Recommendation system, recommendation method, and recommendation program

Abstract

To provide a recommendation system for determining, by using user information such as use frequency and the like, whether a recommendation algorithm is good or bad.SOLUTION: The present invention is directed to providing a recommendation system having an output unit for outputting a plurality of recommended movement destinations using a mobility for a user to present them on a display screen of a terminal held by the user, a result specifying unit for specifying a result showing that the user selected or did not select any one of the plurality of recommended movement destinations after presentation, a user information acquiring unit for acquiring user information included in a process from the presentation to specification, and an influence specifying unit for specifying influence of a way of presentation from the user information.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present disclosure relates to a recommendation system, a recommendation method, and a recommendation program. [Background technology]

[0002] A recommendation system has been developed that allows users to travel by obtaining preferred destination information and route information. The recommendation system of Patent Document 1 includes a user terminal and a server. The user terminal inputs information about the user's current situation. In the user terminal or server, a recommendation means having artificial intelligence narrows down destination information by taking into account transportation means information based on pre-input user preference information, input information about the user's current situation, and current location information of the user terminal. At the same time, the recommendation means narrows down route information including transportation means information. The recommendation means then displays the narrowed-down route information and destination information as a set on the display unit of the user terminal and makes a recommendation. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Publication No. 2023-12995 Summary of the Invention [Problem to be solved by the invention]

[0004] In the method of Patent Document 1, whether a recommended location matches the user's preferences can only be determined by the user's selection results. Therefore, results that the user did not select were all considered to be due to a bad algorithm. However, results that the user did not select do not necessarily indicate a bad recommendation algorithm, but may also be due to a bad UI (User Interface). Therefore, the purpose of the present disclosure is to provide a recommendation system that determines the quality of a recommendation algorithm using various user information such as frequency of use. [Means for solving the problem]

[0005] The recommendation system of the present disclosure includes: an output unit that outputs a plurality of recommended destinations using mobility to a user so as to present them on a display screen of a terminal carried by the user; a result determination unit that determines whether the user selects or does not select any of the plurality of recommended destinations after the presentation; a user information acquisition unit that acquires user information included in the process from the presentation to the identification; and an influence specifying unit that specifies the influence of the presentation method from the user information.

[0006] The above configuration provides a recommendation system that judges the quality of a recommendation algorithm using various user information such as frequency of use.

[0007] The recommendation system of the present disclosure includes: the plurality of recommended destinations are presented by an algorithm; The algorithm is characterized by being machine-learned from the identified effects of the presentation manner.

[0008] With the above configuration, the system can machine-learn algorithms based on user information and provide further improved recommendations.

[0009] The recommendation system of the present disclosure includes: a tagging unit that assigns tags to the recommended destinations; the output unit outputs tags together with the plurality of recommended destinations for presentation on the display screen; The impact identification unit is characterized in that when the user presses the presented tag and the presented tag includes the tag assigned to the recommended destination, it determines that there is a problem with the image of the recommended destination.

[0010] With the above configuration, it is possible to distinguish between problems with the algorithm and problems with the displayed image by using the recommended image and the tag attached to it.

[0011] The recommendation method of the present disclosure includes: outputting a plurality of recommended destinations using mobility to the user for presentation on a display screen of a terminal carried by the user; Identifying a result of whether the user selects or does not select any of the plurality of recommended destinations after the presentation; acquiring user information included in the process from the presentation to the identification; The recommendation method identifies the influence of the presentation method from the user information.

[0012] The above configuration provides a recommendation method that determines the quality of a recommendation algorithm using various user information such as frequency of use.

[0013] The recommendation program of the present disclosure is outputting a plurality of recommended destinations using mobility to the user for presentation on a display screen of a terminal carried by the user; Identifying a result of whether the user selects or does not select any of the plurality of recommended destinations after the presentation; acquiring user information included in the process from the presentation to the identification; The recommendation program causes an information processing device to identify the influence of the presentation method from the user information.

[0014] The above configuration provides a recommendation program that causes an information processing device to determine the quality of a recommendation algorithm using various user information such as frequency of use. [Effects of the Invention]

[0015] The present disclosure provides a recommendation system that determines whether a recommendation result is good or bad using various user information such as frequency of use. [Brief explanation of the drawings]

[0016] [Figure 1] 1 is a block diagram showing a configuration of a recommendation system according to an embodiment; [Figure 2] 1 is a flowchart of a recommendation method according to an embodiment. [Figure 3] 10 is a table illustrating an example of user information according to an embodiment; [Figure 4] 10 is a table showing a method for determining the quality of an algorithm and a recommended destination image using recommended images and tags attached thereto according to an embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0017] Embodiment Hereinafter, embodiments of the present invention will be described with reference to the drawings. However, the invention according to the claims is not limited to the following embodiments. Furthermore, not all of the configurations described in the embodiments are necessarily essential means for solving the problems. For clarity of explanation, the following description and drawings have been omitted and simplified as appropriate. In each drawing, the same elements are given the same reference numerals, and duplicate explanations are omitted as necessary.

