JPH10254340A - Device for inspecting study and method therefor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve an efficiency of processing and reliability at the time of determining and answer inputted by hand writing as a character, symbol, etc., by recognizing non-coded information inputted as an answer and extracting a group of character candidates and selecting a hand written answer with a priority given to a recognition result within a range permitted by the correct answer. SOLUTION: What a learner answers is inputted as non-coded information. When the non-coded information inputted by this input processing is recognized and converted into characters expressively by coded information, a recognition result conforming to a fixed condition is selected as a group of character candidates. And an answer is outputted with a priority given to a recognition result within a permissible range by a correct answer to the extracted group of character candidates. In a category 'Reading of kanji', for example, a contracted sound which cannot compose 'Reading of Kanji' in Hiragana in set as a inhibition code 72, and characters corresponding to the inhibition code 72 are excluded, and non-excluded characters are determined as answer characters.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a learning test apparatus and a learning test method, and more particularly, to inputting a learner's answer to a problem for which a correct answer is prepared in advance, and determining the validity of the answer based on the correct answer. The present invention relates to a learning test device and a learning test method.

[0002]

2. Description of the Related Art In recent years, various learning and examining apparatuses using excellent processing functions of computers have been proposed. As a method of learning test of these devices, a predetermined problem is written
Generally, a method is used in which a response is output by voice or the like, an answer is requested from the answerer, and it is determined whether or not the input answer matches the correct answer to the question. The input of the answer in this case is performed by an input device such as a keyboard, a mouse, and a tablet, and the determination as to whether or not the input answer matches the correct answer for the previously prepared question is based on the correct answer prepared in advance. Is compared with the input information of the answer, and it is determined whether or not these two pieces of information match.

Recently, various proposals have been made to enable the user to input the answer to the question by hand using a tablet or the like. Since the person who is going to learn is not always proficient in using a computer or a keyboard or mouse attached thereto, being able to answer with handwritten characters is desirable in terms of expanding the range of learners. From infants who cannot handle the keyboard to adults who are not accustomed to keyboard operation, the learning test device can be used widely. In such a learning test apparatus, as a step before judging the coincidence between the answer and the correct answer, after recognizing the answer input by handwriting, it is determined as a character / symbol by referring to a dictionary prepared in the apparatus. It is necessary to do.

[0004]

However, since the individual differences of handwritten characters are large, it is quite difficult to recognize the answer input by handwriting and determine it as a character or a symbol, etc. May be recognized as a different character or symbol, and it takes time to correct the character, thereby reducing the learning efficiency. In addition, recognition requires a large amount of information to be referred to, and the time required for recognition is long, and it takes a considerable amount of time from when a learner answers a question until a test is performed to determine whether the answer is correct. I couldn't overlook the problem of taking time. For example, in the case of learning contents such as reading and writing of kanji, spelling of English, writing of map symbols, etc., prepare information of all commonly used kanji, kana, English words, map symbols, etc. in a dictionary etc. in advance. In addition, by referring to all the information stored in the dictionary, it is necessary to determine what kind of character, symbol, or the like corresponds to the answer input by handwriting. In such a case, in the conventional learning test apparatus, a large amount of information must be accessed in order to determine the answer input by handwriting as characters, symbols, etc., and it takes time to determine the answer. There was a problem that it would.

When an answer to a question is input by handwriting, whether the recognition target is a kana character representing a tongue or a kana character which is not a tonality, such as "fan" and "fan", is determined by the size of the character. There are times when you can only judge by that. In such a case, both "@" and "a" are the same in recognizing kana, and it is difficult to narrow down the input answer to only one character only by referring to the dictionary. Often. Therefore, in order to determine the answer input by handwriting as character information, both are displayed as correct answer candidates and the learner is allowed to select one of them again to confirm, or the size of the character prepared in a dictionary or the like is determined. In the case of the former, it is inevitable to assume that the character closest to the information of
There is a problem in that the operation of the learner is complicated, and in the latter case, correct / incorrect judgment can be made based on an answer contrary to the intention of the learner.

Accordingly, the learning test apparatus and the learning test method of the present invention solve the above-mentioned problems, and improve the efficiency and reliability of processing when determining a handwritten input answer as a character or a symbol. The aim was to improve the usability for learners.

[0007]

Means for Solving the Problems and Actions / Effects The learning test apparatus of the present invention, which has been made to achieve the above object, inputs a learner's answer to a question for which a correct answer is prepared in advance, and outputs the correct answer. A learning test device for determining validity in light of the correct answer, wherein the learner inputs non-code information as non-code information, and recognizes the input non-code information. A candidate character group extracting means for extracting a recognition result meeting a predetermined condition as a candidate character group when converting to a character that can be represented by the code information, within a range permitted by the correct answer for the candidate character group. And answer output means for giving priority to the result of recognition and outputting the result as the answer.

A learning test method corresponding to the learning test apparatus is a learning test method in which a learner's answer to a question for which a correct answer is prepared in advance is input and the validity of the answer is determined in light of the correct answer. When the learner inputs the answer as non-code information and recognizes the input non-code information and converts the non-code information into characters that can be represented by the code information, a recognition that satisfies predetermined conditions The gist is that a result is extracted as a candidate character group, and the recognition result within the range permitted by the correct answer is prioritized and output as the answer to the candidate character group.

