JP2018128849A - Determination program, determination method, and determination device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To enable a precise determination on whether or not an area where a difference in population between in the daytime and the night exceeds a reference is a disaster site.SOLUTION: As an example case in which an area where a population in the daytime is larger than a population in the night by a reference, cited is SHIBUYA district. A determination device 101 reflects, on a determination process 102 for determining whether or not SHIBUYA district is a disaster site, a number of tweets relating to a disaster in NAKANO district, SHINJUKU district, MINATO district, SHINAGAWA district, MEGURO district, SETAGAYA district, and SUGINAMI district that are peripheral city, district, town and village to SHIBUYA district. Since an occurence of disaster at SHIBUYA district has an influence on the number of tweets at the peripheral city, district, town and village to SHIBUYA district, the determination device 101 executes the determination process 102 on which the number of tweets relating to the disaster at the peripheral city, district, town and village, thereby precisely determining whether or not SHIBUYA district is a disaster site.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a determination program, a determination method, and a determination apparatus.

  2. Description of the Related Art Conventionally, there is a technology for analyzing disaster information from the content posted by a social media user and determining the disaster area where the disaster occurred from the profile information of the user who posted the information related to the disaster. As related prior art, for example, the occurrence of a disaster in a monitored area is estimated from the number of disaster information transmitted in a monitored area including a plurality of municipalities according to the selection rule of the municipality, and disaster information for each municipality including the monitored area Some select local governments where disasters are estimated based on the number of messages sent. Also, the location of the event is extracted from the messages related to the predetermined event obtained from social media, and the number of messages at each location is accumulated in descending order of the number of messages. The rate of increase in the number of messages is the average rate of increase. Some output the location of occurrence from the change point.

JP 2006-153971 A JP2015-207106A

  However, according to the prior art, the determination of a disaster area may be erroneous in an area where the difference in population between day and night exceeds the standard. Specifically, the user's profile information often describes the area where the user resides rather than the current location of the user. And when a disaster occurs in an area where the daytime population is larger than the nighttime population, it is determined that the information about the disaster posted by the user is information on the area where the user lives, so the daytime population is higher than the standard. It becomes difficult to determine that many areas are disaster areas.

  In one aspect, an object of the present invention is to provide a determination program, a determination method, and a determination apparatus that can accurately determine whether an area where the population difference between day and night exceeds a standard is a disaster area. To do.

  In one embodiment, the determination program, the determination method, and the determination apparatus determine whether a disaster area is a disaster area when determining a disaster area according to a disaster information transmission situation in social media. When responding to an area where the population difference between day and night exceeds the standard, the transmission status of disaster information in the surrounding area of the target area is reflected in the determination process for determining whether the target area is a disaster area.

  In one aspect, the present invention makes it possible to accurately determine whether an area where the population difference between day and night exceeds a standard is a disaster area.

FIG. 1 is an explanatory diagram illustrating an operation example of the determination apparatus 101 according to the present embodiment. FIG. 2 is an explanatory diagram illustrating a configuration example of the disaster area determination system 200. FIG. 3 is an explanatory diagram illustrating a hardware configuration example of the determination apparatus 101. FIG. 4 is an explanatory diagram illustrating a functional configuration example of the determination apparatus 101. FIG. 5 is an explanatory diagram showing an example of the number of tweets when Shibuya Ward is devastating. FIG. 6 is an explanatory diagram showing an example of the number of tweets when Shibuya Ward does not have a disaster but one of the surrounding municipalities has a disaster. FIG. 7 is an explanatory diagram showing an example of class classification by SVM. FIG. 8 is a flowchart illustrating an example of a processing procedure for creating a list of surrounding municipalities. FIG. 9 is a flowchart illustrating an example of an SVM learning process procedure. FIG. 10 is a flowchart illustrating an example of the target municipality disaster determination process.

  Embodiments of a disclosed determination program, determination method, and determination apparatus will be described below in detail with reference to the drawings.

(Operation example of determination apparatus 101 according to the present embodiment)
FIG. 1 is an explanatory diagram illustrating an operation example of the determination apparatus 101 according to the present embodiment. The determination apparatus 101 is a computer that determines a disaster area that is an area where a disaster has occurred. Specifically, the determination apparatus 101 determines a disaster area according to the transmission status of disaster information on social media. The determination apparatus 101 is a server, for example. Examples of disasters include river flooding, flooding disasters, earth and sand disasters, fires, and earthquakes. Moreover, the unit of the area determined as a disaster area is, for example, an administrative unit such as a municipality or a special ward in Japan, or an administrative unit corresponding to a Japanese municipality or special ward in a country other than Japan. Hereinafter, the municipalities and special wards are collectively referred to as “municipalities”. The regional unit may be a superlative administrative unit that includes municipalities such as prefectures.

  Social media is a general term for websites, network services, and the like having elements that create content based on information transmitted by users and connections between users. For example, social media includes SNS (Social Networking Service), blog, video sharing service, photo sharing service, and the like. The SNS is a service that enables construction of a social network, that is, a social network on the Web. SNS includes, for example, Twitter (registered trademark) and Facebook (registered trademark).

