correct answer options varies greatly due to the influence of other options (incorrect answer options) and question sentences, and is not suitable for use in predicting the correct answer rate. By selecting only past questions whose corresponding options are incorrect answer options in step S5, it is possible to eliminate the influence of such correct answer options.

  In the present embodiment, step S3 and step S5 are explicitly described separately. For example, step S3 is described as “an option that has the same or similar option as the erroneous answer option of interest, and the option is an incorrect answer. Of course, step S3 and step S5 may be combined into one step by transforming such as “select past problem as option”.

  Subsequently, the correct answer rate prediction calculation device 4 determines whether or not one or more past questions have been selected in step S5 (step S6). As a result, if it is determined that it has been selected, the process proceeds to step S7. If it is determined that it has not been extracted, the process proceeds to step S8.

  In step S <b> 7, the correct answer rate prediction calculation device 4 calculates a predicted selection rate of the attention error answer option based on the selection rate of the option that is the same as or similar to the attention error answer option in the selected past problem. Specifically, the average value of the selection rates of options that are the same as or similar to the attention error answer option in the selected past question is calculated and set as the predicted selection rate of the attention error answer option. In FIG. 5, the calculation of the prediction selection rate based on the calculation of the average value is indicated by an arrow 1. For example, when the erroneous answer option of attention corresponds to option 3, selection of the same or similar option as option 3 in the selected past question (five corresponding to the test execution order 1, 3, 5, 6, 10) The average value of the rates ((6 + 8 + 6 + 6 + 6) / 5≈6) is the predicted selection rate of the attention incorrect answer option. Similarly, (21 + 23) / 2 = 22 when the focused erroneous answer option corresponds to option 8 and (7 + 12) / 2≈ when the focused erroneous answer option corresponds to option 15 10 is the predicted selection rate of each of the attention incorrect answer options.

  In this embodiment, a simple average of past selection rates is used as a predicted selection rate of attention incorrect answer options, but a value derived by another method can be used as a predicted selection rate of attention incorrect answer options. It is. For example, the selection rate of the option that is the same as or similar to the attention error answer option in the latest past problem among the past questions selected in step S5 may be used as the prediction selection rate of the attention error answer option. A weighted average calculated by weighting (for example, a weighted average calculated by weighting recently implemented one) may be used as the predicted selection rate of the erroneous answer choice. It is preferable that the specific calculation method of the predicted selection rate of the attention incorrect answer option is appropriately determined for each test question according to the nature of the test question.

  Step S8 is a step provided for the case where the past problem is not extracted or selected in steps S4 and S5. In this case, since the selection rate in the past problem cannot be referred to, it is necessary to determine the predicted selection rate of the attention incorrect answer option by another means. Specifically, as shown in FIG. 4, it is preferable to set a default value for the predicted selection rate of the attention error answer option. This default value may be a predetermined value, or it may be the average value of the selection rate when each option is presented as a wrong answer option for the first time, although it may not be said that the content is substantially the same. It is good also as calculating from the selection rate when the other choices which can be judged to be similar in light of other criteria are set as wrong answer choices.

When the loop process of step S2 ends, the correct answer rate prediction calculation device 4 calculates a predicted value of the correct answer rate of the new question based on the total value of the predicted selection rates calculated or set for each incorrect answer option of the new question ( Step S9, correct answer rate calculation step). The predicted value calculated in this way becomes the correct answer rate predicted value P shown in FIG. More specifically, the correct answer rate prediction calculation device 4 calculates the correct answer rate prediction value P by the following equation (1). However, SUM in the formula is a total value of predicted selection rates calculated or set for each wrong answer option of the new question.
P = 100-SUM (1)

The correct answer rate predicted value P can also be calculated by other methods. For example, the correct answer rate predicted value P may be calculated by the following equation (2). However, a and b in the equation are constants obtained by examining the correlation between the past correct answer rate predicted value and the correct answer rate obtained by actually performing the test after prediction. Note that the sum of the correct answer rate predicted value P obtained by the expression (2) and the total value SUM of predicted selection rates calculated or set for each incorrect answer option of the new question is not necessarily 100. Since the target of prediction by 4 is only the correct answer rate, it is not a problem.
P = a (100−SUM) −b (2)

  Moreover, as a correct answer rate predicted value P, a value obtained by multiplying a correct answer rate predicted value P obtained by the formula (1) or the formula (2) by a coefficient based on a past selection rate of correct answer options is output. Also good.