[0018] (Description of Recommendation System According to Embodiment) FIG. 1 is a block diagram showing a configuration of a recommendation system according to an embodiment. FIG. 3 is a table showing an example of user information according to an embodiment. The recommendation system according to an embodiment will be described with reference to FIGS. 1 and 3. The recommendation system 100 is a system that learns recommendation results using various user information such as frequency of use.

[0019] As shown in FIG. 1, the recommendation system 100 includes an output unit 101, a result identification unit 102, a user information acquisition unit 103, and an influence identification unit 104. The recommendation system 100 is configured with an information processing device. The information processing device is configured with a processor, such as a CPU (Central Processing Unit), that executes a program, and a memory that stores the program. The information processing device may be configured with one device or multiple devices. The information processing device may be configured with a cloud server in which some or all of its functions are distributed.

[0020] The output unit 101 outputs a plurality of recommended destinations using mobility to the user for presentation on a display screen of a terminal carried by the user. A mobility is a vehicle for transporting people, such as an automobile or a motorcycle. A user is a person who travels by riding on the mobility. The terminal is, for example, an electronic device terminal such as a smartphone or a tablet. The terminal can connect to an external server via a mobile phone communication line such as 3G (3rd Generation), LTE (Long Term Evolution), 4G (4th Generation), or 5G (5th Generation), or via Wi-Fi (registered trademark). Application software (hereinafter referred to as an app) that recommends destinations and displays them on the display screen is installed on the terminal. The app is, for example, an outing destination suggestion app. The destinations are displayed as images. The user rides the mobility and selects a destination from the recommended destinations specified on the display screen of the terminal to travel to. The output unit 101 may be located on an external server or on the terminal. It is preferable that the recommended destinations are stored on an external server and downloaded to the terminal for presentation. If the output unit 101 is in an external server, the output unit 101 transmits the recommended destination to the terminal via communication. If the output unit 101 is in a terminal, the output unit 101 acquires the recommended destination from the external server and outputs it to be presented on the display screen. The recommended destination is output using an algorithm using AI (Artificial Intelligence).

[0021] The result identification unit 102 identifies the result in which the user selects or does not select one of the multiple recommended destinations after the presentation. Selecting a recommended destination means that the user indicates their intention to go there by, for example, tapping a destination from the multiple recommended destinations. The app then proceeds to a step of presenting map information or navigating to the destination. Not selecting a recommended destination means that the user does not select a destination from the multiple recommended destinations. The result identification unit 102 may be located in an external server or in the terminal.

[0022] The user information acquisition unit 103 acquires user information included in the process from presentation to identification. For example, as shown in Fig. 3, the user information included in the process from presentation to identification includes the frequency of use of functions within the app, the total number of presses within the app, the time spent within the app, eye movements while using the app, emotional changes while using the app, and body movements while using the app. The user information acquisition unit 103 may be located in an external server or in the terminal.

[0023] The influence identification unit 104 identifies the influence of the presentation method from the user information. The influence identification unit 104 may be located in an external server or in the terminal. If a recommended destination for movement is not selected, it is considered that the recommendation is bad, that is, the AI algorithm is bad. Therefore, it is possible to consider machine learning the AI algorithm, but this is not always the case. Such cases are listed below.

[0024] The impact of the presentation method is that, for example, if the frequency or number of times various functions within the app are used is low, the UI may not be familiar and it may be difficult to use. Also, if the total number of presses within the app is low, the UI may be difficult to use. If the time spent in the app is short, it is possible that the user left before becoming familiar with the UI.

[0025] If the user's eyes do not move as expected when using the app, if they do not look at the icons they are supposed to look at, or if they move their eyes too much, the UI may be difficult to understand and their focus may become unstable. If the user's level of frustration increases while using the app, as estimated by facial expressions, the app may be difficult to use. If the user's body freezes for a short time while using the app, they may not be able to concentrate and move around a lot, making the UI difficult to understand.

[0026] In this way, if the algorithm is not bad, no matter how much the algorithm is modified, the problem will not be solved. Therefore, we weight the algorithm factors based on the influence of the presentation method. Then, we calculate an evaluation index for the algorithm factors. In other words, we eliminate factors that are not algorithm factors. The algorithm is machine-learned based on the influence of the identified presentation method, and presents the next recommended destination.

[0027] The frequency of feature use within an app, the total number of presses within the app, and the time spent within the app can be used to quantify the need for UI improvements, so when retraining the algorithm, users who have a strong UI factor can be used to reduce the evaluation weight.

[0028] Eye movements, emotional changes, and body movements while using an app can be used to present an appropriate UI according to the driver's state, such as a navigation display while driving.

[0029] In this way, a recommendation system is provided that judges the quality of recommendation results using various user information such as frequency of use.

[0030] (Explanation of recommendation method according to embodiment) 2 is a flowchart of the recommendation method according to the embodiment, which will be described with reference to FIG.

