Automobile recommendation system

Abstract

An automobile recommendation system 1 for recommending an automobile to a user includes a memory for storing instructions and one or more processors. When the instructions are executed, the one or more processors cause the automobile recommendation system to: determine a first group of users belonging to attributes that are the same or similar to attributes of a particular user at a predetermined time; according to the difference between the vehicle type characteristic value required by the first group of users and the vehicle type characteristic value required by the specific user at the preset moment, calculating an index representing the value of the user; determining a second group of users belonging to attributes identical or similar to the current attribute of the specific user; calculating a vehicle type characteristic value after combined calculation according to the vehicle type characteristic value required by the second group of users and the index representing the value of the specific user; determining candidate vehicle types according to the vehicle type characteristic values after merging calculation and the characteristic values of all the vehicle types; and recommending the candidate vehicle model to the specific user.

Description

Technical Field

[0001] This invention relates to a technology for recommending goods to users, and more particularly to an effective technology for recommending cars in a car recommendation system. Background Technology

[0002] The industry is researching a mechanism that can recommend suitable products based on users' needs and values ​​during the sales process. However, unless users are directly questioned through methods such as questionnaires, their needs and values ​​cannot be directly observed.

[0003] For example, Japanese Patent Application Publication No. 2022-90717 (Patent Document 1) describes the following: Based on the user's attribute information, the content (demand) of the characteristics that the user seeks for the predetermined characteristic items of a product (such as a car) is inferred, and the inferred demand is compared with the actual characteristic content of the product, thereby selectively recommending products that meet the user's needs.

[0004] [Existing Technical Documents]

[0005] [Patent Literature]

[0006] Patent Document 1: Japanese Patent Application Publication No. 2022-90717 Summary of the Invention

[0007] [The problem the invention aims to solve]

[0008] Based on existing technology, it is possible to infer a user's needs based on their attributes and recommend products that meet those needs.

[0009] In this scenario, to infer user needs (preferences) based on user attributes, for example, user groups with the same or similar attributes would be segmented, and the needs of these users would be averaged. However, even users with the same or similar attributes are not necessarily recommended the same products. Even users with the same attributes may have different values ​​(preferences), so it is necessary to recommend products that align with their values. However, these values ​​are usually not directly observable unless the user is directly asked.

[0010] Therefore, the purpose of this invention is to provide a car recommendation system that can predict the differences and changes in demand caused by differences in user attributes, as well as the values ​​of each user, and thus recommend products (especially cars).

[0011] The present invention and other objects and novel features will become clear from the description and drawings herein.

[0012] [Solutions for solving the problem]

[0013] The invention disclosed in this application is briefly described below as a representative summary.

[0014] A typical embodiment of the car recommendation system of the present invention includes memory for storing instructions and one or more processors. When the instructions are executed, the one or more processors cause the car recommendation system to perform the following operations: determine a first group of users belonging to attributes that are the same as or similar to those of a specific user at a predetermined time; calculate an index representing the user's values ​​based on the difference between the vehicle characteristic values ​​required by the first group of users and the vehicle characteristic values ​​required by the specific user at the predetermined time; determine a second group of users belonging to attributes that are the same as or similar to those of the specific user at present; calculate a combined vehicle characteristic value based on the vehicle characteristic values ​​required by the second group of users and the index representing the specific user's values; determine candidate vehicle models based on the combined vehicle characteristic value and the characteristic values ​​of each vehicle model; and recommend the candidate vehicle models to the specific user.

[0015] [Invention Effects]

[0016] The effects that can be obtained through representative content in the invention disclosed in this application are briefly described below.

[0017] In other words, according to a typical embodiment of the present invention, it is possible to recommend cars by inferring the differences and changes in needs caused by differences in user attributes, as well as the values ​​of each user. Attached Figure Description

[0018] Figure 1 This is a schematic diagram illustrating a structural example of a car recommendation system in one embodiment of the present invention.

[0019] Figure 2 (a) Figure 2 (b) is a schematic diagram illustrating an example of a car recommendation approach based on inferring user needs and values ​​in one embodiment of the present invention.

[0020] Figure 3 This is a schematic diagram illustrating an example of a data structure for performance parameters categorized by vehicle type in one embodiment of the present invention.

[0021] Figure 4 This is a schematic diagram illustrating an example of a data structure for classifying vehicle type characteristic values ​​in one embodiment of the present invention.

[0022] Figure 5 This is a schematic diagram illustrating an example of a user-owned vehicle model data structure in one embodiment of the present invention.

