JP2007108961A - Illicit act detection server, system, and data processing method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To automatically and effectively identify the examinees who may have executed the illicit acts, by discriminating possible actions of illicit acts from action patterns of examinees. <P>SOLUTION: A server 300 for processing data received from a pen device 200 comprises a stroke information acquisition section 312 for acquiring stroke information which is information for actions of the pen device 200 on a space of an answer sheet on the basis of received data, a database registration section 330 for registering the stroke information action times of the pen device in database by sorting them for each examinee, and an illicit answer pattern determining section 360 for comparing the stroke information of a plurality of examinees by referring to the information registered in the database and detecting answer patterns of possible illicit acts on the basis of its similarity. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a system and method for mechanically detecting and preventing an examinee's fraud when performing a test.

  In the exam, it is necessary to effectively prevent fraudulent behaviors such as teaching answers between examinees and stealing others' answers. In recent years, since small communication devices such as mobile phones have become widespread, it is possible to share answers even when seats or venues are separated, and such illegal acts must also be prevented.

Various proposals have been made for methods of preventing fraud in tests.
For example, a computer network connecting the test director and the test taker will be installed at the test site, and a network terminal will be prepared for the test supervisor to monitor the test taker. Some screens are captured and this captured image is used as evidence of fraud (see Patent Document 1).
Also, in the learning system (e-learning) via the Internet, the student and student's exam answer information and the course time and record are sent to the server, and fraudulent activity based on the course time and course record information sent to the server There is a method of checking and preventing (see Patent Document 2).
In addition, by preparing multiple exam questions with different arrangements of questions and answer options and distributing them so that different test forms are passed to each examinee, cheating is simply done by copying the answers. A method of making it difficult has also been proposed (see Patent Document 3).

JP 2003-66826 A JP 2003-345907 A JP 2004-144903 A

As described above, various methods for preventing fraud in tests have been proposed. However, the prior art described in the above literature has the following problems.
For example, in the method described in Patent Document 1, it is necessary for the test supervisor to check the monitor information in a large amount of time in order to check the fraudulent behavior of each examinee. Burden.
Further, in the method described in Patent Document 2, when the test answer information is exchanged without using the network on the learning system, or the examinees answer by shifting their history time and history record, There is a problem that such illegal acts cannot be prevented.
Further, in the method described in Patent Document 3, when a plurality of examinees are in a situation where they can exchange information on test questions and the like, information on an array of sub-questions and answer options is exchanged, so that an answer can be obtained. There is a problem that fraudulent behavior that can be taught cannot be prevented.

  Furthermore, it is difficult to prevent the above-described prior art from performing an illegal act in which answers are taught using a small communication device such as a mobile phone that has been widespread in recent years. According to the prior art described in Patent Document 1, it is possible to capture a screen for a test taker who is determined to be suspected of fraud, but the test supervisor monitors the monitor. Because the burden of doing so is high, there is a high possibility that fraudulent acts will be overlooked.

These problems are caused by the fact that it is difficult for the prior art to efficiently discriminate actions that may be fraudulent by the examinee.
The present invention has been made in order to solve the technical problems as described above. The purpose of the present invention is to discriminate a possible fraudulent behavior from the examinee's behavior pattern and to cheat. It is to automatically and efficiently identify candidates who may have performed the test. This is also to support the monitoring of examinees and to reduce the burden on examination supervisors.

  The present invention for achieving the above object is realized as a system including the following servers. This server includes a stroke information acquisition unit that acquires stroke information, which is information related to the operation of the pen device on the answer sheet, based on the data received from the pen device, and the stroke information and the operation time of the pen device. The database registration unit that sorts and registers in the database and compares the stroke information of multiple examinees with reference to the information registered in the database, and based on the similarity, answers that may be fraudulent An incorrect answer pattern determination unit for detecting a pattern.

More specifically, the incorrect answer pattern determination unit determines that the stroke information is similar at least on a part of the answer sheet between the plurality of examinees, and the operation time of the pen device for the part is within a predetermined threshold. In some cases, an answer pattern including stroke information is determined as an incorrect answer pattern for at least one of a plurality of examinees.
In addition, the incorrect answer pattern determination unit obtains the answer order for each examinee from the identification information in the answer column and the operation time of the pen device with respect to the part, and the answer order is irregular for a given examinee, and other When the stroke information corresponding to a certain or more answer column is similar to the examinee, the answer pattern including the stroke information is determined to be an incorrect answer pattern for the predetermined examinee. In this case, the server further includes a target answer field determination unit that identifies an answer field of the answer sheet to which the stroke information acquired by the stroke information acquisition unit corresponds. Then, the database registration unit registers the identification information of the answer column specified by the target answer column determination unit in the database in association with the stroke information.
Furthermore, the incorrect answer pattern determination unit determines that the time required for answering a given question is shorter than the average for a given examinee, and the stroke information in the answer column for this question is similar to that for other examinees. In this case, an answer pattern including stroke information is determined as an incorrect answer pattern for a predetermined examinee. Also in this case, the server is configured to further include a target answer column determination unit. Then, the database registration unit obtains time information required for the answer for each question from the answer field specified by the target answer field determination unit and the operation time information of the pen device, and registers it in the database in association with the stroke information. I will do it.