In the learning test apparatus and the learning test method of the present invention having the above-mentioned configuration, for a question for which a correct answer has been prepared in advance, a question answered by a learner is input as non-code information, and the non-code information is recognized. When converting to a character that can be represented by the code information, a recognition result that meets a predetermined condition is extracted as a candidate character group. With respect to this candidate character group, an answer handwritten by the learner is selected with priority given to a recognition result within a range permitted by the correct answer, and the result of the selection is output as an answer.

As the non-code information, image information of handwritten characters, voice information input from a microphone or the like, gestures and hand gestures taken by a video camera or the like can be considered. The predetermined condition required for recognizing non-code information is not particularly limited in the present invention, and for example, condition setting such as a degree of coincidence at the time of recognition may be considered. Also, regarding the "criterion for determining whether or not the correct answer is within a permissible range" which is a criterion when narrowing down the recognition result from the candidate character group and outputting the answer, attention is paid to various properties possessed by the correct answer prepared in advance. And can be set freely. For example, if the correct answer is a letter or symbol, the code number set for the letter or symbol, etc.
When the correct answer is to select from predetermined options, a code number of a code displayed as an option can be considered.

In the learning test apparatus and the learning test method of the present invention having the above-described configuration, the answers are narrowed down based on the information held by the correct answers prepared in advance. Even with an input method in which the information tends to vary, the result of the answer recognition can be appropriately narrowed. Therefore, when recognizing characters that can be represented by code information from non-code information, erroneous recognition due to recognition accuracy can be reduced. As a result,
Learners can concentrate on learning.

If the character group (recognition result) extracted by the extraction of the candidate character group contains a single character, the processing for giving priority to the recognition result in the range permitted by the correct answer is not performed. It is also preferable to output the recognition result as it is as an answer. In this case, the non-code information input as the answer can be determined more quickly as characters.

After recognizing the non-coded information, a character type of the non-coded information to be recognized is specified, and the candidate character group is generated from the non-coded information by referring to a dictionary prepared for each of the specified character types. By doing so, it is also preferable to extract the candidate character group. In the learning test device,
For example, in the case where the reading is answered in the learning of the kanji test, kana characters are input, and the character type constituting the answer may be determined for each learning content. Therefore, in such a case, when recognizing the input non-code information, it is sufficient to refer to the dictionary for only the corresponding character type. By adopting such a configuration, it is possible to quickly and efficiently narrow down the answer of the handwritten input to characters and the like, and to improve the accuracy of character recognition by reducing the number of characters to be referred to. be able to.

[0014]

Other Embodiments of the Invention The present invention includes the following other embodiments. A first mode is a medium storing a program for inputting a learner's answer to a question for which a correct answer has been prepared in advance, and implementing a learning test apparatus for determining the validity of the answer in light of the correct answer, When the learner makes the answer, non-code information input processing to input as non-code information, when recognizing the non-code information input by the input processing and converting it to a character that can be represented by the code information,
A candidate character group extraction process for extracting a recognition result that meets a predetermined condition as a candidate character group, giving priority to the recognition result within the range permitted by the correct answer for the candidate character group extracted by the candidate character group extraction process And a computer-readable recording medium storing a program for executing an answer output process for outputting the answer. Such recording media include flexible disks, CD-
ROMs, magneto-optical disks, punch cards, paper on which codes such as bar codes are printed, and the like can be considered.

These recording media are mounted on a flexible disk device or a magneto-optical disk device of a computer, read a program recorded by magnetic means or optical means, and transfer the read program to a memory of the computer.
Alternatively, by directly executing this, the function of the learning test described above is realized on a computer.

According to a second aspect, a learner's answer to a question in which a correct answer is prepared is input by being loaded into a memory of a computer and executed by a microprocessor of a computer system, and the validity of the answer is determined. The present invention is configured as a device for supplying a program for realizing a learning test device to be determined in light of the correct answer via a communication line. Also in this case, the function of the learning test described above can be realized on the computer.

[0017]

DESCRIPTION OF THE PREFERRED EMBODIMENTS In order to further clarify the configuration and operation of the present invention described above, embodiments of the present invention will be described below based on examples. FIG. 1 is a schematic configuration diagram showing the overall configuration of a kanji learning test apparatus 20 according to a preferred embodiment of the present invention. As shown in the figure, the kanji learning test apparatus 20 of this embodiment is realized by executing a predetermined program in a computer main body 38.
On the computer body 38, a CRT 2 for visually displaying the result of the determination of the consistency between the question and the answer and the correct answer is displayed.
6. A keyboard 24 as a means for inputting an instruction to execute each process such as starting the test, selecting a question level, and finishing the test.
A tablet 3 as a means for the user to manually write the answer to the question sentence displayed on the CRT 26 and to input the answer
5, a pen 34, a mouse 36 and the like are connected. The tablet 35 is a device that inputs characters written on the tablet with the pen 34 as coordinate information. In this device, the input of the answer can be performed by the mouse 36. In this case, the learner operates the mouse 36,
Trace the shape of the character you are trying to enter as an answer. When the mouse 36 is operated while holding down its button (called dragging), the traced character is displayed in a predetermined area of the CRT 26 and is input as image information. Image information is non-coded information that is not coded.