[0023] Figure 6This is a schematic diagram illustrating an example of the data structure for user attributes at the start of holding, according to one embodiment of the present invention.

[0024] Figure 7 This is a schematic diagram illustrating an example of a data structure for classifying vehicle characteristics by user in one embodiment of the present invention.

[0025] Figure 8 This is a schematic diagram illustrating an example of a data structure for classifying vehicle characteristics by user attributes in one embodiment of the present invention.

[0026] Figure 9 This is a schematic diagram illustrating an example of a data structure for classifying user values ​​in one embodiment of the present invention.

[0027] Figure 10 This is a schematic diagram illustrating an example of a data structure for current user characteristics in one embodiment of the present invention.

[0028] Figure 11 This is a schematic diagram illustrating an example of a data structure for the vehicle model characteristics required by the current user in one embodiment of the present invention.

[0029] Figure 12 This is a schematic diagram illustrating an example of the data structure for recommending vehicle models in one embodiment of the present invention.

[0030] Figure 13 This is an example diagram of the hardware structure in one embodiment of the present invention.

[0031] Figure 14 This is a flowchart illustrating a series of operations performed by the car recommendation system in one embodiment of the present invention. Detailed Implementation

[0032] Hereinafter, embodiments of the present invention will be described in detail based on the accompanying drawings. Furthermore, in all the drawings describing the embodiments, the same reference numerals are generally used for the same parts, and repeated descriptions are omitted. On the other hand, there are cases where parts described with reference numerals in a certain drawing may be mentioned in other drawings but not illustrated again, although they are referred to with the same reference numerals.

[0033] <Summary>

[0034] As mentioned above, the needs and values ​​of users, resulting from differences in user attributes, are usually not directly observable. This requires a clear definition of "needs" and individual "values," and even if they can be defined, they are still difficult to observe directly unless asked directly by the users.

[0035] In response, one embodiment of the car recommendation system of the present invention, when recommending vehicles to users, assumes that the user's needs and values ​​are reflected in the performance indicators of the models the user currently owns. In other words, considering that users typically choose cars that align with their values, the performance indicators of their currently owned models can be used to explore the user's needs and values.

[0036] Furthermore, it is believed that the demand arising from differences in user attributes can be obtained by finding representative values ​​(such as average, median, etc.) of the car performance indicators required by user groups with the same or similar attributes (hereinafter collectively referred to as "same attributes"). Each user's values ​​can be assessed by the difference between their desired performance indicators and the aforementioned representative values ​​(e.g., "This user is more concerned about fuel consumption than the average user in their 30s"). Moreover, these user values ​​are considered innate and unlikely to change significantly even as life stages change.

[0037] Figure 2 This is a schematic diagram illustrating an example of a method for recommending cars by inferring user needs and values, as described in one embodiment of the present invention. Figure 2 (a) illustrates an example of inferring user values. The horizontal axis in the figure represents the characteristics of the car (in...). Figure 2 In the example, there are four feature items ("Feature 1" to "Feature 4"). The vertical axis represents the feature intensity categorized by feature item. For each feature item, the hollow portion represents the vehicle model feature (representative value) required by the user group with the same attributes as the target user at the time of initial ownership (when ownership began). Additionally, the shaded portion shows the vehicle model feature required by the user at the time of ownership.

[0038] In this embodiment, the difference between the hollow and shaded portions categorized by characteristic items in the chart—that is, the difference between the representative value (preference) of a user group with the same attributes and the characteristic value required by the target user—can be considered to reflect the user's values. Figure 2 In example (a), there are differences between "Characteristic 1" and "Characteristic 3" (indicated by the solid black arrows in the diagram). These differences represent the user's values. Furthermore, the initial ownership date of the user's current car is used as the benchmark because at that time, through sales activities and various procedures, there is an opportunity to obtain information about the user's attributes and other aspects. However, it is not mandatory to use the initial ownership date as the benchmark; other dates can be chosen.

[0039] Figure 2 (b) shows the basis Figure 2(a) Using the obtained user value data, examples of vehicle model characteristics to recommend to the current user are derived. The hollow portion shows the vehicle model characteristics (representative values) required by the user group with the same attributes as the current user. The characteristic strength of each characteristic item is... Figure 2 The hollow portion in (a) is different, reflecting the difference in intensity due to differences and changes in attributes such as age (e.g., from the 20s to the 30s, needs may also differ due to changes in life stages).