  In addition, this server obtains information on the propagation time of radio waves transmitted and received between the wireless tags provided at a plurality of locations on the answer sheet and the pen device from the pen device, and transmits information on the propagation time on the sheet of the answer sheet. The time-position information conversion part which converts into the information of the position of the pen device in can be comprised further. In this case, the stroke information acquisition unit generates stroke information based on the position information of the pen device generated by the time-position information conversion unit.

Further, the server includes an unauthorized test taker specifying unit that specifies an examinee corresponding to an answer pattern determined as an answer pattern having a possibility of an illegal act by the incorrect answer pattern determining unit based on identification information of the pen device. And a notifying unit for notifying information on the examinee specified by the unauthorized examiner specifying unit.
Also, an image acquisition means (camera) for photographing the examinee and a display means (monitor) for displaying an image of the examinee photographed by the image acquisition means are prepared, and the notification section is operated by the unauthorized examiner identification section. An image for notifying the specified examinee's information can be generated, combined with the examinee's image taken by the image acquisition means, and displayed on the display means.

  The present invention can also be understood as a data processing method for a computer that receives data from a pen device and performs processing. In this method, a computer obtains stroke information, which is information related to the operation of the pen device on the answer sheet, based on data received from the pen device, and the stroke information is sorted for each examinee in a database. And comparing the stroke information of a plurality of examinees with reference to the information registered in the database, and detecting an answer pattern with a possibility of fraud based on the similarity. .

  According to the present invention configured as described above, based on the action when the examinee writes the answer on the answer sheet using the pen device, the examinee performs the action with the possibility of fraud. It can be mechanically determined whether or not it has been taken. This makes it possible to automatically and efficiently identify a candidate who may have performed an illegal act. And it can support the monitoring of examinees and reduce the burden on examination supervisors.

Embodiments of the present invention will be described below in detail with reference to the accompanying drawings.
In the present embodiment, a candidate who may have performed an illegal act is mechanically determined based on the answer pattern of the examinee with respect to the test question. Here, the answer pattern includes not only the correctness and content of the answer but also the action pattern when answering, specifically the order of answers and the time interval when answering. In this embodiment, in order to acquire the answer pattern in real time, a means for mechanically reading the writing operation on the answer sheet is used.

FIG. 1 is a diagram showing a basic configuration for realizing the system of the present embodiment.
As shown in FIG. 1, the system of this embodiment includes an answer sheet 100, a pen device 200, and a server 300. The pen device 200 has a function of acquiring electronic information (data) for recognizing how the pen device 200 has moved on the surface of the answer sheet 100, and a function of transmitting the acquired electronic information to the server 300. Prepare. Therefore, when the examinee writes an answer on the answer sheet 100 using the pen device 200, electronic information representing the writing operation of the pen device 200 at the time of writing is acquired and sent to the server 300.

  The server 300 recognizes the position and operation of the pen device 200 on the answer sheet 100 based on the electronic information sent from the pen device 200. Further, the server 300 holds information on the layout of the answer sheet 100 (hereinafter, layout information), and by comparing the layout information with the position and operation of the pen device 200, It has a function to analyze what answer is written in which answer column. In addition, it has a function of discriminating an action with a possibility of fraud by processing this analysis result.

  As described above, in the present embodiment, a function of acquiring electronic information (stroke information) representing the position and operation of the pen device 200 on the answer sheet 100 is essential. The function of acquiring the stroke information is realized by using an existing technique, and will be described in detail below with a specific example.

Here, as a specific example of the technique for acquiring stroke information, a method of transmitting / receiving radio waves between the wireless tag provided on the answer sheet 100 and the pen device 200 and obtaining the stroke information based on the propagation time will be described. .
FIG. 2 is a diagram illustrating a configuration example of the answer sheet 100 used in the present embodiment.
As shown in FIG. 2, the answer sheet 100 is provided with wireless tags 110a to 110d at the four corners, respectively. When the radio tags 110a to 110d receive radio waves (hereinafter, request radio waves) from the pen device 200, the radio tags 110a to 110d emit radio waves for response (hereinafter, reflected radio waves). Identification information of each of the wireless tags 110a to 110d is transmitted by this reflected radio wave.

  The pen device 200 according to the present embodiment transmits a request radio wave to the wireless tags 110a to 110d, receives the reflected radio waves from the wireless tags 110a to 110d, and acquires identification information. In addition, the pen device 200 measures the transmission time of the request radio wave and the reception time of the reflected radio wave. Then, based on the received identification information, for each of the wireless tags 110a to 110d, the time required from when the request radio wave is transmitted until the reflected radio wave is received is calculated, and this time information (hereinafter, propagation time information) is calculated. Send to server 300.

  The server 300 receives the propagation time information from the pen device 200, converts it into position information, and recognizes the position of the pen device 200 on the surface of the answer sheet 100. The conversion from the propagation time information to the position information can be performed by an interpolation calculation using a lookup table. That is, several virtual representative points are set on the surface of the answer sheet 100. For example, a virtual grid representing orthogonal coordinates with the upper left corner as the origin can be set on the surface of the answer sheet 100, and each intersection of the grid can be used as a representative point. Then, the propagation time information for each of the wireless tags 110a to 110d when the pen device 200 is at the position of each representative point is measured in advance, and the position (in the previous example, the coordinate value in orthogonal coordinates with the upper left corner as the origin is used. ) And registered in the lookup table. The current position of the pen device 200 is calculated by performing an interpolation operation using the propagation time information of the representative point registered in the lookup table and the propagation time information received from the pen device 200.