In this embodiment, the tablet 35 and the CRT 2
6 are configured separately, but it is of course possible to configure them integrally. In the kanji learning test apparatus 20, a learner writes a solution to a problem on a tablet 35 using a pen 34, inputs the solution as image information, extracts candidate characters matching the image information, and determines candidate characters. After that, processing such as judging the coincidence between the answer and the correct answer is performed. These processes are performed by a program loaded in the main memory of the computer main body 38. The program executed by the kanji learning test apparatus 20 is pre-written in a CD-ROM, and the CD-ROM is stored in a CD-ROM drive unit 39.
Is set, the preparation for executing various processes of the learning test in the computer main body 38 is completed.

FIG. 2 is a block diagram showing an internal configuration of a computer main body 38 which is hardware of the kanji learning test apparatus 20 according to a preferred embodiment of the present invention. As shown in FIG. 2, this device mainly includes a CPU 21 for executing various arithmetic processing for controlling operations related to the kanji learning and examining device 20 according to a preset program, and a bus 31.
The following components are connected to each other. ROM22
Stores in advance programs and data necessary for the CPU 21 to execute various arithmetic processes.
Similarly, it is a memory for temporarily reading and writing various programs and data related to the kanji learning test device 20 necessary for the CPU 21 to execute various arithmetic processes. The keyboard interface 25 controls input and output of signals from the keyboard 24, and a pointing device interface 37.
Controls input and output of signals from the mouse 36 and the tablet 35. The CRTC 27 is a CRT 2 capable of color display.
6 and the printer interface 29
Controls output of data to the printer 28. The hard disk 32 stores various programs loaded into the RAM 23 and executed, various programs provided in the form of device drivers, various conversion dictionaries, and the like. The control of the hard disk 32 is performed by a hard disk controller (HDC) 30. The CD-ROM drive unit 39 is a CD-RO drive unit.
This is a device for reading the recorded contents recorded in M. The timer 33 generates date and time information indicating a predetermined time, such as a current time, a date, and the like.

In the hardware configured as described above, the CD-ROM on which the kanji learning program is recorded is
When installed in the D-ROM drive unit 39, the operating system running on the computer main body 38 recognizes the CD-ROM and sets the state in which the kanji learning test program can be executed. By starting this program, the user who wants to perform the kanji learning test can display the screen of the question sentence, input the answer, recognize the character of the input answer, judge the correctness of the answer recognized by the character, and check the correctness The display or the like is performed, and the computer main body 38 functions as the kanji learning test apparatus 20.

Next, the details of the kanji learning test performed by the hardware configured as described above will be described. First, when a key operation for giving an instruction to execute a kanji learning test is performed on the keyboard 24, the kanji learning test program recorded on the CD-ROM by the CPU 21 is loaded onto the RAM 23, and the program is ready to be executed.
The program on the CD-ROM is stored on the hard disk 3
2, and may be started from the hard disk 32.

The outline of the kanji learning test program will be described. FIG. 3 is a block diagram showing the contents of the kanji learning test realized by the kanji learning test apparatus 20. In the kanji learning test apparatus 20 of this embodiment, a kanji test simulation test is mainly performed for each class. Rank RA represents the difficulty of the practice test. In the present embodiment, three ranks R, Grade 3, and Grade 4 are provided so that learning suitable for each ability can be performed.
A is set, and the less difficult the series is, the more difficult the problem is. Each rank RA is divided into first to eight test TIs, and different questions are prepared for each test TI. The content of the test questions prepared by each test round TI is further divided into 11 different genres GE. In this embodiment, in order to be able to acquire a wide range of knowledge related to kanji, it is necessary to read a question, the same kanji, a sentence kana, a idiom, a radical, a synonym / synonym, a four-letter idiom, a typo, a typo, and a dictation. , And various genres GE such as the number of strokes. The CD-ROM records a predetermined number of questions for each of these ranks RA, test times TI, and genres GE. In the present embodiment, by default, the test times TI are set from the first time to the eighth time, but the learner can arbitrarily select from a predetermined number of questions recorded on the CD-ROM. , And registering it as a test question, it is also possible to create a new ninth or later practice test.

When the kanji learning test program is started,
First, a process of displaying the start screen shown in FIG. 4 on the CRT 26 is performed. On the start screen, a button KB for instructing the learner to move to a screen describing the kanji test, a button EX for instructing the start of the simulation test, a button QT for instructing the end of the program, and auxiliary information on how to use the program Is displayed, and credits such as the author are displayed. The learner operates the mouse 36 to select these buttons, so that the process assigned to each button can be selected. When the learner selects the button EX for executing the kanji test simulation test on the start screen, the detailed item setting screen WD shown in FIG.
Processing for displaying on the RT 26 is performed. In the detailed item setting screen WD, a name input field N for inputting the name of the answerer is provided.
I, a rank selection button group RB for selecting a rank RA, a pop-up list TP for selecting a test time TI, and other setting items are displayed. The learner operates the keyboard 24 to input his / her name, or operates the mouse 36 to select one of the ranks RA and the test times TI.
By selecting, a desired kanji test simulation test can be set. The name of the answerer is NI in the name input field.
Is registered by clicking the “Registration” button RS, and is displayed in the form of a pop-up in the name input field NI.