[0040] In this embodiment, the characteristics of the vehicle model required by the user group with the same attributes are reflected relative to the hollow portion. Figure 2 The user's values ​​(solid black arrows) obtained in (a) can be used to determine the characteristics of recommended car models that reflect those values ​​(shaded areas). Then, the car model that actually possesses characteristics similar to those of the recommended car models obtained in this way is the target recommended car model.

[0041] In this way, it's possible to differentiate user preferences based on attributes and individual values, enabling car recommendations that adapt to changing life stages. For example, recommendations could be made as follows: "Due to age changes, the next car should be a model generally favored by the user. However, because the user values ​​environmental performance, a car with lower fuel consumption is recommended." Or, "A user who previously drove a car in Tokyo is now recommended a car with a different model after moving to Hokkaido." "A user who previously drove a car in Tokyo is now recommended a car with a different model after marriage." "A user who enjoys camping is recommended a car with a greater emphasis on space compared to other users with similar attributes."

[0042] <System Structure and Processing Flow>

[0043] Figure 1 This is a schematic diagram illustrating a structural example of a car recommendation system according to one embodiment of the present invention. The car recommendation system 1 is an information processing system, which may consist of, for example, server equipment, a virtual server built on cloud computing services, or an information processing terminal. It executes middleware such as an OS (operating system), DBMS (database management system), and Web server programs, as well as software running on them, from storage devices such as HDDs (hard disk drives) and SSDs (solid-state drives) into memory, via a CPU (central processing unit) not shown, thereby realizing various functions related to car recommendation.

[0044] The car recommendation system 1 includes the following components: for example, a vehicle model characteristic calculation unit 11 installed as software, a user-required vehicle model characteristic calculation unit 12, a user-attribute-classified vehicle model characteristic calculation unit 13, a user-classified-value calculation unit 14, a current user-required vehicle model characteristic calculation unit 15, and a recommended vehicle model calculation unit 16. Additionally, it includes the following data repositories consisting of databases and files: vehicle model classification performance parameters 21, user-held vehicle models 22, user attributes at the time of initial ownership 23, current user attributes 24, vehicle model classification characteristic values ​​31, user-classified required vehicle model characteristics 32, user-attribute-classified required vehicle model characteristics 33, user-attribute-classified required vehicle model characteristics 34, current user-required vehicle model characteristics 35, and recommended vehicle models 36.

[0045] In the process of recommending cars to users, the characteristic values ​​of each model from each manufacturer are first calculated in advance by the model characteristic calculation unit 11. Specifically, the performance parameters of each model stored in the model classification performance parameter 21 are used as input, and the characteristic values ​​of each model are calculated based on a predetermined model obtained through machine learning, and output as the model classification characteristic value 31. Furthermore, there are no particular restrictions on the tools or libraries used for machine learning; suitable tools or libraries can be selected and used.

[0046] Figure 3 This is a schematic diagram illustrating an example of a data structure for performance parameters 21 categorized by vehicle model, according to one embodiment of the present invention. Performance parameters 21 categorized by vehicle model is a table storing performance indicators / parameters for various vehicle models from different manufacturers. Examples include key items such as manufacturer, body shape, model name, year and model number (items marked with ※ in the diagram, the same below), new car price, horsepower, fuel consumption, vehicle length / width / height, and other performance parameter-related items. Furthermore, the performance parameter items are not limited to the above and may include other items. This data, such as from automobile manufacturers' websites, product catalogs, and market data, can be collected automatically or manually through various media.

[0047] Figure 4 This is a schematic diagram illustrating an example of a data structure for vehicle model classification characteristic value 31 in one embodiment of the present invention. The vehicle model classification characteristic value 31 is a table used to store characteristic values ​​calculated for each vehicle model in the vehicle model characteristic calculation unit 11 based on the vehicle model classification performance parameter 21. It includes key items such as manufacturer, body shape, vehicle model name, and year / model, as well as characteristic value-related items such as characteristic 1, characteristic 2, characteristic 3, ... Each item (characteristic 1, characteristic 2, characteristic 3, ...) is calculated based on a predetermined model obtained from machine learning in this embodiment; therefore, the meaning of each item and its value is a "black box" and is essentially uninterpretable by humans. The number of items is not particularly limited and can be determined based on the model design.