  FIG. 5 shows the correct answer rate prediction value P calculated according to the equation (1). First, as shown by the arrow 2, the total value SUM in the example of FIG. 5 is the total value 38 of the predicted selection rates 6, 22, and 10 of the options 3, 8, and 15 that are incorrect answer options of the new problem. . Then, the correct answer rate predicted value P is a value 62 obtained by substituting 38 into the equation (1), as indicated by an arrow 3.

  As described above, according to the method for predicting the correct answer rate of alternative questions according to the present embodiment, the selection rate of correct answer options that may vary greatly due to the influence of incorrect answer options is calculated when calculating the predicted selectivity. (Steps S2 to S8 in FIG. 4, arrow 1 in FIG. 5) and when calculating the percentage of correct answers for new questions (step S9 in FIG. 4, arrows 2 and 3 in FIG. 5). In addition, since the correct answer rate is calculated based only on the selection rate of erroneous answer options with little fluctuation in the past data, it is possible to predict the correct answer rate of a new question to some extent correctly. Therefore, it becomes possible to keep the difficulty level of a new problem at a certain level.

  Another example of calculating the correct answer rate predicted value P will be described. 6 shows an example of data (past data) stored in the past test data storage device 3 and another example of a new question input to the correct answer rate prediction calculation device 4 for the option group of question number 1. FIG. The past data is the same as that shown in FIG. On the other hand, the new question according to this example is configured to have the same or similar options as the options 2, 6, and 15 as wrong answer options and the same or similar options as the option 12 as correct answer options.

  In the example of FIG. 6, for option 2, the past questions corresponding to the test execution orders 9 and 13 in step S5 of FIG. 4 are selected (selection rates of options that are the same as or similar to option 2 are 28 and 25, respectively). The predicted selection rate calculated in step S7 is (28 + 25) / 2≈27. Similarly, with regard to option 15, the past questions corresponding to the test execution orders 1 and 12 (selection rates of options that are the same as or similar to option 15 are 7 and 12 respectively) are selected in step S5 of FIG. The calculated prediction selectivity is (7 + 12) / 2≈10. On the other hand, for option 6, since no past problem is selected in step S5 of FIG. 4, the default value is set to the predicted selection rate in step S8. In FIG. 6, 20 is used as a default value.

  After the predicted selection rate of each wrong answer option is obtained, the correct answer rate predicted value is obtained by substituting the total value 57 of the predicted selection rates 27, 20, and 10 of the options 2, 6, and 15 into the formula (1). P = 43 is obtained. Thus, also in the example of FIG. 6, it is possible to obtain the correct answer rate predicted value P without using the correct answer option selection rate.

  Hereinafter, the fact that the present invention can obtain a substantially correct correct answer rate prediction value will be described with reference to two examples in which the present invention is applied to actual test results.

  FIG. 7 is a diagram illustrating a calculation example of the correct answer rate predicted value P for the same past data as that illustrated in FIGS. 5 and 6. The figure shows an example in which the correct answer rate predicted value P is calculated according to the process shown in FIG. 4 only for the past problems corresponding to the test execution orders 1 to 8. For example, the correct answer rate predicted value P for the past problems corresponding to the test execution order 8 is obtained by using only the past problems corresponding to the test execution orders 9 to 14 performed in the past from the test execution order 8 as past data. It was obtained as a result of carrying out the processing shown in. The reason why the correct answer rate predicted value P is not calculated for the past questions in the test execution order 9 to 14 is that the past data is too small.

  FIG. 7 also shows numerically an error (prediction error) between the actually measured correct answer rate and the correct answer rate predicted value P, and shows changes in the actually measured correct answer rate and the correct answer rate predicted value P. FIG.

  As understood from the graph of FIG. 7, the actually measured correct answer rate and the correct answer rate predicted value P show substantially the same changes. This indicates that the correct answer rate is obtained by the present invention. Looking at the prediction error, the average of 8 prediction errors (average error) is 11, and it can be said that a good value is obtained.