FIG. 3 is a diagram for explaining the concept of the interpolation calculation for obtaining the position of the pen device 200.
3, P1 to P4 are four adjacent representative points at predetermined positions on the answer sheet 100, and Px represents the current position of the pen device 200. That is, four representative points existing around the position Px of the current pen device 200 are P1 to P4. The propagation time information and position information regarding the representative points P1 to P4 are registered in the lookup table,

P1: (Tul1, Tur1, Tll1, Tlr1, CO1, LO1)
P2: (Tul2, Tur2, Tll2, Tlr2, CO2, LO1)
P3: (Tul3, Tur3, Tll3, Tlr3, CO1, LO2)
P4: (Tul4, Tur4, Tll4, Tlr4, CO2, LO2)

It is. Here, Tul is propagation time information for the radio tag 110a in the upper left corner shown in FIG. 2, Tur is propagation time information for the radio tag 110b in the upper right corner, Tll is propagation time information for the radio tag 110c in the lower left corner, and Tlr is right This is propagation time information for the wireless tag 110d in the lower corner. CO is a horizontal offset position (Cross Offset) in orthogonal coordinates with the upper left corner of the answer sheet 100 as the origin, and LO is a vertical offset position (Length Offset).

Regarding the current position of the pen device 200, the propagation time information acquired by the pen device 200 transmitting and receiving radio waves to and from the wireless tags 110a to 110d is Tulx, Turx, Tllx, Tlrx, and the position information is COx, LOx.

Px: (Tulx, Turx, Tllx, Tlrx, COx, LOx)

It can be expressed as.

In general, when the distance between the wireless tag 110a and the pen device 200 in the case of the propagation time information Tul is F (Tul),

F (Tul) = (CO ² + LO ² ) ^1/2

It is. Assuming that the horizontal size of the answer sheet 100 is C and the vertical size is L, similarly,

F (Tur) = ((C−CO) ² + LO ² ) ^1/2
F (Tll) = (CO ² + (L−LO) ² ) ^1/2
F (Tlr) = ((C−CO) ² + (L−LO) ² ) ^1/2

It can be expressed. Therefore, for example, when comparing the propagation time information Tul1 of the representative point P1 with the propagation time information Tulx of the current position Px of the pen device 200,

Tulx−Tul1 = (COx ² + LOx ² ) ^1/2 − (CO1 ² + LO1 ² ) ^1/2

This relationship holds. Similarly, a predetermined relational expression is obtained for the propagation time information Tur, Tll, and Tul. Further, the same relationship holds between the representative points P2 to P4 and the current position Px of the pen device 200. Therefore, position information COx and LOx of the current position Px can be obtained from these relational expressions.

  The pen device 200 transmits and receives radio waves to and from the wireless tags 110a to 110d as needed, and periodically transmits the acquired propagation time information to the server 300. The server 300 sequentially recognizes the position of the pen device 200 at each time point from the received propagation time information. Therefore, by arranging the positions of the pen devices 200 recognized by the server 300 in time series, it is possible to recognize the movement trajectory of the pen devices 200 on the answer sheet 100.

  As described above, the method of transmitting / receiving radio waves between the wireless tags 110a to 110d provided on the answer sheet 100 and the pen device 200 and obtaining the stroke information based on the propagation time has been described. It is just an example of how to obtain. In the above example, the pen device 200 acquires the propagation time information and transmits it to the server 300, and the server 300 converts the propagation time information into the position information of the pen device 200. However, a processor with a sufficiently high processing speed is incorporated. For example, the pen device 200 may calculate its own position information and transmit the calculated position information from the pen device 200 to the server 300.

  Further, as in the technique described in Japanese Patent Application Laid-Open No. 2004-94907, for example, a two-dimensional code representing position information is formed on the paper surface of the answer sheet 100, and the position on the paper surface is specified by reading this two-dimensional code. It is also possible to obtain stroke information of the pen device 200 by using the technique. In this case, the pen device 200 is provided with optical reading means for reading the two-dimensional code, and the position of the pen device 200 is recognized by reading and analyzing the image of the two-dimensional code and decoding the position information. It becomes.

Next, the configuration of the pen device 200 in the present embodiment will be described.
FIG. 4 is a diagram illustrating a configuration of the pen device 200.
The pen device 200 has a writing unit 210 that records characters and figures on the answer sheet 100 by the same operation as a normal pen, and the movement of the writing unit 210 is monitored and the pen device 200 is pressed against the answer sheet 100. And a writing pressure detection unit 220 for detecting. In addition, a radio wave detection unit 230 that transmits and receives radio waves to and from the wireless tags 110 a to 110 d (see FIG. 2) provided on the answer sheet 100, and a control unit 240 that controls the entire electronic operation of the pen device 200. And a communication unit 250 for transmitting data to the server 300. The control unit 240 includes a processing unit 241 that performs arithmetic processing, a clock 242 for obtaining time information, a memory 243 that stores a program that controls the operation of the control unit 240 and identification information (device ID) of the pen device 200. Is provided.