  The transmission status of disaster information is indicated, for example, by the number of information transmitted regarding the disaster. The information transmitted regarding the disaster is, for example, a sentence posted to the SNS and includes a character string indicating the disaster. Alternatively, the information transmitted regarding the disaster may be image information posted to the photo sharing service and image information showing the state of the disaster. In addition, the transmission status of the disaster information may be the number of people who have transmitted the information transmitted regarding the disaster. In the following description, it is assumed that the transmission status of disaster information is a short sentence posted to the SNS and is indicated by a short sentence including a character string indicating the disaster. A short sentence posted to the SNS is referred to as “tweet”.

  As a method for determining the disaster area according to the transmission status of the disaster information in social media, for example, the following first to fourth processes can be considered.

First process: Collecting tweets about disasters.
2nd process: The place contributed with respect to each collected tweet is estimated.
Third process: Count the number of tweets related to disasters for each prefecture, compare the number of tweets normalized by population in statistical processing, and the prefectures determined to be likely to occur with a certain threshold or less. Judge that the disaster has occurred.
Fourth processing: In the prefectures determined to be in disaster, the number of tweets is arranged in descending order for each municipality to form a cumulative composition ratio curve of the number of tweets, and the increase in the number of tweets becomes gentle, Specifically, it is determined that the municipality up to the point where the increase rate is equal to or less than the threshold is a disaster area.

  Here, in the third process and the fourth process, in order to determine that the area has a disaster, in the first process and the second process, the disaster in the corresponding area You will collect tweets about. And as a method of estimating the posted location in the second process, the region registered in the profile among the region information registered in the user's profile and the word related to the region included in the tweet text Prioritize information. The reason for not giving priority to the words related to the region included in the tweet text is that the tweet text is likely to be swayed, such as a character string with the word related to the region omitted. This is because it is difficult.

  As described above, as a result of estimating the area information registered in the profile as the place where the tweet is posted, the determination of the disaster area may be erroneous in an area where the population difference between day and night is large. Specifically, the user's profile information often describes the area where the user resides rather than the current location of the user. And, when a disaster occurs in an area where the population difference between day and night is large, specifically in an area where the daytime population is larger than the standard than the nighttime population, the tweet posted by the user is a tweet for the area where the user lives Will be judged. As a result, in areas where the daytime population is greater than the standard than the nighttime population, many tweets will not be tweeted. As a result, an area where the daytime population is larger than the standard than the nighttime population is less likely to be determined as a disaster area. For example, a tweet posted by a person coming from outside is not considered to be from the above area even if the disaster actually occurs in an area where there are many people moving from outside area in the daytime such as Tokyo. There are cases where it differs from the actual tweet posting location.

  Therefore, in the present embodiment, when executing the determination process for determining the disaster area by regarding the place of residence as the posting location of the tweet, if the area to be determined is an area where the daytime population is larger than the reference than the nighttime population, The following describes how to reflect the number of tweets in the surrounding area of the target area in the determination process.

  An operation example of the determination apparatus 101 will be described with reference to FIG. In FIG. 1, an example of Shibuya Ward is shown as an area where the daytime population is greater than the standard than the nighttime population. Even if there is actually a disaster in Shibuya Ward, Shibuya Ward has many people from the surrounding area, and the population that actually lives is small, so if you perform the above first to fourth processing, Instead of Shibuya Ward, it is determined that the surrounding municipalities are in disaster. In FIG. 1, it is assumed that the target area for performing the determination process 102 for determining whether or not a disaster area is Shibuya Ward.

  When the target area corresponds to an area where the difference in population between day and night exceeds the standard, the determination apparatus 101 determines whether or not the target area is a disaster area. This is reflected in the determination processing 102. Here, the reason for reflecting the transmission status of the disaster information in the surrounding area in the determination processing 102 is that the occurrence of the disaster in the target city affects the transmission status of the disaster information in the surrounding area. Hereinafter, the area is a unit of the municipality, the target area is referred to as “target municipality”, and the surrounding area is referred to as “neighboring municipality”.

  Here, as a municipality exceeding the standard of the difference in daytime and nighttime, the determination apparatus 101 may, for example, select a municipality in which the difference in daytime and nighttime population exceeds a predetermined value set by the administrator of the determination apparatus 101. It is determined that the municipality exceeds the standard of population difference between day and night. In addition, to correspond to a municipality that exceeds the standard of the population difference between day and night, the target municipality may be an area that exceeds the standard of population difference between day and night, or the target municipality It may be included in an area that exceeds the standard of population difference between day and night.

  The surrounding municipalities are, for example, municipalities included in a predetermined range from the target municipality. Here, the predetermined range is a value set by an administrator of the determination apparatus 101, for example. In the example of FIG. 1, the municipalities where the government office is included in the range Larea within the radius L [km] of Shibuya Ward are the surrounding municipalities. Specifically, the surrounding areas are Nakano Ward, Shinjuku Ward, Minato Ward, Shinagawa Ward, Meguro Ward, Setagaya Ward and Suginami Ward. In the example of FIG. 1, Nakano Ward, Shinjuku Ward, Minato Ward, Shinagawa Ward, Meguro Ward, Setagaya Ward, and Suginami Ward are areas determined to have been affected. In FIG. 1, the area to which shading is given is an area that has been determined to have occurred.