[0048] return Figure 1 Next, taking the user's owned vehicle model 22 and the user's current owned vehicle model and user attribute information stored in the user attributes 23 at the time of ownership as input, the user's required vehicle model characteristic calculation unit 12 calculates the characteristics of the vehicle model owned by each user, i.e. what kind of vehicle model characteristics each user needs, by referring to the vehicle model classification characteristic value 31 as described above, and outputs it as the user classification required vehicle model characteristic 32.

[0049] Figure 5 This is a schematic diagram illustrating an example of the data structure for a user-held vehicle model 22 in one embodiment of the present invention. The user-held vehicle model 22 is a table storing information about the vehicle models held by each user, including, for example, items used to identify the vehicle model such as user ID (key item), manufacturer of the held vehicle model, body shape of the held vehicle model, name of the held vehicle model, and year and model number of the held vehicle model.

[0050] Figure 6 This is a schematic diagram illustrating an example of the data structure for the user attribute 23 at the time of initial ownership, as described in one embodiment of the present invention. The user attribute 23 at the time of initial ownership is a table storing the attribute information of each user at the time they begin owning the vehicle model. For example, it includes user ID (a key item), gender, region (prefecture) at the time of initial ownership, and age category at the time of initial ownership, among other user attribute-related items. However, the items related to user attributes are not limited to the above; for example, they may also include other attribute items such as occupation, family composition, and interests.

[0051] and Figure 7 This is a schematic diagram illustrating an example of a data structure for user-classified vehicle model characteristics 32 in one embodiment of the present invention. User-classified vehicle model characteristics 32 is a table storing the attributes of each user and the characteristic information of the vehicle models they own, i.e., the characteristic requirements of each user for the currently owned vehicle model. For example, it includes user ID (key item), gender, region (prefecture) at the time of initial ownership, age classification at the time of initial ownership, and other items related to user attributes; vehicle model manufacturer, vehicle model body shape, vehicle model name, vehicle model year and model, and other items used to determine the vehicle model; and characteristic value-related items such as characteristic 1, characteristic 2, characteristic 3, etc. The above items are identical to the corresponding items in the user-owned vehicle model 22, user attributes at the time of initial ownership 23, and vehicle model classification characteristic values ​​31, which are respectively used as input.

[0052] return Figure 1Next, the vehicle characteristics 32 required for user classification are taken as input, and the calculation unit 13 for vehicle characteristics required for user attribute classification summarizes and statistically processes them according to the user's attributes, and outputs them as vehicle characteristics 33 required for user attribute classification.

[0053] Figure 8 This is a schematic diagram illustrating an example of a data structure for classifying vehicle characteristics 33 by user attributes in one embodiment of the present invention. Classifying vehicle characteristics 33 by user attributes involves summarizing and statistically processing the characteristics of each user's owned vehicle according to the user's attributes at the time of initial ownership. Specifically, it stores tables related to demand (preference) information resulting from differences in user attributes. For example, it includes key items such as gender, region (prefecture) at the time of initial ownership, and age classification at the time of initial ownership, as well as characteristic values ​​such as characteristic 1, characteristic 2, characteristic 3, etc.

[0054] In this embodiment, the user attributes used as the aggregation units are gender and region (prefecture), age classification (e.g., "20s", "30s", "40s", etc.), but are not limited to these. A user's residential area does not have to be a prefecture, but a smaller region, or conversely, it can be a larger region such as Kanto, Tohoku, or Chubu. Other attribute items such as occupation and family composition can also be used. The values ​​of each item (Attribute 1, Attribute 2, Attribute 3, ...) can be, for example, the average value after aggregation by user attribute unit, or other statistical indicators such as the median.

[0055] return Figure 1 Next, the vehicle characteristics 32 required for user classification and the vehicle characteristics 33 required for user attribute classification are taken as inputs, and the value of each user (more accurately, "indicators representing values", but simply "values" here) is calculated by the user classification value calculation unit 14 and output as the user classification value 34.

[0056] Figure 9 This is a schematic diagram illustrating an example of a data structure categorized by user values ​​34 in one embodiment of the present invention. User values ​​34 is a table storing the difference between the representative values ​​of each user's values ​​(i.e., the vehicle characteristics sought by a user group with the same attributes as that user) and the vehicle characteristics required by that user. For example, it includes user ID (a key item) and various items related to characteristic values ​​such as characteristic 1, characteristic 2, characteristic 3, ... The values ​​of characteristic 1, characteristic 2, characteristic 3, ... are as follows: Figure 2As shown in (a), the difference between the values ​​of each item of the vehicle characteristics 33 required by user attributes (based on the user's attribute needs (preferences) at the time of initial ownership) and the values ​​of each item of the vehicle characteristics 32 required by user attributes (the vehicle characteristics required by the user at the time of initial ownership).