When characters and symbols are entered on the answer sheet 100 using the pen device 200, the writing pressure detection unit 220 detects that the recording is performed by the pen device 200, and the detection signal is transmitted to the radio wave detection unit 230. Send to.
When receiving the detection signal from the writing pressure detection unit 220, the radio wave detection unit 230 repeatedly transmits a request radio wave and receives reflected radio waves from the wireless tags 110a to 110d provided on the answer sheet 100. The event is notified to the processing unit 241 of the control unit 240 when the request radio wave is transmitted and the reflected radio wave is received, and the received reflected radio wave is transmitted to the processing unit 241.

  When receiving a radio wave transmission / reception event from the radio wave detection unit 230, the processing unit 241 of the control unit 240 refers to the clock 242 to acquire the transmission time of the request radio wave and the reception time of the reflected radio wave and calculate the interval. . The interval from the transmission time of the request radio wave to the reception time of the reflected radio wave is propagation time information. Then, identification information of the wireless tags 110a to 110d is extracted from the reflected radio waves, and the propagation time information is sorted for each of the wireless tags 110a to 110d based on the identification information.

  The communication unit 250 periodically transmits the propagation time information sorted for each of the wireless tags 110a to 110d to the server 300. Further, when transmitting the propagation time information, the communication unit 250 refers to the clock 242 of the control unit 240 to acquire the transmission time, accesses the memory 243 to acquire the device ID, and transmits these information to the propagation time information. Together with the server 300. Note that the transmission means may be wired or wireless, and any communication method may be used.

Next, the configuration of the server 300 will be described.
FIG. 5 is a diagram schematically illustrating a hardware configuration example of a computer suitable for realizing the server 300.
The computer shown in FIG. 5 includes a CPU (Central Processing Unit) 301 that is a computing means, a main memory 302 that is a storage means, and a magnetic disk device (HDD: Hard Disk Drive) 303. In addition, a network interface 304, a graphics display mechanism 305 including a video memory and a display, and an input device 306 such as a keyboard and a mouse are provided.

FIG. 6 is a diagram illustrating a functional configuration of the server 300 in the present embodiment.
As illustrated in FIG. 6, the server 300 includes a time-position information conversion unit 311 and a stroke information acquisition unit 312 for processing the propagation time information transmitted from the pen device 200 to generate stroke information. In addition, based on this stroke information, a target answer column determination unit 320 and a database registration unit 330 for creating a database of answer patterns for each examinee, and a storage unit 340 are provided. Furthermore, the server 300 includes a stroke information correction unit 350 and an incorrect answer pattern determination unit 360 for detecting an answer pattern that may be fraudulent based on the created answer pattern database. In addition, based on the determination result of the incorrect answer pattern determination unit 360, a test taker specifying unit 370 and a notification image generating unit 380 for specifying a test taker who may have performed an illegal act and notifying a test supervisor are provided. Prepare. Among these configurations, the storage unit 340 is realized by, for example, the magnetic disk device 303 illustrated in FIG. Other configurations are realized by the program-controlled CPU 301 and the main memory 302 shown in FIG. 5, for example.

The time-position information conversion unit 311 inputs the propagation time information transmitted from the pen device 200, and uses the interpolation operation using the above-described lookup table to position the pen device 200 on the answer sheet 100. Information (COx and LOx described above) is obtained.
The stroke information acquisition unit 312 arranges the position information calculated by the time-position information conversion unit 311 in time series, and generates stroke information representing a trajectory of the pen device 200 moving on the answer sheet 100. The stroke information also includes the time (hereinafter, writing time) required from the start to the end of the movement of the pen device 200 in the stroke.

The target answer field determination unit 320 determines which answer field in the answer sheet 100 the stroke information generated by the stroke information acquisition unit 312 is based on the position information calculated by the time-position information conversion unit 311. . In the present embodiment, the layout information of the surface of the answer sheet 100 is stored in the storage unit 340 in advance. In this layout information, position information of the answer column provided on the answer sheet 100 is described.
FIG. 7 is a diagram illustrating an example of layout information.
Referring to FIG. 7, the layout information includes each value of position information (COMin) at the left end of the answer column, position information (COMMax) at the right end, position information (LOMIN) at the upper end, and position information (LOMax) at the lower end. Each answer field is recorded in association with a question number that identifies the answer field.
The target answer column determination unit 320 reads out the layout information from the storage unit 340 and compares it with the position information calculated by the time-position information conversion unit 311. Thereby, it can be determined to which answer column of the answer sheet 100 the movement trajectory (writing operation) of the pen device 200 indicated by the stroke information is made. Instead of the layout information as shown in FIG. 7, image data including information such as the size of the answer sheet 100, the position and size of the answer column is prepared, and the image data and the time-position information conversion unit 311 The answer field corresponding to the stroke information may be specified by comparing with the calculated position information.

The database registration unit 330 receives the information obtained by the processes of the time-position information conversion unit 311, the stroke information acquisition unit 312, and the target answer column determination unit 320 together with the propagation time information from the pen device 200. To be registered in the database constructed in the storage unit 340.
FIG. 8 is a diagram illustrating a configuration example of a database.
Referring to FIG. 8, this database includes a transmission time (SentTime), propagation time information (Tul, Tur, Tll, Tlr), position information (CO, LO), and question number (Q) for each device ID of the pen device 200. ), Answer time (TQ) is registered. Here, the transmission time is the transmission time received from the pen device 200 together with the propagation time information. The answer time is a difference between the transmission time of the stroke information for the answer field specified by the question number and the transmission time of the stroke information for the answer field answered immediately before the question number.