  In the example of FIG. 1, as shown in FIG. 1 (1), the determination apparatus 101 determines the transmission status of disaster information in the surrounding municipalities of Shibuya Ward, and determines whether or not Shibuya Ward is a disaster area 102. To reflect. Here, in FIG. 1, tweets tw1, tw2,... Transmitted regarding disasters in Setagaya Ward are shown as information indicating the transmission status of disaster information in surrounding municipalities in Shibuya Ward. For example, the determination apparatus 101 reflects the number of tweets transmitted regarding disasters in Nakano Ward, Shinjuku Ward, Minato Ward, Shinagawa Ward, Meguro Ward, Setagaya Ward, and Suginami Ward in the determination processing 102. Here, to reflect the above-described number of tweets in the determination process 102 is to specifically use the above-described number of tweets as one of the arguments of the determination process 102.

  As a specific example of the determination process 102, the determination apparatus 101 includes the number of tweets for each municipality when the target municipality is in a disaster, and the surrounding municipalities that are not affected by the target municipality. Machine learning is performed using the number of tweets for each city at the time of the disaster as teacher data. As the machine learning, for example, a support vector machine (SVM) can be employed. Next, when the determination apparatus 101 detects a disaster even in one of the target municipalities, the target municipality is also affected based on the result of machine learning and the number of tweets for each neighboring municipality to be reflected. It is determined whether or not More specific details of the determination process 102 will be described with reference to FIG.

  As described above, since the disaster of the target municipality affects the number of tweets in the surrounding cities, the determination apparatus 101 executes the determination process 102 by reflecting the number of tweets in the surrounding cities. As a result, the determination accuracy can be improved. The reason for using the number of tweets of the surrounding municipalities within the radius L [km] is that the surrounding municipalities are in the vicinity of the target municipalities. This is because the number of tweets in a municipality can be a meaningful increase, not a coincidence.

  In the description of FIG. 1, in an area where the difference in daytime and nighttime population is large, the area where the daytime population is larger than the standard is determined as the target area of the determination process 102, but the nighttime population is the standard rather than the daytime population. More areas, so-called bed towns, may be targeted areas. Next, a system including the determination apparatus 101 will be described with reference to FIG.

(Configuration example of the disaster area determination system 200)
FIG. 2 is an explanatory diagram illustrating a configuration example of the disaster area determination system 200. The disaster area determination system 200 includes a determination device 101, a social media management server 201, and a mobile terminal 202. The determination apparatus 101, the social media management server 201, and the mobile terminal 202 are connected via a network 210 such as the Internet, a LAN (Local Area Network), and a WAN (Wide Area Network).

  The social media management server 201 is a server that provides a social media service. Specifically, the social media management server 201 accumulates tweets transmitted from the mobile terminal 202 by the operation of the user U, and determines a tweet corresponding to the search condition of the search request according to the search request of the determination device 101. Return to 101. The portable terminal 202 is a computer operated by the user U. The mobile terminal 202 is, for example, a smartphone or a mobile phone. Next, the hardware configuration of the determination apparatus 101 will be described with reference to FIG.

(Example of hardware configuration of determination apparatus 101)
FIG. 3 is an explanatory diagram illustrating a hardware configuration example of the determination apparatus 101. In FIG. 3, the determination apparatus 101 includes a CPU (Central Processing Unit) 301, a ROM (Read-Only Memory) 302, and a RAM (Random Access Memory) 303. The determination apparatus 101 includes a disk drive 304 and a disk 305, a communication interface 306, and a display 307. Further, the CPU 301 to the disk drive 304, the communication interface 306, and the display 307 are connected by a bus 310, respectively.

  The CPU 301 is an arithmetic processing device that governs overall control of the determination device 101. The ROM 302 is a nonvolatile memory that stores programs such as a boot program. A RAM 303 is a volatile memory used as a work area for the CPU 301.

  The disk drive 304 is a control device that controls reading and writing of data with respect to the disk 305 according to the control of the CPU 301. As the disk drive 304, for example, a magnetic disk drive, an optical disk drive, a solid state drive, or the like can be adopted. The disk 305 is a nonvolatile memory that stores data written under the control of the disk drive 304. For example, when the disk drive 304 is a magnetic disk drive, a magnetic disk can be adopted as the disk 305. Further, when the disk drive 304 is an optical disk drive, an optical disk can be adopted as the disk 305. When the disk drive 304 is a solid state drive, a semiconductor memory formed by a semiconductor element, that is, a so-called semiconductor disk can be used as the disk 305.

  The communication interface 306 is a control device that controls an internal interface with the network 210 and controls input / output of data from other devices. Specifically, the communication interface 306 is connected to another device via the network 210 through a communication line. As the communication interface 306, for example, a modem or a LAN adapter can be employed.

  The display 307 displays data such as a document, an image, and function information as well as a cursor, an icon, or a tool box. For example, the display 307 can employ a CRT (Cathode Ray Tube), a TFT (Thin Film Transistor) liquid crystal display, a plasma display, or the like.