Other items that can be set on the detailed item setting screen WD include the contents of the setting of the time limit during the simulation test, the time limit until the end of the entire simulation test, and the permissible time for each question. The answer time and the setting of whether the test format is to be performed in a simulation test format or in a one-by-one format. In response to these items, a detailed item setting screen WD includes an end time setting field TT for setting a time limit, a button FQ for setting to forcibly end the simulation test after the time set here has elapsed, and one button. An answer time setting field AT for setting an answer time for each question, a button FT for forcibly setting to move to the next question after the time set here has elapsed, and a practice test format selection for selecting a practice test format as a test format. A button PB and a question-and-answer format selection button QA for selecting a question-and-answer format are also displayed.

Further, in the detailed item setting screen WD, when the answer is made by handwritten characters, the completion of the input of the answer by the handwritten characters is set based on how many seconds have elapsed since the input was stopped. A recognition time setting column IR is provided. The answerers were tablet 35 and pen 3
The time until recognition can be set freely according to the level of proficiency in handwriting input using the mouse 4 or handwriting input using the mouse 36. Also, at the bottom of the detailed item setting screen, a “return” button RT for instructing return to the start screen,
A "learning history" button PH for displaying a learning history, a "help" button HP for displaying help, and a "simulation test start" button ST for instructing start of a simulation test are displayed. By operating these buttons, the learner can start a simulation test and the like according to the conditions set on the detailed item setting screen WD. "Mock test start button" S
When T is operated, a mock test is started. The first mock test is a "reading" test for answering the reading of kanji (see FIG. 6).

When the simulation test is started, a question is displayed on the CRT 26 for each genre GE. Learners
The answer to each question can be input by drawing handwritten characters on the tablet 35 with the pen 34. Alternatively, a handwritten character can be input by operating the mouse 36. Further, the user may operate the keyboard 24 to input the answer. When the practice test is started, a series of processing from displaying these questions to inputting the answer is performed for all questions prepared for each genre GE in the test time TI selected by the learner during the time limit. Do
Thereafter, all of the answers are collectively scored, and the score is displayed together with the mock test passing border line. In the present embodiment, the borderline that passes the simulation test is set to a correct answer rate of 80% or more. As a result, the learner can select the rank RA and the test time T
It can be determined whether or not the kanji test simulation test for I has passed. When the question-and-answer format is selected, scoring is performed for each question, but the score is displayed in the same manner as in the simulation test format.

The kanji learning test apparatus 20 of this embodiment has a function of accumulating and storing the past learning history of each learner. When the learner name is selected, the "learning history" button PH is selected. Then, for the simulation test that the learner has performed in the past, a process of displaying the score transition for each test time TI and the correct answer rate for each genre GE in the specific test time TI is performed. The display of the score transition and the correct answer rate is displayed in a graph and color-coded according to the score transition and the correct answer rate.

The outline of the kanji learning test and the setting screen have been described above. Next, the kanji learning test apparatus 20 of this embodiment
A series of processes performed from the start to the end of the kanji test simulation test will be described with reference to the flowchart of FIG. When the kanji test simulation test routine shown in FIG. 7 is started, the rank RA of the mock test selected by the learner operating the mouse 36 on the detailed item setting screen WD displayed on the CRT 26 after the start, and the test time TI Is performed (step S100), and the simulation test time limit T is similarly set based on the setting on the detailed item setting screen WD.
A process for setting the TL and the answer limit time PTL for each question is performed (step S110). Practice test time limit TTL
Represents the time from the start to the end of the simulation test. When this time has elapsed, the simulation test ends regardless of whether or not unanswered questions remain. The simulation test limit time TTL can be set in units of one minute, with the maximum being 120 minutes, and it is also possible that no limit is imposed by the selection of the learner. The answer limit time PTL indicates the time from the setting of each question to the input of the answer. When this time has elapsed, regardless of whether the answer to the question being set has been answered, the answer limit time elapses. Proceed to the next problem. The answer limit time PTL can be set in units of one second, with the maximum being 100 seconds, and it is also possible that no limit is imposed by the selection of the learner.
In the present embodiment, the following processing will be described on the assumption that the learner has selected to set the practice test limit time TTL and the answer limit time PTL for each question to a fixed time.

Next, a process of outputting the first question from among the questions stored in advance on the CD-ROM and displaying it on the CRT 26 is performed (step S120), and a process of initializing the number N of answer characters to a value of 1 (Step S1
30). Since the kanji learning test apparatus 20 of the present embodiment inputs the learner's answer every time one character is written on the tablet 35, the answer character number N is incremented by one each time the answer is input by one character. Is done. Note that an answer using handwritten characters can be made by using the mouse 36. However, in the following description, it is assumed that input of handwritten characters is performed by the tablet 35.