The stroke information correction unit 350 corrects the stroke information generated by the stroke information acquisition unit 312 in order to increase the accuracy of processing by the subsequent incorrect answer pattern determination unit 360. In this correction, a graphic represented by the movement locus of the pen device 200 recognized based on the stroke information is converted into an image. An existing conversion algorithm can be appropriately used as the correction content and algorithm. Specifically, for example, there are a function for rotating an image based on stroke information, a function for enlarging or reducing, and the like. Further, when the pen pressure at the time of writing on the answer sheet 100 by the pen device 200 is detected and recognized as the density, the image based on the stroke information is also expressed in gray scale, so that the gray scale is converted into a binary image. You may make it provide the function to perform.
FIG. 9 is a diagram illustrating an example of correction by the stroke information correction unit 350. 9A shows an image based on the stroke information before correction. FIG. 7B is a diagram illustrating a state in which the image in FIG. (C) is a figure which shows a mode that the image of (B) was binarized.

  The fraudulent answer pattern determination unit 360 conducts fraudulent activities such as peeping at the answer of another examinee or teaching the answer to another examinee when an answer pattern by a particular examinee satisfies a predetermined condition. Is determined to have been performed. In order to make this determination, the incorrect answer pattern determination unit 360 includes a stroke information comparison unit 361, an inter-taker answer time comparison unit 362, a question-specific answer time comparison unit 363, and an answer order analysis unit 364.

  The stroke information comparison unit 361 compares the stroke information of each answer column for each examinee, and obtains the similarity of the shapes. As an algorithm for obtaining the similarity, an existing algorithm such as one using a mesh feature amount or a peripheral feature amount can be applied. The examinee is specified by the device ID. In this embodiment, since character recognition is not performed on the written image described in the answer column, the similarity in the shape of the image obtained from the stroke information is determined, but character recognition is performed on this image. It is also possible to go and determine the semantic similarity of written content. In this case as well, an existing algorithm may be used for determining the similarity.

The inter-taker answer time comparison unit 362 compares answer entry times of a plurality of examinees with respect to a specific answer column.
The question-specific answer time comparison unit 363 compares the time required for answering each question in a specific examinee. The time required for answering a predetermined question is obtained by obtaining the difference between the answer entry time for the question and the answer entry time for the question answered before that time for each examinee.
These answer entry times can be, for example, the transmission time of stroke information for this answer field.
The answer order analysis unit 364 compares the answer times to the answer fields in a specific examinee and identifies the order in which the answers were made. For example, as in the above example, the transmission times of the stroke information for each answer column are compared and sorted in order from the earliest time, and the question number corresponding to this stroke information is specified.

Based on the processing results of the above-described configurations, the incorrect answer pattern determination unit 360 may select an answer pattern (hereinafter referred to as an incorrect answer) that may be fraudulent from the answer patterns of each examinee stored in the storage unit 340. Pattern). Various methods can be considered as a method for determining an incorrect answer pattern. As a specific example, in the present embodiment, an incorrect answer pattern is determined in the following case.
1. The stroke information in the same answer field is similar among multiple examinees, and the answer entry time for the answer field is close.
In this case, there is a possibility that the answer entry time may be near due to a given examinee stealing another examinee's answer and answering it, so this may be an incorrect answer pattern. Judge. In determining whether the stroke information is similar, an appropriate threshold is set in advance, and if the similarity of the stroke information is within the threshold, it is determined that they are similar. Similarly, whether or not the answer entry time is near is also determined by comparison with a preset threshold value.

2. When the order of answers is irregular for a given examinee, and the stroke information in a certain number of answer columns is similar to other given examinees.
In this case, the answering candidate may have received an answer from a number of other candidates and the answering may have become irregular, so it is determined that this is an incorrect answering pattern. . The similarity determination of stroke information is performed by comparison with a threshold value as described above.

3. When the time required for answering a given question is significantly shorter than the average of all examinees and the stroke information in the answer column for that question is similar to that of another given examinee .
In this case, the applicable examinee will answer in a shorter time than the average of all the examinees because he / she has seen the answers of other examinees regarding the relevant questions or has been taught by other examinees. This is considered to be an incorrect answer pattern. Specifically, for all examinees, the time required to answer each question is obtained, and an average value is calculated for each question. Then, when the time required for answering a predetermined test taker is less than or equal to a predetermined threshold (for example, several percent or less) with respect to this average value, the test taker's answer pattern is determined as an incorrect answer pattern. It becomes.

  As mentioned above, although the determination method of three types of incorrect answer patterns was shown, it cannot be overemphasized that these are only illustrations. Actually, a determination method that is considered effective for determining an incorrect answer pattern may be applied as appropriate. Therefore, it is possible to add a method other than the above-described determination method, and it is not always necessary to apply all the above-described three types of determination methods.

In the present embodiment, after the incorrect answer pattern is detected as described above, the test supervisor corresponding to the incorrect answer pattern, that is, the examiner who may have performed the wrong act is identified and notified to the test supervisor. Encourage surveillance.
The fraud examiner identifying unit 370 refers to the database stored in the storage unit 340, acquires a device ID corresponding to the fraud answer pattern detected by the fraud answer pattern determination unit 360, and uses the pen device of this device ID. A test taker who uses 200 is specified, and identification information (test take number etc.) of the test taker is listed.
The notification image generation unit 380 images the list of test taker identification information created by the unauthorized test taker specifying unit 370 and transmits it to a monitor monitored by the test supervisor. As a result, the test supervisor can monitor the examinees registered in the list displayed on the monitor with particular attention.