  When the administrator of the determination apparatus 101 directly operates the determination apparatus 101, the determination apparatus 101 may include hardware such as a keyboard and a mouse. The social media management server 201 has a CPU, ROM, RAM, disk drive, disk, and communication interface. Similarly, the mobile terminal 202 includes a CPU, ROM, RAM, disk drive, disk, communication interface, and touch panel. A touch panel is a device that combines a sensor that detects a tap operation or a flick operation and a display that displays data such as a document, an image, and function information including a cursor, an icon, or a tool box. Specifically, the sensor is arranged on the display. The sensor is a sensor that employs, for example, a resistance film method, a surface type or a projection type capacitance method. For example, a TFT liquid crystal display can be adopted as the display. Next, a functional configuration example of the determination apparatus 101 will be described with reference to FIG.

(Example of functional configuration of determination apparatus 101)
FIG. 4 is an explanatory diagram illustrating a functional configuration example of the determination apparatus 101. The determination apparatus 101 includes a control unit 400. The control unit 400 includes a tweet collection unit 401, a reflection unit 402, a determination unit 403, a determination unit 404, a specification unit 405, and an output unit 406. The control unit 400 realizes the functions of the respective units when the CPU 301 executes a program stored in the storage device. Specifically, the storage device is, for example, the ROM 302, the RAM 303, the disk 305, etc. shown in FIG. The processing results of each unit are stored in the RAM 303, the CPU 301 register, the CPU 301 cache memory, or the like.

  Further, the determination apparatus 101 can access the storage unit 410. The storage unit 410 is stored in a storage device such as the RAM 303 and the disk 305.

  The tweet collecting unit 401 collects tweets about disasters from the social media management server 201. The collected tweets are accumulated in the storage unit 410. Thereby, the tweet collecting unit 401 can provide the collected tweet to the reflecting unit 402, the determining unit 404, and the specifying unit 405.

  The reflection unit 402 causes the determination process 102 to reflect the target municipality that evaluates whether it is a disaster area or not in a region where the difference in population between day and night exceeds the standard. Specifically, if the target city corresponds to an area where the difference in population between day and night exceeds the standard, the reflection unit 402 adds the number of tweets of surrounding cities to the determination process 102 executed by the determination unit 404. Add to argument. Thereby, the reflection unit 402 can cause the determination unit 404 to execute the determination process 102 in consideration of the number of tweets in the surrounding municipalities.

  The determination unit 403 determines whether to reflect in the determination process 102 according to the time zone when the disaster occurred. For example, the determination unit 403 reflects the number of tweets of the surrounding municipalities in the determination process 102 if the time when the disaster occurred is daytime, and if the time zone where the disaster occurred is nighttime, The number of tweets is not reflected in the determination process 102. In addition, when the time when the disaster occurred is a weekday daytime, the determination unit 403 reflects the number of tweets of the surrounding municipalities in the determination processing 102, and the time when the disaster occurred is a holiday, a holiday, or a weekday. At night, the number of tweets in the surrounding municipalities may not be reflected in the determination process 102. Thereby, the determination part 403 can perform the suitable determination process 102 according to the time slot | zone when the disaster occurred, for example.

  When the reflecting unit 402 reflects the number of tweets of the surrounding municipalities, the determination unit 404 uses the number of tweets of the surrounding municipalities of the target municipality to determine whether the target municipality is a disaster area. A determination process 102 for determining whether or not is executed. Moreover, the determination part 404 may perform the determination process 102, when the determination part 403 determines to reflect the number of tweets of a surrounding municipality. Moreover, the determination part 404 performs the 1st-4th process shown in FIG. 1, when not reflecting the number of tweets of a surrounding municipality. Thereby, the determination unit 404 can provide a determination result as to whether or not the target city is a disaster area.

  The identifying unit 405 refers to the storage unit 410, executes SVM from a data group belonging to either the first class or the second class, and classifies the SVM for classifying the first class and the second class. By executing, a boundary for classifying the first class and the second class is specified. SVM is a pattern recognition technique for classifying given vector data into two classes. The data belonging to the first class is the number of tweets transmitted regarding past disasters in surrounding municipalities when a past disaster occurred in the target municipality. In addition, data belonging to the second class was transmitted regarding past disasters in the surrounding municipalities when past disasters occurred in the surrounding municipalities, although no past disasters occurred in the target municipalities. The number of tweets.

  Further, the specifying unit 405 uses vector data whose component is the number of tweets for each neighboring city. Therefore, the vector data is data of the dimension of the number of surrounding municipalities. For example, the vector data is two-dimensional vector data if there are two neighboring municipalities, and is three-dimensional vector data if there are three neighboring municipalities. Therefore, the boundary specified by the specifying unit 405 is a boundary that divides a dimension space of the number of surrounding municipalities. For example, the boundary specified by the specifying unit 405 is a straight line or a curve if there are two neighboring municipalities, and a plane or a curved surface if there are three neighboring municipalities. More specifically, if there are two surrounding municipalities, the specifying unit 405 specifies a straight line expression indicating a boundary for classifying the first class and the second class. As a result, when there is a new disaster, the identification unit 405 determines whether the number of tweets for each neighboring city is classified into the first class or the second class. It can be specified whether or not is a disaster area.

