JP6229354B2 - Demand forecasting device, demand forecasting method, and demand forecasting program - Google Patents

Description

  The present invention relates to a demand forecasting device, a demand forecasting method, and a demand forecasting program for forecasting demand for a predetermined service such as a taxi service.

  In certain services, such as taxi services, if the actual demand and the number of taxis to be supplied are significantly different from each other, the number of taxis that run in an empty state (decrease in operating rate) and the waiting time will be prolonged. (Deterioration of customer convenience) is a concern.

  In view of this, various methods have been proposed to grasp the demand for taxis and to efficiently allocate taxis.

  For example, Patent Document 1 has a description of dispatching vehicles based on population density information and weather information.

  Further, in Patent Document 2, for a certain area, the demand level according to the Fijii rule is obtained from the setting conditions according to the weekend, the beginning / end of the month, the day of the week, the day of the night time, and the date and time when the usage frequency increases. Describes a vehicle demand forecasting system that calculates the final forecast demand amount in consideration of the above.

JP2003-288687A (Page No. 3-4) JP 09-153098 A (Page No. 4)

  In general, information on weather conditions, transportation operation conditions, event holding conditions, and the like is provided from specialized information providers (regardless of profit / non-profit, public / private).

  However, the information finally provided from the above-mentioned organization is information that is delayed a little because it undergoes analysis and aggregation processes for individual measurement results (reports from monitors, sensor device measurement results, camera images, etc.). . Furthermore, the above information is not information that reflects local needs because of a wide spatial detection unit.

  For example, an event information provider may not be able to provide information about the actual event end time. Alternatively, the weather information providing organization may not be able to provide information on local weather irregularities.

  That is, in demand prediction based on information from the information providing organization, it is difficult to immediately and locally grasp an increase in demand for a predetermined service.

  The present invention has been made to solve the above-described problems, and provides a demand prediction apparatus, a demand prediction method, and a demand prediction program capable of immediately and locally grasping an increase in demand for a predetermined service. There is to do.

  The demand prediction apparatus of the present invention is an apparatus that predicts demand for a predetermined service based on dialogue information exchanged on a social network.

  The demand prediction method of the present invention is characterized in that a demand for a predetermined service is predicted based on dialogue information exchanged on a social network.

  The demand prediction program of the present invention is a program for causing a computer of a demand prediction apparatus to execute a process of predicting demand for a predetermined service based on dialogue information exchanged on a social network.

  According to the present invention, it is possible to immediately and locally grasp an increase in demand for a predetermined service.

It is a block diagram which shows the structural example of the service provision system which concerns on the 1st Embodiment of this invention. It is a block diagram which shows the structural example of the demand prediction apparatus which comprises the service provision system shown in FIG. It is a list which shows an example of the 1st judgment keyword. It is a list which shows an example of the 2nd judgment keyword. It is a block diagram of high demand information. It is a flowchart for demonstrating the operation example of the dialogue information selection part which comprises a demand prediction apparatus. It is a flowchart for demonstrating the operation example of the high demand information preparation part which comprises a demand prediction apparatus. It is a timing chart for demonstrating the operation | movement which transmits high demand information for every predetermined transmission period. It is a spatial image figure of the high demand information in the case of FIG. It is a block diagram which shows the structural example of the demand prediction apparatus which concerns on the 2nd Embodiment of this invention. It is a block diagram which shows the structural example of the demand prediction program memorize | stored in the memory of FIG.

[First Embodiment]
[Description of configuration]
FIG. 1 is a block diagram illustrating a configuration example of a service providing system 10 according to the first embodiment of the present invention. The service providing system 10 is a system that provides a predetermined service according to the demand prediction. In the present embodiment, a case where the service providing system 10 is a system for dispatching a taxi will be described as an example. The service providing system 10 includes a demand prediction device 20 and a vehicle allocation control device 30.

  The demand prediction device 20 is connected to an SNS (abbreviation of Social Network System, Social Network Service, Social Network Site, etc.) 40 on the Internet 50. The SNS 40 is also called a social network, and is a membership website that forms a community on the Internet 50 and allows users to communicate with each other in various forms. Examples of the SNS include Twitter (registered trademark), Facebook (registered trademark), mixi (registered trademark), and the like.

  The demand prediction device 20 analyzes the “dialog information A” exchanged on the SNS 40 and finally generates high demand information D. The dialogue information A indicates one message having a predetermined data structure (for example, ID + text + time stamp). ID is an abbreviation for Identification. For example, in the case of Twitter information, the text is 140 characters or less. The time stamp indicates the posting date of the dialog information A.

  The vehicle allocation control device 30 controls taxi allocation based on the high demand information D.

  FIG. 2 is a block diagram illustrating a configuration example of the demand prediction apparatus 20. The demand prediction device 20 includes a dialogue information selection unit 22 and a high demand information creation unit 24.