FIGS. 10-1 to 10-5 are flowcharts showing the operation of the system of the present embodiment configured as described above.
As shown in FIG. 10A, in this system, first, the pen device 200 transmits a request radio wave (step 1001), and the wireless tags 110a to 110d receive the request radio waves and transmit reflected radio waves (step 1002, 1003), the pen device 200 receives this reflected radio wave (step 1004). When the reflected radio waves have been received from all the wireless tags 110a to 110d, the pen device 200 transmits the propagation time information Tul, Tur, Tll, Tlr, the transmission time Ts, and the device ID to the server 300. (Steps 1005 and 1006).

  When the server 300 receives each piece of information from the pen device 200, the time-position information conversion unit 311 performs an interpolation operation using the lookup table for the propagation time information Tul, Tur, Tll, Tlr, and the position information of the pen device 200. CO and LO are calculated (FIG. 10-2, step 1007). Then, the stroke information acquisition unit 312 generates the stroke information S based on the temporally continuous position information CO and LO (step 1008).

Next, the target answer column determination unit 320 performs a series of operations for determining whether or not the stroke information is an answer to a predetermined answer column in the answer sheet 100 from the position information CO and LO obtained in step 1007. .
In this operation, the target answer column determination unit 320 first sets a parameter i representing a question number, and sets i = 1 (step 1009). Here, the number of questions is n (that is, i: 1 ≦ i ≦ n). Then, by referring to the layout information as shown in FIG. 7, the position information COMin_I, COMax_I, LOMin_I, and LOMax_I in the answer column of the question Qi is acquired, and whether the position information CO and LO correspond to the position of the answer column. No, that is,

LOMin_i ≦ LO ≦ LOMax_i and COMin_i ≦ CO ≦ COMax_i?

It is checked whether or not this relationship holds (step 1010).

If the above relationship does not hold, the target answer field determination unit 320 next checks whether or not all answer fields have been processed, i.e., i = n (step 1011). If i has not reached n, i + 1 is set as a new i, and the determination in step 1010 is repeated (step 1012). On the other hand, if i = n, it means that the position where writing by the pen device 200 is performed is not included in any answer column. Therefore, the propagation time information Tul, Tur, Tll, Tlr received from the pen device 200 and the information obtained in the processing after step 1007 are discarded without being registered in the database, and waiting for the next input from the pen device 200 ( Step 1013).
In step 1010, if the position information CO, LO corresponds to the position of the answer column of question Qc (cεi), it is determined that the stroke information S obtained in step 1008 is an image representing the answer of question Qc. (Step 1014).

Next, the database registration unit 330 accesses the storage unit 340 based on the device ID sent together with the propagation time information, and reads a record of this device ID (step 1015). Then, assuming that the transmission time of the last record (the record for the data being processed) is Ts1 and the transmission time immediately before it is Ts0, the time TQc required to answer the question Qc is

TQc = Ts1-Ts0

(Step 1016). Then, the question number Qc and the answer time TQc are registered in the last record (step 1017).

  Next, the stroke information correction | amendment part 350 performs the correction process of the stroke information S produced | generated by step 1008 (FIG. 10-3, step 1018). This correction process is, for example, a process of rotating, enlarging or reducing an image based on the stroke information S, or binarizing, and can be performed using an existing conversion algorithm as appropriate as described above. After performing the correction process, the stroke information correction unit 350 stores the corrected image in the storage unit 340.

Next, the incorrect answer pattern determination unit 360 detects an incorrect answer pattern based on each piece of information generated so far.
First, the stroke information comparison unit 361 selects a pair of examinees A and B, obtains a difference between the stroke information SA of the examinee A and the stroke information SB of the examinee B, and compares it with a preset threshold value ST. . Specifically, a parameter k representing the question number (k: 1 ≦ k ≦ n) is set, and k = 1 is set (step 1019). And about the stroke information SA and SB for the answer column of the question Qk, the following formula

SA-SB <ST

Is checked (step 1020).

If the above relationship is established, the fact that the stroke information SA and the stroke information SB are similar in the answer to the question Qk is stored in the memory (step 1021), and another pair of examinees is newly taken as examinees A and B. The determination in step 1020 is repeated for question Qk (step 1022).
On the other hand, if the above relationship does not hold, the stroke information comparison unit 361 next refers to the database in the storage unit 340 to determine whether or not the determination in Step 1020 for the question Qk has been made for the pair of examinees in all combinations. Check (step 1023). If an unprocessed pair of examinees remains, the unprocessed pair of examinees is newly set as examinees A and B, and the determination in step 1020 for question Qk is repeated (step 1022).
If the determination in step 1020 for the question Qk is made for all the pairs of examinees, the stroke information comparison unit 361 next checks whether or not all the answer columns have been processed, that is, whether k = n. (Step 1024). If k has not reached n, the operation from step 1020 is repeated with k + 1 as a new k (step 1025).

  When k = n, the similarity of the stroke information is determined for all the answer columns in all pairs of examinees, and the stroke information similar to the threshold ST or more (hereinafter referred to as similar stroke information) is stored in the memory. It is remembered. Then, the process for detecting an incorrect answer pattern is performed by the answer time comparison part 362 between test takers, the answer time comparison part 363 by question, and the answer order analysis part 364 next.