  Then, when a disaster occurs, the determination unit 404 determines whether the target municipality is a disaster area based on the identified boundary and the number of information transmitted regarding disasters in other regions as determination processing. Determine if.

  In addition, the identification unit 405 uses the data belonging to the first class, except for the areas where the difference in population between day and night in the surrounding municipalities when the past disaster occurred in the target municipality exceeds the standard. It may be the number of tweets sent about disasters in the area. Similarly, the specifying unit 405 uses the past data in the above-mentioned other area when the past disaster has occurred in the other city but the past disaster has not occurred in the target city to the data belonging to the second class. It may be the number of tweets sent about the disaster. In this case, when a disaster occurs, the determination unit 404 determines whether the target city is a disaster area based on the identified boundary and the number of tweets transmitted regarding the disaster in the other area described above as the determination process 102. It is determined whether or not.

  The output unit 406 outputs a determination result of whether the target city is a disaster area by the determination unit 404. The output format is, for example, display on the display 307, transmission to an external device via the communication interface 306, or storage in a storage area such as the RAM 303 or the disk 305. In addition, the administrator of the determination apparatus 101 can browse the determination result based on the output of the output unit 406, and can ensure safety and confirm safety according to the determination result. For example, when the administrator of the determination apparatus 101 browses a determination result indicating that the target city is a disaster area, the administrator of the target device can ensure safety and confirm safety.

  Next, explanation will be made with reference to FIGS. The municipalities within the radius Lkm of Shibuya Ward are Nakano Ward, Shinjuku Ward, Meguro Ward, Suginami Ward, Setagaya Ward and Minato Ward.

(An example of the number of tweets in the target city and surrounding cities)
FIG. 5 is an explanatory diagram showing an example of the number of tweets when Shibuya Ward is devastating. FIG. 6 is an explanatory diagram showing an example of the number of tweets when Shibuya Ward has not occurred but one of the surrounding municipalities has occurred.

  A table 500 shown in FIG. 5 shows a list of the number of tweets when the Shibuya Ward actually occurs in the daytime. A table 500 illustrated in FIG. 5 includes records 501-1 to 50-1. As Table 500 shows, the number of tweets is relatively large in Setagaya Ward, Suginami Ward, Nakano Ward, and Meguro Ward.

  A table 600 shown in FIG. 6 shows an example of the number of tweets when the Shibuya Ward does not have a disaster in the daytime but any of the surrounding municipalities has a disaster. A table 600 illustrated in FIG. 6 includes records 601-1 to 601-7. As Table 600 shows, the number of tweets in Nakano Ward and Meguro Ward is not increasing, and the number of tweets is relatively large only in Setagaya Ward and Suginami Ward.

  As can be seen from the results of comparing Table 500 and Table 600, it is unclear whether Shibuya Ward is also devastating simply by showing a tendency for the number of tweets to increase only in Setagaya Ward and Suginami Ward. Then, when there is a tendency for the number of tweets to increase in Setagaya Ward, Suginami Ward, Nakano Ward, and Meguro Ward, it is possible to determine that Shibuya Ward is also affected. As a specific determination method, SVM can be adopted. The SVM will be described with reference to FIG.

(An example of classification by SVM)
FIG. 7 is an explanatory diagram showing an example of class classification by SVM. A graph 701 illustrated in FIG. 7 is a diagram illustrating the concept of SVM. In the learning stage, the determination apparatus 101 obtains a boundary ω that divides the two classes as shown in the graph 701 from the teacher data including data that is examples of the two classes. In the graph 701, data indicated by white circles indicates data belonging to the first class of the two classes, and data indicated by black triangles indicates the second class of the two classes. Indicates data belonging to. As shown by the graph 701, it can be seen that the data is classified into the first class and the second class by the boundary ω. The graph 701 shows two vector data of x and y, but the boundary ω can be obtained for three or more vector data.

  In the determination process 102, the determination device 101 uses the boundary ω calculated in advance when the actual data is given to which one of the first class and the second class. Classify or calculate. In the present embodiment, the determination apparatus 101 uses vector data in which the number of tweets for each neighboring city is arranged.

  For example, in the learning stage, the determination apparatus 101 sets the number of tweets for each surrounding city when Shibuya Ward occurs as the first class, and Shibuya City is not affected but is one of the surrounding cities. Teacher data is created with the number of tweets in each neighboring city at the time of the disaster as the second class. For the determination, the determination apparatus 101 calculates which class to classify from the boundary ω for class classification and the number of tweets for each surrounding municipality that is actual data.

  Next, flowcharts performed by the determination apparatus 101 will be described with reference to FIGS. Moreover, before the surrounding municipalities list creation process shown in FIG. 8, the determination apparatus 101 obtains the nighttime population and the daytime population, and the daytime population is larger than the reference and the difference is equal to or greater than a certain amount than the nighttime population. A list of municipalities, that is, a list of municipalities that are difficult to determine as having occurred is stored. For example, the list of target municipalities may be created by inputting the list of target municipalities by an administrator of the determination apparatus 101, for example. Or the determination apparatus 101 may acquire a night population and a daytime population from the result of a national tax survey etc. via the network 210, and may produce | generate the list of object municipalities.