After the initial setting process for the number N of answer characters is performed, the learner can answer the question set. In this state, first, it is determined whether or not the answer limit time PTL for each question has elapsed (step S135). If the answer limit time PTL for each question has not elapsed, the answer is continued. It is determined whether or not the first character has been written on the tablet 35 by the learner (step S140). In the example shown in FIG. 6, “reading” is used as the first genre GE when the rank RA is the second level.
The question is given, and it shows that you are answering this question. When you write a character on the tablet 35 with the pen 34,
The shape as handwritten is displayed in the handwritten character input box HA of the CRT 26. After the input of handwritten characters from the tablet 35 starts, if handwritten input is interrupted for a predetermined time,
It is determined that the answer character handwriting input has been completed for one character. This determination is made based on whether or not the time during which no input is performed on the tablet 35 has elapsed. The time used for this determination is set in the detailed item setting screen WD
Are set arbitrarily by the learner by using the recognition time setting column IR. Before the determination based on the elapse of this time, the learner operates the mouse 36 to operate the CRT2.
When the recognition button RR displayed on 6 is selected, it is regarded that the entry of the answer character is completed, and the image recognition of the inputted answer is started. If the answer text has been entered,
The Nth handwritten answer character is input as image information (step S145), and the input image information is determined as the Nth answer character (step S145).
150). In FIG. 6, a question asking the reading of the kanji "hotaru" is displayed, the answer for two characters up to "hota" has already been input, and the character "ru" has been handwritten as the third character. The state is shown. Details of the process of recognizing handwritten characters will be described later.

The determined answer character is N which constitutes the answer.
It is stored in the area A (N) on the RAM 23 as the first character. Thereafter, the variable N is incremented by 1 (step S155), and the process returns to step S135 until the learner instructs the completion of the answer, and the above processing is repeated (step S160). In the meantime, if the scoring button AQ displayed on the screen is clicked or the arrow button AR for instructing to proceed to the next question is clicked, it is regarded that the answer is completed, and the input processing of the answer for the question is stopped. (Steps S160 and S165). If it is determined in step S135 that the answer limit time PTL for each question has elapsed, the input process of the answer for the question is stopped regardless of the learner's intention (steps S135 and S165).

After the answer input process is stopped, it is determined whether or not the simulation test time limit TTL has elapsed. If it is determined that the simulation test time limit TTL has not elapsed, it is determined whether a problem remains. If there are unanswered questions left (step S175), the above process is repeated from the process of outputting the next question and displaying it on the CRT 26 (steps S120 to S175). If it is determined that the simulation test time limit TTL has elapsed, or if all the questions have been answered (step S170 or S175), the simulation test is terminated and all the input After performing the process of determining the correctness of the answer (step S180), the process of recording the score is performed (step S190), and the present routine is terminated.
Even if all the questions have been answered, if there is time, the learner may check the questions that have already been solved. Ending is also realistic. In such a case, the learner can end the simulation test by operating the simulation test end button ED shown in FIG.

The series of processes performed by the kanji learning test apparatus 20 of this embodiment from the start to the end of the kanji test simulation test has been described. Next, the kanji learning examination device 20
In the series of processes performed by the user, details of the process of inputting the answer by handwritten characters and the process of determining the correctness of the answer, which are characteristic portions of the present embodiment, will be described. FIG. 8 is an explanatory diagram showing the functional units of the kanji learning test apparatus 20 that executes the correctness judgment processing of the answer from the input processing of the answer by the handwritten character.
The learner operates the mouse 36 to rank RA and test times TI.
Is designated, the designated information is transmitted via the pointing device interface 37 to the genre recognition unit 142 realized by a series of processes by the CPU 21.
Sent to When the start of the test is instructed, the genre recognizing unit 142 determines which genre GE among the questions for the questions prepared in each test time TI will be set.
It is a functional unit that recognizes whether it should belong to. Questions are set for the genre recognized by the genre recognition unit 142, but the functional units for setting the questions are not shown in FIG. The genre recognizing unit 142 sends the information of the recognized genre GE to a question function unit (not shown) and also sends it to a character type narrowing unit 144. A character type narrowing unit 144 determines a character type that can be input as an answer to a question belonging to the genre GE, and instructs the DPM processing unit 150 to refer to a dictionary stored in the hard disk 32 only for the character type. It is. For example, when the type of the genre GE is “reading”, the character input as the answer is “hiragana”.
Therefore, upon receiving an instruction from the character type narrowing unit 144,
At the time of character recognition, the DPM processing unit 150 refers only to the area in the dictionary where the hiragana is stored.

When the learner writes one of the answer characters on the tablet 35 with the pen 34 and the handwritten character is input, the pointing device interface 37 sends the input information to the handwritten answer recognition unit 146 in the CPU 21. . The handwritten answer recognition unit 146 is a functional unit that recognizes characters input by handwriting as image information. Recognition as image information is performed by converting handwriting of a handwritten character drawn on the tablet 35 into coordinates of the pen 34 at predetermined time intervals, and storing the handwritten one as a large number of pieces of coordinate information. The handwritten answer recognition unit 146 temporarily stores a large amount of coordinate information in the image information storage unit 148 secured in the RAM 23. The coordinate information is sent to the CRTC 27, and the handwriting of the drawn handwritten character is displayed so as to be visible in the handwritten character input box HA of the CRT 26.