  The dialogue information selection unit 22 is dialogue information that suggests an increase in demand for a predetermined service (in the present embodiment, taxi service) from among a plurality of dialogue information A exchanged on the SNS 40. Demand dialogue information B ”is selected. Specifically, the dialogue information sorting unit 22 sorts dialogue information having the first determination keyword (see FIG. 3) in the text as high demand dialogue information B. In this case, the data structure of the high demand dialogue information B is not different from the dialogue information A, and includes, for example, an ID, a text, and a time stamp.

  FIG. 3 is merely an example, and the first determination keyword is any keyword as long as it is at least one of a keyword indicating that demand for services has already increased and a keyword indicating that demand will increase in the future. There may be.

  The number of first determination keywords may be singular or plural. When a plurality of first determination keywords are set, an OR determination (selecting dialogue information A having any one of the first determination keywords as high demand interaction information B) may be performed, and an AND determination (having a plurality of first determination keywords simultaneously) Dialog information A may be selected as high demand dialog information B).

  It is assumed that the first determination keyword is registered in advance in the dialog information selection unit 22 or a predetermined storage unit (not shown in FIG. 2) of the demand prediction device 20 prior to the above processing.

  The high demand dialogue information B is transmitted from the dialogue information selection unit 22 to the high demand information creation unit 24 every time it is detected.

  The high demand information creation unit 24 further analyzes the received high demand dialogue information B and finally creates “high demand information D”. First, the high demand information creation unit 24 confirms the time stamp of the received high demand dialogue information B, and determines whether or not the high demand dialogue information B has been posted “most recently”. Here, the latest means a time range in which immediacy is not lost (for example, within 15 minutes).

  Next, the high demand information creation unit 24 extracts information on “location” from the high demand dialogue information B posted most recently. Specifically, the high demand information creation unit 24 determines whether or not a second determination keyword (see FIG. 4) that is a keyword for specifying a location exists in the text of the high demand dialogue information B that has been posted most recently. Determine. When the second determination keyword exists, the high demand information creation unit 24 extracts a noun including the second determination keyword (that is, a noun that identifies a place, usually a proper noun). For example, when the second determination keyword is “station”, the extracted noun is, for example, “Shinjuku station”. Hereinafter, the extracted noun is referred to as “high demand point information C”.

  That is, in this embodiment, the “location” extracted from the high demand dialogue information B selected as the dialogue information that suggests an increase in demand for a predetermined service (for example, taxi service) has already increased in demand. Or a place with high probability that demand will increase in the future.

  Note that FIG. 4 is merely an example, and the second determination keyword may be any keyword as long as the keyword can specify the place. For example, the second determination keyword can be a place name, a station name, a facility name, a store name, or an intersection name.

  The number of second determination keywords may be singular or plural. When a plurality of second determination keywords are set and high demand point information C for each second determination keyword is extracted (for example, “Shinjuku Station” and “Kasumigaseki Building”), the respective high demands thereof. The point information C may be included in the high demand information D as it is, or one of them may be selected and included in the high demand information D based on a predetermined standard. It is assumed that the second determination keyword is registered in advance in the high demand information creation unit 24 or a predetermined storage unit (not shown in FIG. 2) of the demand prediction device 20 prior to the above processing.

  The high demand information creation unit 24 creates high demand information D (see FIG. 5) including at least one high demand point information C. The created high demand information D is transmitted to the vehicle allocation control device 30.

The vehicle allocation control device 30 controls the taxi allocation based on the received high demand information D. Specifically, for example, the vehicle allocation control device 30 allocates a taxi to a point indicated by the high demand point information C included in the high demand information D. Alternatively, the vehicle allocation control device 30 collects more taxis at the above points than at other points. In addition, since dispatch control itself based on a demand prediction result is a well-known technique, more detailed description is abbreviate | omitted.
[Description of operation]
FIG. 6 is a flowchart for explaining an operation example of the dialog information selection unit 22.

  The dialogue information selection unit 22 performs a match determination with the first determination keyword on the plurality of dialogue information A exchanged on the SNS 40 (step S1). If they match (Yes determination in step S1), the dialogue information sorting unit 22 sorts the dialogue information A as the high demand dialogue information B (step S2). The dialogue information selection unit 22 transmits the high demand dialogue information B to the high demand information creation unit 24 (step S3). If they do not match (No determination in step S1), and if the process in step S3 ends, the above-described match determination process (process in step S1) for the next interaction information A is executed.

  FIG. 7 is a flowchart for explaining an operation example of the high demand information creation unit 24.

  The high demand information creation unit 24 receives the high demand dialogue information B from the dialogue information selection unit 22 (step S10). The high demand information creation unit 24 confirms the time stamp of the received high demand dialogue information B, and determines whether or not the high demand dialogue information B has been most recently posted (step S11).