(Neighboring city list creation process procedure)
FIG. 8 is a flowchart illustrating an example of a processing procedure for creating a list of surrounding municipalities. The surrounding city list creation process is a process of creating a surrounding city list for the target city.

  The determination apparatus 101 selects the first target city among all the target cities (step S801). Next, the determination apparatus 101 determines whether or not a list of neighboring cities has been created for all target cities (step S802). If there is still a target municipality for which a list of neighboring municipalities has not been created (step S802: No), the determination apparatus 101 acquires latitude / longitude information of the government office of the selected target municipality (step S803). . Next, the determination apparatus 101 creates a list of neighboring cities, towns and villages having a government office within a radius L [km] from the acquired latitude / longitude information (step S804). In the process of step S804, the determination apparatus 101 may or may not include the target city in the surrounding city list. Then, the determination apparatus 101 selects the next target city from all the target cities (step S805), and proceeds to the process of step S802.

  When a list of surrounding cities is created for all target cities (step S802: Yes), the determination apparatus 101 ends the surrounding city list creation process.

(SVM learning process procedure)
FIG. 9 is a flowchart illustrating an example of an SVM learning process procedure. The SVM learning process is a process in which teacher data is created and learning is performed in the SVM when the target municipality is in a disaster and when the target city is not in a disaster.

  The determination apparatus 101 selects the first target city among all the target cities (step S901). Next, the determination apparatus 101 determines whether or not class classification by SVM has been performed for all target municipalities (step S902). If there is a target municipality that has not been classified by SVM (step S902: No), the determination apparatus 101 creates teacher data (step S903). The above-mentioned teacher data has the first class as the number of tweets for each neighboring city when the selected target city is struck, and the selected target city is not struck, but the neighboring city is This is data in which the number of tweets for each neighboring city when any of the disasters occurred is a second class. Next, the determination apparatus 101 specifies an expression representing the boundary ω for class classification by using SVM from the created teacher data (step S904). And the determination apparatus 101 selects the next target municipality among all the target municipalities (step S905).

  When classifying by SVM is performed for all target municipalities (step S902: Yes), the determination apparatus 101 ends the SVM learning process.

(Target municipality disaster assessment process)
FIG. 10 is a flowchart illustrating an example of the target municipality disaster determination process. The target municipality disaster determination is a process for determining whether or not the target municipality has occurred, and corresponds to the determination process 102.

  The determination apparatus 101 determines whether or not the time when the disaster occurred is daytime (step S1001). When the time when the disaster occurred is daytime (step S1001: Yes), the determination apparatus 101 selects the first target municipality among all target municipalities (step S1002). Next, the determination apparatus 101 determines whether or not the disaster determination has been performed for all target municipalities (step S1003). When there is a target municipality that has not yet been subjected to the disaster determination (step S1003: No), the determination device 101 determines that at least one of the surrounding municipalities of the selected target municipality has occurred. It is determined whether or not (step S1004).

  When it is determined that one or more neighboring municipalities have caused the disaster (step S1004: Yes), the determination apparatus 101 determines the boundary ω for classifying the selected target municipality and the selected target city. It is determined from the number of tweets for each neighboring municipality of the municipality whether or not the selected target municipality has a disaster (step S1005). Specifically, when the determination apparatus 101 determines that the number of tweets for each of the surrounding municipalities of the selected municipality is data belonging to the first class by the boundary ω, the selected municipality is It is determined that there is a disaster. In addition, when the determination apparatus 101 determines that the number of tweets for each neighboring municipality of the selected municipality is data belonging to the second class, based on the boundary ω, the selected municipality is in a disaster. Judge that there is no.

  After the process of step S1005 is completed, or when it is determined that no surrounding municipalities have occurred (step S1004: No), the determination apparatus 101 determines the next target among all target municipalities. A municipality is selected (step S1006), and the process proceeds to step S1003.

  Further, when the disaster determination is performed for all the target municipalities (step S1003: Yes), the determination apparatus 101 ends the target municipality disaster determination processing. When the time when the disaster occurred is nighttime (step S1001: No), the determination apparatus 101 executes a disaster determination process at night (step S1007). The target city / town / town disaster determination processing is terminated. Here, the disaster determination processing at night is the first to fourth processing shown in FIG.

  Further, when it is determined in step S1005 that the target municipality has a disaster, the determination apparatus 101 may perform the disaster determination again for the surrounding municipalities of the target city. Here, when the disaster process is performed again, the determination apparatus 101 may relax the condition for determining that a disaster has occurred, such as lowering the threshold value for determining whether or not a disaster has occurred. For example, as a result of the increase in the number of tweets sent about disasters in the surrounding municipalities due to the disaster of the target municipalities, the surrounding municipalities have been devastated even though they did not actually occur It may be determined that. In this case, since the determination apparatus 101 changes the condition for determining that a disaster has occurred and performs the disaster determination again, the accuracy of determining whether or not the surrounding municipalities have occurred is improved. be able to.