After performing the above processing, handwritten answer recognition section 146 outputs a character recognition execution command to DPM processing section 150. The DPM processing unit 150 is a functional unit that recognizes handwritten objects as characters based on the coordinate information stored by the handwritten answer recognition unit 146. This character recognition is performed by performing pattern matching based on the coordinate information stored in the image information storage unit 148, and specifying a character having a high rate of matching with the character stored in the dictionary stored on the hard disk 32. Done. If handwritten characters vary in size, tablet 35
Since the position drawn above is also different every time, the pattern matching is performed by using a so-called dynamic pattern matching method after converting into a relative positional relationship from the center of gravity of the handwritten figure. As a result of the matching, a character determined to have a high matching rate is temporarily stored in the candidate character storage unit 152 secured on the RAM 23 in the form of character code information as a candidate character.

After performing the above processing, DPM processing section 150 instructs code collation section 154 to collate character codes. The code collating unit 154 searches for a candidate character stored in the candidate character storage unit 152, which is a predetermined area of the RAM 23, compares it with a predetermined condition according to the answer, and estimates the answer from the handwritten one. This is a functional unit for determining what is to be performed. The code collating unit 154 holds the prohibition code information that records the shift JIS code of the character that cannot form the answer. Of the candidate characters stored in the candidate character storage unit 152, By excluding candidate characters having the same shift JIS code as the prohibition code information from the answer candidates, characters estimated as the answer are narrowed down. After performing the above process, the code collating unit 154 sends the narrowed down characters as the answer to the input answer determining unit 156. The input answer determination unit 156 converts the answer information calibrated from the characters narrowed down as candidate characters by the code matching unit 154 into CRT as a result of character recognition.
This is a control unit that is displayed on the answer candidate display unit EC (see FIG. 6) of C27 and provided to the correctness determination unit 158. The first candidate is also visibly displayed in the answer column AN.

When the character displayed as the first candidate is different from the learner's intention, the learner clicks the recognition candidate display field EC with the mouse 36, and the input answer determination section 156 stores the other candidate stored in the input answer determination section 156. Characters are displayed as selectable. Therefore, learning is done by using mouse 3
If the user designates the answer by 6, the character to be handled as the answer can be easily replaced. When all the characters that make up the answer have been entered in this way and the correctness is determined,
The correct / wrong determining unit 158 determines whether the input answer matches the correct answer prepared in advance. Correctness judgment section 158 receiving answer information from input answer determination section 156
Is a functional unit that determines the coincidence between the answer information and the correct answer prepared in advance. The correctness / incorrectness determination unit 158 sets the question-and-answer format Q as a test format when the learner clicks a scoring button or on the detailed item setting screen WD.
When A is selected, when the transition to the next problem is instructed, the matching is determined. The result of the coincidence determination is notified to the learner by superimposing and displaying an "○" mark for a correct answer and an "x" mark for an incorrect answer in the answer column AN of the CRT 26.

In the foregoing, the functional unit for executing the answer correctness determination process from the answer input process using handwritten characters has been described. Next, among the processes executed by the CPU 21 corresponding to these functional units, an answer character determination routine (FIG. 7, step S150) which is a process until a handwritten one is determined as an answer character is shown in a flowchart of FIG. This will be described with reference to FIG. When this routine is started, first,
A process of acquiring the Nth handwritten answer character input as image information is performed (step S200). This processing is performed by the DPM processing unit 1 in the functional block shown in FIG.
50 corresponds to receiving image information from the image information storage unit 148. As described above, since N is initially set to the value 1 (FIG. 7, step S130), the first handwritten answer character is obtained first. Next, this N
Based on the second handwritten answer character, a filter 1 process for narrowing down answer candidate characters is executed, and a process of recognizing a character from a handwritten figure is performed (step S210). The processing performed by the filter 1 corresponds to the processing in the DPM processing unit 150, and this processing will be described later in detail.

Next, it is determined whether or not the number of characters narrowed down by the filter 1 is one (step S).
220) If it is not one character, a process of storing a plurality of recognized characters as answer candidate characters in a predetermined area on the RAM 23 is performed (step S230). This process corresponds to the candidate character storage unit 152 shown in FIG. In order to further narrow down the answer candidate characters recognized by the filter 1, the process of the filter 2 is performed (step S240). This processing is performed by the code collating unit 154 as a tank. The processing performed by the filter 2 will also be described later.
The character group narrowed down by the processing of the filter 2 is called a limited candidate character. Unless otherwise specified, the first candidate among the limited candidate characters is determined as an answer character as it is (step S250). If the number of characters narrowed down by the filter 1 is one, the input Nth handwritten answer character has been narrowed down to only one answer candidate character by the filter 1. Processing is not performed, and processing for determining this character as the answer character is performed (steps S220, S250). By adopting such a configuration, an answer character can be determined more quickly from a handwritten input.