  Next, FIG. 8 is a diagram illustrating an example of calculating the correct answer rate predicted value P for past data different from those illustrated in FIGS. 5 and 6. The past data in the figure includes 16 past questions (test execution order 1 to 16). Moreover, the figure has shown the example which calculated the correct answer rate estimated value P according to the process shown in FIG. 4 only about the past problem corresponding to test implementation order 1-8. Further, the prediction error and the graph are also illustrated in the same manner as in FIG.

  As can be understood from the graph of FIG. 8, in this example as well, the actually measured correct answer rate and the correct answer rate predicted value P show substantially the same changes. This indicates that the correct answer rate is obtained by the present invention. Looking at the prediction error, the average (average error) of the prediction errors for 8 times is 6, and it can be said that a good value is obtained.

  As described above with reference to FIGS. 7 and 8, when the present invention is applied to actual test results, a substantially correct correct answer rate prediction value can be obtained. Therefore, according to the present invention, it can be said that the correct answer rate of the alternative problem can be predicted to some extent correctly, and as a result, it can be said that the difficulty level of the alternative problem can be maintained at a certain level.

  As mentioned above, although preferable embodiment of this invention was described, this invention is not limited to such embodiment at all, and this invention can be implemented in various aspects in the range which does not deviate from the summary. Of course.

  For example, in the above-described embodiment, the present invention is configured as a method for predicting the correct answer rate of an alternative question. However, the present invention provides a correct answer option when the same or similar options are presented in the past for each of a plurality of option candidates. And an information processing apparatus connected to a database that stores type information indicating which of the answer options was wrong, and a selection rate when the same or similar option was asked in the past, and constitutes an alternative problem The selection rate when the same or similar option is given as an incorrect answer option for each of one or more options that are erroneous answer options among the multiple options When reading from the database and one or more selection rates are read from the database, a predicted selection rate is calculated based on the read one or more selection rates. On the other hand, when the selection rate is not read from the database, a prediction selection rate acquisition unit that sets a predetermined value as the prediction selection rate, and one or more calculated or set by the prediction selection rate acquisition unit It is also possible to configure as an information processing apparatus including a correct answer rate calculation unit that calculates the correct answer rate of the alternative question based on the predicted selection rate.

  In addition, the present invention provides, for each of a plurality of option candidates, type information indicating whether the option is a correct answer option or an incorrect answer option when the same or similar option is presented in the past, and the same or similar option in the past. A program for causing a computer to function as an information processing apparatus connected to a database that stores a selection rate when a question is presented, and an input receiving unit that receives input of a plurality of options constituting an alternative problem, For each of one or more options that are incorrect answer options among a plurality of options, the selection rate when the same or similar option is presented as an incorrect answer option is read from the database, and one or more selection rates are obtained from the database. When read, the prediction selectivity is calculated based on the read one or more selection rates, while the database A prediction selection rate acquisition unit that sets a predetermined value as the prediction selection rate when the selection rate is not read, and one or more prediction selection rates calculated or set by the prediction selection rate acquisition unit On the basis of the above, it is also possible to configure as a program that causes the computer to function as a correct answer rate calculation unit that calculates the correct answer rate of the alternative question.

1 alternative question correct answer rate prediction system 2 past test data addition device 3 past test data storage device 4 correct answer rate prediction calculation device D1 completed question data D2 newly created question data P correct answer rate prediction value

Claims (1)

For each of a plurality of option candidates, type information indicating whether the option was a correct answer option or an incorrect answer option when the same or similar option was given in the past, and when the same or similar option was given in the past A method for predicting a correct answer rate of an alternative question executed by an information processing apparatus connected to a database that stores a selectivity,
An input reception step for receiving input of a plurality of options constituting the alternative problem;
For each of one or more options that are incorrect answer options among the plurality of options, the selection rate when the same or similar option is presented as an incorrect answer option is read from the database, and the one or more selection ratios from the database Is read based on one or more of the read selection ratios, and when the selection ratio is not read from the database, a predetermined value is selected as the prediction selection. Predictive selection rate acquisition step to set as a rate,
A correct answer rate predicting method for an alternative question, comprising: a correct answer rate calculating step for calculating a correct answer rate of the alternative question based on one or more of the predictive select rates calculated or set in the predictive select rate acquiring step.

Images