  Further, in the process of step S903, the determination apparatus 101 uses the other cities except for the municipalities in which the difference in population between day and night exceeds the standard as the data belonging to the first and second classes. It may be the number of tweets sent about disasters in towns and villages. At this time, the determination apparatus 101 determines “the number of tweets for each of the surrounding municipalities of the selected target municipality” in the process of step S1005 as “the tweet for each of the other selected municipalities described above”. Change to "number" and execute.

  In step S1004, the determination apparatus 101 executes the process of step S1005 when it is determined that one of the surrounding municipalities has caused a disaster. Even when no disaster has occurred, the process of step S1005 may be executed.

  As described above, the determination apparatus 101 reflects the number of tweets in the surrounding municipalities of the target municipality in the determination process 102 if the target municipality is an area where the daytime population is larger than the standard than the nighttime population. Thereby, the determination apparatus 101 can eliminate the state where it is difficult to determine that a disaster has occurred in an area where the difference between the daytime population and the nighttime population exceeds the standard, and accurately determines that the corresponding area is the disaster area. be able to.

  Further, the determination apparatus 101 may determine whether or not to reflect the occurrence state of disaster information in the surrounding municipalities in the determination process 102 according to the time zone when the disaster occurred. Specifically, the determination apparatus 101 determines to reflect the occurrence state of the disaster information in the surrounding municipalities in the determination process 102 if the time zone in which the disaster occurred is daytime. On the other hand, if the time zone in which the disaster occurred is nighttime, it is determined not to reflect the occurrence status of the disaster information in the surrounding municipalities in the determination process 102. As a result, the determination apparatus 101 can accurately determine that an area where the daytime population is greater than the reference than the nighttime population is a disaster area in the daytime when erroneous determination is likely to occur.

  In addition, when the disaster has occurred in the past, the determination apparatus 101 determines the number of tweets for each surrounding city when the target city has occurred, and the target city does not have a disaster. The SVM may be executed based on the number of tweets for each surrounding municipality when the municipality has a disaster. Then, when a disaster occurs, the determination device 101 determines, as the determination process 102, that the target municipality has occurred based on the boundary obtained from the SVM and the number of tweets for each neighboring municipality when the disaster occurs. It may be determined whether it is ground. Thereby, the determination apparatus 101 can determine with high precision using SVM.

  In addition, the method of executing the SVM described above is not the same as the method of determining that the target municipality is also affected when the surrounding municipalities are affected by the disaster. The number of tweets for each past municipality when the disaster occurred and when it did not occur is used. Therefore, by executing the above-described SVM, the determination apparatus 101 can more accurately determine whether the target city is a disaster area as compared to the above-described method.

  In addition, in the method for executing the SVM described above, the administrator designates whether or not the number of tweets in the surrounding cities increases at the time of the disaster when generating the list of surrounding cities in the process of step S804. You don't have to. As described above, the determination apparatus 101 may mechanically add to the list of neighboring cities and towns without causing the administrator to specify which neighboring city or town will increase the number of tweets in the event of a disaster.

  Moreover, in the method for executing the SVM described above, the determination apparatus 101 may set the city where the disaster area is determined as a bed town. In this case, the data belonging to the first class is the number of tweets transmitted regarding past disasters in the municipalities around Bedtown when a past disaster occurred in Bedtown. The data belonging to the second class is disseminated regarding past disasters in the surrounding municipalities when no past disaster occurred in Bedtown but past disasters occurred in the surrounding municipalities in Bedtown. The number of tweets. Thereby, the determination apparatus 101 can determine whether a bed town is a disaster area.

  In addition, the determination apparatus 101 may execute the SVM by creating teacher data in other municipalities that exclude the area exceeding the standard of population difference between day and night from the neighboring municipalities of the target municipality. . In this case, also in the determination process 102, the determination apparatus 101 determines whether the target city is the disaster area from the boundary obtained from the SVM and the number of tweets for each other city when the disaster occurs. Determine if. Here, as shown in FIG. 5 and FIG. 6, the number of tweets does not change greatly in areas that exceed the standard of population difference between day and night, even if it occurs in the daytime or not. Therefore, the determination apparatus 101 can suppress the load on the SVM because the teacher data is reduced while maintaining the determination accuracy by excluding an area that exceeds the standard of population difference between day and night.

  The determination method described in the present embodiment can be realized by executing a program prepared in advance on a computer such as a personal computer or a workstation. This determination program is recorded on a computer-readable recording medium such as a hard disk, a flexible disk, a CD-ROM (Compact Disc-Read Only Memory), a DVD (Digital Versatile Disk), etc., and is read from the recording medium by the computer. Executed. The determination program may be distributed through a network such as the Internet.

  The following additional notes are disclosed with respect to the embodiment described above.

(Supplementary note 1) In a determination program for determining a disaster area according to the transmission status of disaster information on social media,
When the target area for determining whether or not it is a disaster area corresponds to an area where the difference in population between day and night exceeds the standard, the transmission status of disaster information in the surrounding area of the target area is indicated as the target area. Is reflected in the judgment process of whether or not it is a disaster area,
A determination program that causes a computer to execute processing.

(Additional remark 2) The determination program of Additional remark 1 which makes the said computer perform the process which determines whether it reflects in the said determination process according to the time slot | zone when the said disaster occurred.