After the Nth answer character determination process is completed (step S250), this character is displayed in the answer column A on the CRT 26.
N is displayed (step S260). The above processing corresponds to the input answer determination unit 156. After these processes, this routine ends, and the system waits for the next handwritten answer character to be input as image information. In the example shown in FIG. 6, two handwritten characters up to “hota” are sequentially determined as answer characters by the present routine and displayed in the answer column AN, but are marked or moved to the next question. Until is designated, this routine is repeatedly executed. Therefore, when the learner handwrites on the tablet 35, the next character (the character "ru" in the example of FIG. 6) is recognized by this routine.

Next, the filter 1 processing and the filter 2 processing for narrowing down the answer characters in the answer character determination routine described above will be described in detail with reference to the flowcharts of FIGS. FIG.
0 is a flowchart showing a processing routine corresponding to the filter 1. When this routine is started, first, a process of recognizing the genre GE of the question set is performed (step S300). The genre GE is a type of question such as “reading”, “identical kanji”, and “radical” in the case of the kanji test, and this processing of recognizing this corresponds to the processing of the genre recognition unit 142.

Subsequently, based on the information of the genre GE,
A process of determining a character type X to be searched for a dictionary is performed (step S320). For example, as shown in FIG. 12, if the information of the genre GE is “reading”, the character type that can be input by handwriting as an answer is limited to Hiragana, so the character type X to be searched in the dictionary is determined as “Hiragana”. This processing corresponds to the character type narrowing unit 144. Thereafter, the handwritten characters obtained in step S200 are extracted as image A (step S340), and only the area in the dictionary where character type X is stored is searched (step S36).
0), a process of extracting characters having a pattern similar to image A as answer candidate characters (step S3)
80). The determination as to whether or not the pattern is similar to the pattern A is performed by the dynamic pattern matching method as described above. However, since a large number of such character recognition methods are already known, detailed description is omitted. I do.

FIG. 12 is an explanatory diagram showing an example of a process when a question asking the reading of a kanji character “anxiety” is set. In this case, the image 62 of the handwritten characters “F”, “A”, and “N” sequentially written on the tablet 35 by the learner is stored in the hiragana storage area 6 in the dictionary.
Processing for searching for characters of similar patterns is performed with reference to only 4. As a result, "F" is extracted as the first character, "A" and "@" is extracted as the second character, and "N" is extracted as the candidate character as the third character. .

As described above, in the kanji learning test apparatus 20 of the present embodiment, the kana character is always input when the reading is answered in the kanji test learning. Is determined, the genre GE is recognized, and at the same time, a character type to be searched in the dictionary is set corresponding to the genre GE. Therefore, it is sufficient to refer to the dictionary for only the corresponding character type in recognizing the answer of the handwritten input character. By adopting such a configuration, it is possible to quickly and efficiently process narrowing down the answer input by handwriting to characters without the learner previously specifying the character type to be referred in the dictionary. In addition, since the number of characters to be referred to decreases, the accuracy of character recognition can be improved.

Next, the details of the processing of the filter 2 will be described.
The processing routine will be described with reference to the flowchart of FIG. When this routine is started, first, the shift JIS of the answer candidate character obtained by the processing of the filter 1 is performed.
A process for obtaining a code is performed (step S400). In the example shown in FIG. 12, each of the answer candidate characters "fu", "a",
"82C3", "83A0", "829F",
“82F1” is required. Next, it is determined whether or not there is a code in the shift JIS code 70 that matches a preset prohibition code 72 based on the genre GE (step S420), and there is no code that matches the prohibition code 72. In this case, it is determined that there is no change in the answer candidate characters obtained by the already performed filter 1 process, and the process proceeds to the next process. If there is a code that matches the prohibition code 72, the character having this code is deleted from the answer candidate characters (step S440).
Move to the next process. The prohibition code 72 is set in advance for each genre.

In the example shown in FIG. 12, since the problem is the genre GE of "reading kanji", the kanji learning test apparatus 20 uses the repetitive character "kanon", which is a character that cannot constitute kanji reading in hiragana.ぁ ”,“ ぃ ”,“ ぅ ”,“ ぇ ”,
The shift JIS code of "@" is set as a prohibition code. The set prohibition codes are shift JIS codes “829F”, “82A1”, “82” corresponding to the repetitive sounds “ぁ”, “ぃ”, “ぅ”, “ぇ”, “ぉ”, respectively.
A3 "," 82A5 ", and" 82A7 "(hexadecimal notation). In the example shown in FIG. 12, among the second answer candidate characters “A” and “ぁ”, “ぁ” is the shift JIS code “8”.
29F ", which corresponds to the prohibition code, and is therefore deleted from the answer candidate characters. By this deletion, the repetitive sound “音” is excluded from the answer candidate characters, and the characters “fu”, “a”, and “n” that are not deleted are determined as the answer characters.

As a result of adopting the above-described configuration, the kanji learning test apparatus 20 of the present embodiment narrows down the answers based on information that is known in advance with respect to the correct answers prepared in advance, so that the answerer such as a handwritten input can be used. Even with an input method that tends to vary every time, it is possible to appropriately narrow down the results of answer recognition. Therefore, erroneous recognition caused by the recognition accuracy of handwritten characters can be reduced. By reducing trouble such as correction of misrecognized characters, the learner can concentrate on learning. In the above example, the murmur was deleted when the kanji reading was answered.However, for example, in the case of selecting the correct answer from a plurality of options prepared as answer candidates, the answer was given to the option. There are cases where symbols "A", "I", "U", etc. are answered. In such a case as well, in the processing of the filter 2, it is only necessary to delete the sound. In the case of showing a kanji character and answering which part is a radical, a shape provided for the kanji in question can be coded and prepared, and the rest can not be accepted as an answer. In any case, as long as the answer candidate characters are further narrowed down according to the range permitted by the correct answer, the filter 2 can be adopted.