[0031] As shown in FIG. 2, first, input information is obtained (step S201). Prior information from the user is obtained to recommend a recommended destination. Next, the AI makes a recommendation using an algorithm (step S202). The AI recommends a recommended destination using an algorithm. Next, the recommendation results are presented (step S203). Next, user actions are measured (step S204). The user's reaction to the recommendation results is measured. The measurement results of user actions and user reactions are the user information described above.

[0032] Next, the weights of the algorithm factors are calculated (step S205). Then, evaluation indices for the algorithm factors are calculated (step S206). In this way, factors other than the algorithm factors are eliminated. The AI machine-learns the algorithm and recommends destinations again.

[0033] In this way, a recommendation method is provided in which the quality of the recommendation result is judged using various user information such as frequency of use.

[0034] (Description of Modifications of Recommendation System According to Embodiment) 4 is a table showing a method for determining the quality of an algorithm and a recommended destination image using recommended images and tags attached thereto according to an embodiment. A modified example of the recommendation system according to the embodiment will be described with reference to FIG. 4.

[0035] The modification of the recommendation system further includes a tagging unit that tags the recommended destinations.

[0036] The output unit 101 outputs tags together with multiple recommended destinations for presentation on a display screen. As shown in Fig. 4, the influence identification unit causes the algorithm to learn that a good recommendation was made when a recommended spot was pressed. The influence identification unit also causes the algorithm to learn that a bad recommendation was made when neither a recommended spot nor a tag was pressed.

[0037] On the other hand, if a tag is pressed but not a recommended spot, it is unclear whether the recommended destination is bad or the image of the recommended spot is bad. Therefore, the influence identification unit determines whether the pressed tag is included in the tag of the recommended spot assigned by the tag assignment unit. If the tag is not included, it is determined that the recommended destination is bad. If the tag is included, it is determined that the image of the recommended spot is bad, and it is clear that there is no problem with the algorithm.

[0038] In this way, the recommended images and the tags attached to them can be used to distinguish between problems with the algorithm and problems with the displayed images.

[0039] Furthermore, part or all of the processing in the recommendation system 100 described above can be realized as a computer program. Such a program can be stored and provided to a computer using various types of non-transitory computer-readable media. Non-transitory computer-readable media include various types of tangible recording media. Examples of non-transitory computer-readable media include magnetic recording media (e.g., flexible disks, magnetic tapes, hard disk drives), magneto-optical recording media (e.g., magneto-optical disks), CD-ROMs (Read Only Memory), CD-Rs, CD-R / Ws, and semiconductor memories (e.g., mask ROMs, PROMs (Programmable ROMs), EPROMs (Erasable PROMs), flash ROMs, and RAMs (Random Access Memory)). The program may also be provided to a computer by various types of temporary computer-readable media. Examples of temporary computer-readable media include electrical signals, optical signals, and electromagnetic waves. The temporary computer-readable media can provide the program to a computer via a wired communication path such as an electric wire or optical fiber, or via a wireless communication path.

[0040] The above configuration provides a recommendation program that causes an information processing device to determine the quality of a recommendation algorithm using various user information such as frequency of use.

[0041] The present invention is not limited to the above-described embodiment, and can be modified as appropriate without departing from the spirit of the present invention. For example, the eye movement of an app, changes in emotions while using the app, and body movements while using the app can be used to instantly change the UI to a different one in a robot that can have an interactive conversation with a person. [Explanation of symbols]

[0042] 100 Recommendation system, 101 Output unit, 102 Result identification unit, 103 User information acquisition unit, 104 Influence identification unit

Claims

1. an output unit that outputs a plurality of recommended destinations using mobility to a user so as to present them on a display screen of a terminal carried by the user; a result determination unit that determines whether the user selects or does not select any of the plurality of recommended destinations after the presentation; a user information acquisition unit that acquires user information included in the process from the presentation to the identification; An influence identification unit that identifies the influence of the presentation method from the user information.

2. the plurality of recommended destinations are presented by an algorithm; The recommendation system of claim 1 , wherein the algorithm is machine-learned from the identified presentation influences.

3. a tagging unit that tags the recommended destination; the output unit outputs tags together with the plurality of recommended destinations for presentation on the display screen; The recommendation system of claim 1, wherein the influence identification unit determines that there is a problem with the image of the recommended destination when the user presses the presented tag and the presented tag includes the tag assigned to the recommended destination.

4. outputting a plurality of recommended destinations using mobility to the user for presentation on a display screen of a terminal carried by the user; Identifying a result of whether the user selects or does not select any of the plurality of recommended destinations after the presentation; acquiring user information included in the process from the presentation to the identification; A recommendation method for identifying the influence of the presentation manner from the user information.

5. outputting a plurality of recommended destinations using mobility to the user for presentation on a display screen of a terminal carried by the user; Identifying a result of whether the user selects or does not select any of the plurality of recommended destinations after the presentation; acquiring user information included in the process from the presentation to the identification; A recommendation program that causes an information processing device to identify the influence of the presentation method from the user information.

Images