(Supplementary note 3) The transmission status of the disaster information is indicated by the number of information transmitted regarding the disaster,
The number of pieces of information transmitted regarding the disaster in the surrounding area when a past disaster occurs in the target area from the disaster is data belonging to a first class, and the past disaster is in the target area. The support vector machine is executed with the number of information transmitted regarding the past disaster in the surrounding area when the past disaster has occurred in the surrounding area, although not occurring, as data belonging to the second class To identify a boundary for classifying the first class and the second class,
When the disaster occurs, as the determination process, based on the identified boundary and the number of information transmitted regarding the disaster in the surrounding area, whether or not the target area is a disaster area. judge,
The determination program according to appendix 1 or 2, which causes the computer to execute processing.

(Supplementary note 4)
The number of pieces of information transmitted regarding the disaster in the other regions other than the region where the population difference between day and night exceeds the standard in the surrounding regions when the past disaster occurred in the target region is the first number. The data that belongs to the class of the information, and the past disaster has not occurred in the target area, but the information transmitted about the past disaster in the other area when the past disaster occurred in the other area A boundary for classifying the first class and the second class by executing a support vector machine as data belonging to the second class
The determination process is as follows.
Whether or not the target area is a disaster area is determined based on the identified boundary and the number of information transmitted regarding the disaster in the other area as the determination process when the disaster occurs. judge,
The determination program according to supplementary note 3, characterized by:

(Supplementary note 5) The determination program according to any one of supplementary notes 1 to 4, wherein the transmission status of the disaster information is indicated by the number of sentences including a character string indicating the disaster.

(Supplementary note 6) In the determination method for determining the disaster area according to the transmission situation of disaster information in social media,
When the target area for determining whether or not it is a disaster area corresponds to an area where the difference in population between day and night exceeds the standard, the transmission status of disaster information in the surrounding area of the target area is indicated as the target area. Is reflected in the judgment process of whether or not it is a disaster area,
A determination method, wherein a computer executes a process.

(Supplementary note 7) In a determination device for determining a disaster area according to the transmission status of disaster information in social media,
When the target area for determining whether or not it is a disaster area corresponds to an area where the difference in population between day and night exceeds the standard, the transmission status of disaster information in the surrounding area of the target area is indicated as the target area. A control unit to be reflected in the determination process of whether or not the disaster area,
The determination apparatus characterized by having.

DESCRIPTION OF SYMBOLS 101 Determination apparatus 102 Determination process 400 Control part 401 Tweet collection part 402 Reflection part 403 Determination part 404 Determination part 405 Specification part 406 Output part

Claims (6)

In a judgment program that determines disaster areas according to the status of disaster information transmission on social media,
When the target area for determining whether or not it is a disaster area corresponds to an area where the difference in population between day and night exceeds the standard, the transmission status of disaster information in the surrounding area of the target area is indicated as the target area. Is reflected in the judgment process of whether or not it is a disaster area,
A determination program that causes a computer to execute processing.   The determination program according to claim 1, wherein the computer is caused to execute a process of determining whether to reflect in the determination process according to a time zone when the disaster occurs. The transmission status of the disaster information is indicated by the number of information transmitted regarding the disaster,
The number of pieces of information transmitted regarding the disaster in the surrounding area when a past disaster occurs in the target area from the disaster is data belonging to a first class, and the past disaster is in the target area. The support vector machine is executed with the number of information transmitted regarding the past disaster in the surrounding area when the past disaster has occurred in the surrounding area, although not occurring, as data belonging to the second class To identify a boundary for classifying the first class and the second class,
When the disaster occurs, as the determination process, based on the identified boundary and the number of information transmitted regarding the disaster in the surrounding area, whether or not the target area is a disaster area. judge,
The determination program according to claim 1, wherein the computer causes the computer to execute a process. The process to specify is
The number of pieces of information transmitted regarding the disaster in the other regions other than the region where the population difference between day and night exceeds the standard in the surrounding regions when the past disaster occurred in the target region is the first number. The data that belongs to the class of the information, and the past disaster has not occurred in the target area, but the information transmitted about the past disaster in the other area when the past disaster occurred in the other area A boundary that classifies the first class and the second class by executing a support vector machine as data belonging to the second class
The determination process is as follows.
Whether or not the target area is a disaster area is determined based on the identified boundary and the number of information transmitted regarding the disaster in the other area as the determination process when the disaster occurs. judge,
The determination program according to claim 3. In the judgment method to determine the disaster area according to the transmission status of disaster information on social media,
When the target area for determining whether or not it is a disaster area corresponds to an area where the difference in population between day and night exceeds the standard, the transmission status of disaster information in the surrounding area of the target area is indicated as the target area. Is reflected in the judgment process of whether or not it is a disaster area,
A determination method, wherein a computer executes a process. In a determination device that determines a disaster area according to the transmission status of disaster information in social media,
When the target area for determining whether or not it is a disaster area corresponds to an area where the difference in population between day and night exceeds the standard, the transmission status of disaster information in the surrounding area of the target area is indicated as the target area. A control unit to be reflected in the determination process of whether or not the disaster area,
The determination apparatus characterized by having.

Images