The embodiments of the present invention have been described above.
The present invention is not limited to such an embodiment at all, and it goes without saying that the present invention can be implemented in various modes without departing from the gist of the present invention. For example, filter 1
And the filter 2 may not be separately performed. For example, a dictionary corresponding to a genre may be further subdivided and processing corresponding to both filters may be executed at a time. Further, the present invention may be applied to a learning test device other than the kanji test. Any device that prepares a question and a correct answer and narrows down the answer by focusing on the properties of the correct answer can be widely applied as a learning test apparatus and its method. In the present embodiment, the input of the answer is performed by the tablet 35 and the mouse 36.
The input means is not limited to these, and OCR
(Optical character reading device) In addition to inputting handwritten characters by detecting the movement of a finger and operating a virtual pen, or inputting an answer by voice and performing voice recognition Naturally, the present invention includes a technology that can be replaced with the answer input unit of the present embodiment, such as a structure for obtaining candidate characters and a structure for recognizing a video of a gesture or gesture word such as sign language to obtain an answer candidate character.

[Brief description of the drawings]

FIG. 1 is a schematic configuration diagram showing an overall configuration of a kanji learning and examining apparatus according to an embodiment of the present invention.

FIG. 2 is a block diagram illustrating an outline of a hardware configuration in which a kanji learning test apparatus according to an embodiment of the present invention is realized.

FIG. 3 is an explanatory diagram showing the contents of a kanji test simulation test that can be realized by the kanji learning test device according to the embodiment of the present invention.

FIG. 4 is an explanatory diagram showing an example of a start screen displayed on a CRT 26 in a kanji test simulation test.

FIG. 5 is an explanatory diagram showing an example of a detailed item setting screen displayed on the CRT 26 in the kanji test simulation test.

FIG. 6 is an explanatory diagram showing an example of a screen during answer input displayed on the CRT 26 in the kanji test simulation test.

FIG. 7 is a flowchart showing a processing routine of a kanji test simulation test executed in the embodiment of the present invention.

FIG. 8 is a block diagram showing functions realized in the kanji learning test apparatus according to the embodiment of the present invention.

FIG. 9 is a flowchart showing an answer character determination routine.

FIG. 10 is a flowchart illustrating a processing routine corresponding to a filter 1;

FIG. 11 is a flowchart illustrating a processing routine corresponding to a filter 2;

FIG. 12 is an explanatory diagram illustrating an example of processing as filters 1 and 2;

[Explanation of symbols]

 Reference Signs List 20 Kanji learning test device 21 CPU 22 ROM 23 RAM 24 Keyboard 25 Keyboard interface 26 CRT 27 CRTC 28 Printer 29 Printer interface 30 HDC 31 Bus 32 Hard disk 33 Timer 34 Pen 35 tablet device 36 mouse 37 pointing device interface 38 computer body 39 CD-ROM drive unit 62 handwritten answer 64 hiragana storage area 70 shift JIS code 72 prohibition code 142 genre recognition section 144 character type narrowing down section 146: Handwritten answer recognition section 148 ... Image information storage section 150 ... DPM processing section 152 ... Candidate character storage section 154 ... Code collation section 156 ... Input answer determination section 158 ... True / false judgment section

Claims (4)

[Claims] 1. A learning test apparatus for inputting a learner's answer to a question for which a correct answer has been prepared in advance and determining the validity of the answer in light of the correct answer, wherein the learner uses the answer as the answer. Non-code information input means for inputting as non-code information; and recognizing the input non-code information and converting the input non-code information into characters that can be represented by the code information. A learning test apparatus comprising: candidate character group extracting means to be extracted; and answer output means for giving priority to a recognition result within a range permitted by the correct answer for the candidate character group and outputting the result as the answer. 2. The learning test device according to claim 1, wherein the answer output unit is configured to output a response within a range permitted by the correct answer when the character group extracted by the candidate character group extraction unit is singular. A learning test device, which outputs the candidate character group as the answer without performing the conversion giving priority to the recognition result. 3. The learning verification apparatus according to claim 1, wherein said candidate character group extracting means is provided with a character type designating means for designating a character type of non-code information to be recognized, and for each of said designated character types. And a candidate character group generating means for generating the candidate character group from the non-code information with reference to the obtained dictionary. 4. A learning test method for inputting a learner's answer to a question for which a correct answer has been prepared in advance, and determining the validity of the answer in light of the correct answer. When the input non-code information is recognized and converted into a character that can be represented by the code information, a recognition result meeting a predetermined condition is extracted as a candidate character group, and the candidate character A learning test method in which a recognition result within a range permitted by the correct answer is preferentially output to the group as the answer.