  When the high demand dialogue information B has been posted most recently (Yes in step S11), the high demand information creation unit 24 extracts the high demand point information C from the text of the high demand dialogue information B (step S11). S12). As described above, the method for extracting the high demand point information C is performed using the second determination keyword. The high demand information creation unit 24 creates high demand information D including the extracted high demand point information C (step S13). The created high demand information D is transmitted to the vehicle allocation control device 30 (step S14). When the process of step S14 is completed, another high-demand dialog information B reception process (process of step S10) and subsequent steps are executed.

When the high demand dialogue information B is not the latest posted (No determination in step S11), the high demand dialogue information B is discarded, and the processing for receiving another high demand dialogue information B (the processing in step S10) is performed. Executed.
[Description of effects]
In the first embodiment described above, the demand prediction device 20 predicts demand for a predetermined service based on dialogue information exchanged on the SNS 40 (for example, Twitter information).

  The SNS 40 can easily perform communication (recent status report, information transmission, etc.) with the other party as compared with mail and telephone. For this reason, the actual information (such as “actual event end time” and “occurrence of downpour”) and the user's feelings (congestion, difficulty in returning home, etc.) are immediately reflected in the dialogue information on the SNS 40. Furthermore, as described above, more local information (for example, “Accident occurs at Shinjuku Station”, “A guerrilla heavy rain occurs near Kasumigaseki Building”, etc.) is exchanged on the SNS 40.

That is, the demand prediction device 20 of the present embodiment can immediately and locally grasp an increase in demand for a predetermined service. When the increase in demand for services can be grasped immediately and locally, the operating rate (merits for service providers) and convenience (merits for customers) can be further improved.
[Explanation about modified examples]
In the above description, the case where the high demand information D is transmitted to the vehicle allocation control device 30 every time the high demand point information C is extracted is taken as an example. However, the transmission timing of the high demand information D is not limited to the above. For example, the high demand information D may be transmitted every predetermined transmission period T. In this case, the high demand information D includes all the high demand point information C extracted in the transmission cycle T.

  FIG. 8 is a timing chart for explaining the operation when the high demand information D is transmitted every transmission cycle T. FIG. 8 shows a case where one high demand point information C1 is extracted in the first period T1, and three high demand point information C2, C3, C4 are extracted in the second period T2. FIG. 9 is a spatial image diagram of the high demand information D in the case of FIG.

  In FIG. 8, in the case of the first period T1, the high demand information D1 includes only the high demand point information C1. The spatial image of the high demand information D1 in this case is as shown in FIG.

  In FIG. 8, in the second period T2, the high demand information D2 includes high demand point information C2, C3, and C4. The spatial image of the high demand information D2 in this case is as shown in FIG.

  In the above, the transmission cycle T may be a cycle (for example, 15 minutes) that does not lose immediacy. Further, when no high demand point information C is extracted within the transmission cycle T, it is not necessary to transmit the high demand information D.

  When demand occurs at the same time and exceeds the supply capacity (the number of taxis), it may be difficult to determine which demand a taxi should be assigned with priority. However, by transmitting the high demand information D at the timing as described above, the vehicle allocation control device 30 can easily grasp the point where the demand is concentrated, for example, as shown in FIG. it can. Therefore, more efficient taxi dispatch is possible even in the above situation. In addition, in this case, since the transmission cycle T is set in the time range as described above, immediacy is not lost (that is, a business opportunity is not missed).

  Moreover, in the above description, the case where the high demand information creation unit 24 automatically creates the high demand information D when the high demand point information C is extracted is taken as an example. However, the present invention is not limited to this, and for example, the high demand information creation unit 24 may generate the high demand information D only when the extracted high demand point information C is within a predetermined area (for example, within the business area of a taxi company). You may make it create. By doing in this way, the burden on the vehicle allocation control device 30 side can be reduced.

  Further, the high demand information creation unit 24 can also build up a demand information database by accumulating all the high demand dialogue information B in a predetermined storage device (not shown in FIG. 2). Future demand can be estimated by analyzing the demand information database.

  In the above description, a case where the demand prediction of a taxi is performed is taken as an example. However, the demand prediction of the demand prediction device 20 is not limited to the above, but other services (temporary bus, home delivery service, etc.) It can also be applied to demand forecasts. In that case, what is necessary is just to set a 1st determination keyword as needed.

  Moreover, in the above description, the case where the output destination (use destination) of the demand prediction result (high demand information D) of the demand prediction device 20 is the vehicle allocation control device 30 that is a taxi company's device is taken as an example. However, the output destination of the above result is not limited to the above, and may be, for example, a store (sales shop, restaurant, playground, etc.) near the place indicated by the high demand point information C according to the service to be provided. You can also.