  The inter-taker answer time comparison unit 362 refers to the database of the storage unit 340 and compares the answer entry times (that is, transmission times Ts) of each examinee for questions corresponding to similar stroke information (FIG. 10-4). Step 1026). If there is a question whose answer entry time difference is equal to or less than a certain threshold, the device ID corresponding to the answer entry time is stored in the memory (step 1027).

  Further, the question-specific answer time comparison unit 363 refers to the database of the storage unit 340 and obtains an average value of answer times for each question among all examinees (step 1028). Next, the answer time TQ for each question of each examinee is compared with the average value of the answer times of all examinees for that question (step 1029). If there is a question whose answer time is equal to or less than a certain threshold, the presence / absence of similar stroke information is examined for the stroke information related to the question (step 1030). If there is similar stroke information, the device ID corresponding to the stroke information is stored in the memory (step 1031).

  The answer order analysis unit 364 refers to the database in the storage unit 340 and checks the answer order for each examinee based on the answer entry time of each question. If the answer order is irregular, the stroke information related to each question is checked for the presence of similar stroke information (FIGS. 10-5, steps 1032 and 1033). If there is similar stroke information, the device ID corresponding to the stroke information is stored in the memory (step 1034).

  In the above operation example, the steps are described in the order of the inter-taker answer time comparison unit 362, the question-specific answer time comparison unit 363, and the answer order analysis unit 364. However, these processes are independent and are executed. The order is not limited to the above order. These processes may be performed in parallel. Furthermore, after the processing of the inter-taker answer time comparing unit 362, the answer time comparing unit by question 363, and the answer order analyzing unit 364, the process is performed in a procedure in which the similarity of the stroke information is judged only for the required examinees and questions. It is also possible to do this.

  Next, based on the device ID stored in the memory by the inter-taker answer time comparison unit 362, the question-specific answer time comparison unit 363, and the answer order analysis unit 364, the incorrect answer pattern determination unit 360 uses the pen device of this device ID. A test taker who uses 200 is specified, and identification information (test take number etc.) of the test taker is registered in the list of fraud test takers (step 1035). Then, the notification image generation unit 380 images the unauthorized examinee list and displays it on the monitor monitored by the test supervisor (step 1036).

In the above operation example, the transmission time when the pen device 200 transmits data to the server 300 is the answer entry time using the pen device 200. However, the time when the writing operation is actually performed in the pen device 200 is used. It may be used. In this case, when the writing pressure detection unit 220 detects that the writing unit 210 is pressed against the answer sheet 100, the processing unit 241 of the control unit 240 refers to the clock 242 and stores the detection time in the memory 243. To do. And when the communication part 250 transmits propagation time information, the detection time of the writing operation corresponding to this propagation time information is read from the memory 243 and transmitted.
Further, in the above operation example, the processing by the incorrect answer pattern determination unit 360, such as comparison of stroke information, is performed for all examinees. The examinees may be divided into groups, and the above operation may be performed on the examinees in the group for each group.

Next, an example of a specific implementation of the system according to this embodiment will be described.
FIG. 11 is a diagram showing the overall configuration of the fraud detection system according to the present embodiment.
As shown in FIG. 11, the examinee enters the answer of the test on the answer sheet 100 provided by the system, using the pen device 200 provided by the system. Information such as propagation time information is transmitted from the pen device 200 to the server 300 via the LAN. An access point 390 is provided at the test site, receives information wirelessly transmitted from the pen device 200, and sends it to the server 300. In addition, a camera 400 is installed at the test venue, and the state of the examinee is photographed. The captured video is displayed on the monitor 500 monitored by the test supervisor. When an incorrect answer pattern is detected by the processing of the server 300, the monitor 500 also displays an image of a list of unauthorized examinees generated by the notification image generation unit 380.

  In the configuration shown in FIG. 11, the notification image generation unit 380 not only generates a list image of unauthorized examinees, but also selects an image of an unauthorized examinee among the examinees photographed by the camera 400 and displays the list image. And can be displayed on the monitor 500. In this case, the test taker can be specified by associating the seat position of each test taker with the device ID of the pen device 200, for example. For example, in the case where control is performed such that images of each examinee are displayed on the monitor 500 while sequentially switching, the examinee is registered in the unauthorized examinee list when displaying the unauthorized examinee. A message can be displayed to notify the test supervisor.

  Thus, according to the present embodiment, examinees who may be cheating can be automatically detected from their answer patterns. Furthermore, by notifying the test supervisor of this detection result, it is possible to call attention to a test taker who may be engaged in a wrongdoing in monitoring the cheating by the test supervisor. Therefore, it is possible to efficiently monitor the examinee's fraud and to reduce the burden on the examination supervisor.

It is a figure which shows the basic composition which implement | achieves the system of this embodiment. It is a figure which shows the structural example of the answer sheet used by this embodiment. It is a figure explaining the concept of the interpolation calculation which calculates | requires the position of a pen device in this embodiment. It is a figure which shows the structure of the pen device used by this embodiment. It is the figure which showed typically the hardware structural example of the computer suitable for implement | achieving the server of this embodiment. It is a figure which shows the function structure of the server in this embodiment. It is a figure which shows the example of the layout information of the answer sheet used by this embodiment. It is a figure which shows the structural example of the database used by this embodiment. It is a figure which shows the example of correction | amendment by the stroke information correction | amendment part of this embodiment. It is a flowchart which shows operation | movement of the system of this embodiment. It is a flowchart which shows operation | movement of the system of this embodiment. It is a flowchart which shows operation | movement of the system of this embodiment. It is a flowchart which shows operation | movement of the system of this embodiment. It is a flowchart which shows operation | movement of the system of this embodiment. It is a figure which shows the whole structure of the fraud detection system by this embodiment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 ... Answer sheet, 110a-110d ... Wireless tag, 200 ... Pen device, 210 ... Writing part, 220 ... Writing pressure detection part, 230 ... Radio wave detection part, 240 ... Control part, 241 ... Processing part, 242 ... Clock, 243 ... Memory, 250 ... Communication unit, 300 ... Server, 311 ... Time-position information conversion unit, 312 ... Stroke information acquisition unit, 320 ... Target answer column determination unit, 330 ... Database registration unit, 340 ... Storage unit, 350 ... Stroke Information correction unit, 360 ... Incorrect answer pattern determination unit, 361 ... Stroke information comparison unit, 362 ... Inter-taker answer time comparison unit, 363 ... Answer time comparison unit by question, 364 ... Answer order analysis unit, 370 ... Unauthorized examinee Identification unit, 380 ... notification image generation unit

Claims (9)

A stroke information acquisition unit that acquires stroke information that is information related to the operation of the pen device on the surface of the answer sheet, based on the data received from the pen device;
A database registration unit that sorts the stroke information acquired by the stroke information acquisition unit and the operation time of the pen device for each examinee and registers them in a database;
An incorrect answer pattern that compares the stroke information of a plurality of examinees with reference to information registered in the database by the database registration unit, and detects an answer pattern that may be fraudulent based on the similarity A fraud detection server comprising a determination unit.   The incorrect answer pattern determination unit is similar to the stroke information in at least a part of the answer sheet between a plurality of examinees, and the operation time of the pen device for the part is within a predetermined threshold. The fraud detection server according to claim 1, wherein an answer pattern including the stroke information is determined as an incorrect answer pattern for at least one of the plurality of examinees. A target answer field determination unit that identifies an answer field of the answer sheet to which the stroke information acquired by the stroke information acquisition unit corresponds;
The database registration unit registers the identification information of the answer field specified by the target answer field determination unit in the database in association with the stroke information,
The incorrect answer pattern determination unit obtains an answer order for each examinee from the identification information in the answer column and the operation time of the pen device for the part, and the answer order is irregular for a predetermined examinee, and other When the stroke information corresponding to a certain number of answer fields is similar to the examinee, the answer pattern including the stroke information is determined as an incorrect answer pattern for the predetermined examinee. Item 12. The fraud detection server according to item 1. A target answer field determination unit that identifies an answer field of the answer sheet to which the stroke information acquired by the stroke information acquisition unit corresponds;
The database registration unit obtains time information required for each question from the answer field specified by the target answer field determination unit and the operation time information of the pen device, and associates the information with the stroke information. Register in the database,
The incorrect answer pattern determination unit, in a predetermined examinee, when the time required for answering a predetermined question is shorter than the average, and the stroke information in the answer column of the question is similar to other examinees The fraud detection server according to claim 1, wherein for the predetermined examinee, the answer pattern including the stroke information is determined as an incorrect answer pattern. Information on propagation time of radio waves transmitted and received between the wireless tags provided at a plurality of locations on the answer sheet and the pen device is acquired from the pen device, and information on the propagation time is obtained on the paper surface of the answer sheet. A time-position information conversion unit for converting the information into the position information of the pen device;
The fraud detection server according to claim 1, wherein the stroke information acquisition unit generates the stroke information based on the position information of the pen device generated by the time-position information conversion unit. An unauthorized test taker specifying unit that specifies an examinee corresponding to an answer pattern determined as an answer pattern having a possibility of an illegal act by the incorrect answer pattern determining unit, based on identification information of the pen device;
The fraud detection server according to claim 1, further comprising: a notifying unit that notifies information on a test taker specified by the test taker specifying unit. Stroke information acquisition means for acquiring stroke information which is information relating to the operation of the pen device on the surface of the answer sheet;
Registration means for sorting the stroke information acquired by the stroke information acquisition means for each examinee and registering them in a database;
Incorrect answer pattern detection that compares the stroke information of a plurality of examinees with reference to the information registered in the database by the registration means, and detects an answer pattern that may be fraudulent based on the similarity Means,
An injustice characterized by comprising: a notifying means for specifying a test taker corresponding to an answer pattern determined as an answer pattern having a possibility of cheating by the incorrect answer pattern detecting means and notifying the test taker's information Action detection system. Image acquisition means for photographing the examinee;
Display means for displaying an image of the examinee taken by the image acquisition means,
The notifying means generates an image for notifying the specified information of the examinee, combines it with an image of the examinee taken by the image acquisition means, and causes the display means to display the image. Item 12. The fraud detection system according to Item 7. A computer data processing method for receiving and processing data from a pen device,
The computer acquires stroke information, which is information related to the operation of the pen device on the surface of the answer sheet, based on data received from the pen device;
The computer sorting the stroke information for each examinee and registering it in a database;
The computer refers to information registered in the database, compares the stroke information of a plurality of examinees, and detects an answer pattern that may be fraudulent based on the similarity. A data processing method.

Images