In addition, when performing the demand forecast described above (that is, demand forecast based on dialogue information exchanged on the SNS 40), a predetermined information provider (event information provider, weather information provider, traffic information provider, etc.) ) Can also be referred to. Although the dialogue information exchanged on the SNS 40 is excellent in immediacy and locality, there is a concern that a typographical error or misunderstanding of the user may occur. On the other hand, the information provided from the information provider has a certain level of reliability, although the immediacy and locality are inferior. Therefore, by combining the both, it is possible to perform more accurate demand prediction while maintaining immediacy and locality. For example, whether or not the extracted high demand point information C is valid can be finally determined based on the information. However, in this case, the above information may be delayed, so care must be taken so that immediacy is not adversely affected.
[Second Embodiment]
FIG. 10 is a block diagram illustrating a configuration example of the demand prediction apparatus 100 according to the second embodiment of the present invention. The demand prediction apparatus 100 includes a CPU (Central Processing Unit) 102 and a memory 104.

  The memory 104 stores the demand prediction program 200. The demand prediction program 200 is executed by the CPU 102. Examples of the memory 104 include non-temporary storage means such as a ROM (Read Only Memory), a hard disk, a removable medium, or a removable disk.

  FIG. 11 is a block diagram illustrating a configuration example of the demand prediction program 200. The demand prediction program 200 includes a first program 202 and a second program 204.

  The first program 202 is a program for selecting high demand dialogue information B, which is dialogue information suggesting an increase in demand for a predetermined service, from a plurality of pieces of dialogue information A. The first program 202 corresponds to the processing of steps S1 to S3 in FIG.

  The second program 204 is a program for extracting high demand point information C, which is information about a place, from the high demand dialogue information B and creating high demand information D including the extracted high demand point information C. The second program 204 corresponds to the processing in steps S10 to S14 in FIG.

  According to the second embodiment described above, as in the first embodiment, an increase in demand for a predetermined service can be grasped immediately and locally.

DESCRIPTION OF SYMBOLS 10 Service provision system 20 Demand prediction apparatus 22 Dialog information selection part 24 High demand information preparation part 30 Vehicle allocation control apparatus 40 SNS
50 Internet 100 Demand forecasting device 102 CPU
104 memory 200 demand prediction program 202 first program 204 second program A dialogue information B high demand dialogue information C, C1 to C4 high demand point information D, D1, D2 high demand information T1 first period T2 second period

Claims (6)

Based on dialogue information exchanged on social networks, the demand for a given service is predicted ,
A dialog information selection unit that selects high demand dialog information, which is dialog information suggesting an increase in demand, from a plurality of the dialog information;
High demand point information that is information about a place is extracted from the high demand dialogue information, high demand information including the extracted high demand point information is created, and the created high demand information is transmitted to a control device. A demand information creation department;
With
The high demand information creation unit includes the high demand point information extracted within a predetermined transmission cycle in the high demand information, and transmits the high demand information creation unit to the control device at each transmission cycle. A demand forecasting device characterized by that. The interactive information selecting unit, from a plurality of said interactive information, interactive information including the first determining keyword is a keyword showing the increase of the demand, according to claim 1, characterized by selected as the high demand interactive information Demand forecasting equipment. The demand prediction device according to claim 2, wherein the first determination keyword is at least one of a keyword indicating that the demand has already increased and a keyword indicating that the demand will increase in the future. The high demand information creation unit determines whether or not a second determination keyword that is a keyword for specifying the location exists in the high demand dialogue information, and when the second determination keyword exists, 2 extracts the noun comprising determining keywords claims extracted the nouns, characterized in that said high demand point information 1 - 3 demand prediction apparatus according to any one of. A demand prediction method in a demand prediction device,
The demand prediction device
Based on dialogue information exchanged on social networks, the demand for a given service is predicted ,
Selecting high demand dialogue information, which is dialogue information suggesting an increase in demand, from a plurality of the dialogue information;
From the high demand dialogue information, extract high demand point information that is information about the location, create high demand information including the extracted high demand point information, and send the created high demand information to the control device,
The high demand information includes the high demand point information extracted within a predetermined transmission cycle, and transmits the high demand information creation unit to the control device every transmission cycle. . A process for predicting demand for a predetermined service based on dialogue information exchanged on a social network with a computer of a demand prediction device ;
A process of selecting high demand dialogue information that is dialogue information suggesting an increase in demand from a plurality of the dialogue information;
Processing for extracting high demand point information which is information about a place from the high demand dialogue information, creating high demand information including the extracted high demand point information, and transmitting the created high demand information to a control device When,
Demand for causing the high demand information to include the high demand point information extracted within a predetermined transmission cycle, and to execute processing to transmit from the high demand information creation unit to the control device for each transmission cycle Prediction program.

Images