[0057] return Figure 1 Next, the data of user classification values ​​34 and current user attributes 24 are used as input. Referring to the vehicle characteristics required by user attribute classification 33, the vehicle characteristics required by current user are calculated by the vehicle characteristics calculation unit 15, that is, the vehicle characteristics that should be recommended to the current user and output as the vehicle characteristics required by the current user 35.

[0058] Figure 10 This is a schematic diagram illustrating an example of the data structure for current user characteristic 24 in one embodiment of the present invention. Current user characteristic 24 is a table storing current attribute information for each user, including, for example, user ID (a key item), gender, current age, and current location (prefecture), etc., all related to user attributes. As mentioned above, the attribute items are not limited to... Figure 10 The content shown can also include other attribute items such as occupation and family composition, and the method of dividing residential areas can adopt a different zoning standard than "prefecture".

[0059] Figure 11 This is a schematic diagram illustrating an example of the data structure for the current user's desired vehicle model characteristic 35 in one embodiment of the present invention. The current user's desired vehicle model characteristic 35 is a table that stores the current attributes of each user and the characteristic information of their desired vehicle model, i.e., information reflecting the current desired vehicle model characteristics of each user. For example, it includes items related to user attributes such as user ID (key item), gender, current age, and current location (prefecture), as well as items related to characteristic values ​​such as characteristic 1, characteristic 2, characteristic 3, ...

[0060] The values ​​of each item here, such as characteristic 1, characteristic 2, characteristic 3, ..., are as follows: Figure 2 As shown in (b), the values ​​of each item (characteristic 1, characteristic 2, characteristic 3, ...) obtained from the user's required vehicle model characteristics 33 based on the target user's current attribute information (based on the user's current attribute needs (preferences)) are added to the values ​​of each item (characteristic 1, characteristic 2, characteristic 3, ...) in the target user's user-categorized values ​​34 (the current user's values). That is, the values ​​of each item (characteristic 1, characteristic 2, characteristic 3, ...) here represent the characteristics of the vehicle model that should be recommended to the user.

[0061] return Figure 1Next, taking the current user's desired vehicle characteristic 35 as input, the recommended vehicle calculation unit 16 calculates candidate specific vehicle models based on vehicle classification characteristic values ​​31, and further outputs them as specific recommended vehicle models 36 based on vehicle classification performance parameters 21. Specifically, one or more vehicle models with similar characteristic values ​​to characteristic 1, characteristic 2, characteristic 3, ... in the current user's desired vehicle characteristic 35 are extracted from the vehicle classification characteristic values ​​31 as candidates for specific recommended vehicle models. Furthermore, the method for judging similarity is not particularly limited; for example, appropriate methods such as vectorizing the values ​​of characteristic 1, characteristic 2, characteristic 3, ... can be used, and the similarity of these vectors can be evaluated.

[0062] Figure 12 This is a schematic diagram illustrating an example of the data structure for recommended vehicle model 36 in one embodiment of the present invention. Recommended vehicle model 36 is a table storing vehicle models and their characteristic information recommended to each user. For example, it includes key items such as user ID, manufacturer, body shape, model name, and year / model year, as well as performance parameters such as new car price, horsepower, fuel consumption, and vehicle length / width / height.

[0063] also, Figures 3 to 12 The data structure examples shown are logical table structures (views). In reality, for example in a database, multiple logical tables can be mounted as one or more physical tables in normalized form.

[0064] Furthermore, this embodiment uses the current car ownership situation of each user as an example, i.e., the user considering changing cars, but it is not limited to this. As long as information that can identify the characteristics of the car model needed by the user can be obtained, whether or not the user currently owns a car is not a necessary condition, and this also applies to the case of first-time car purchase. For example, if it is known from the user's web browsing history that they have frequently viewed information about a specific car model, then the characteristics of that car model can be treated as the characteristics of the car model needed by the user.

[0065] <Example of hardware structure for a car recommendation system>

[0066] Combination Figure 13 The hardware structure of a server device included in the car recommendation system 1 will be described as an example. The server device 1300 includes a memory 1301, a processor 1302, a communication interface 1303, a storage 1304, an input interface 1305, and a bus 1306, etc.

[0067] Memory 1301 is a volatile memory such as DRAM (Dynamic Random Access Memory) used for temporary storage of programs and calculation results of a processor (such as a CPU). Processor 1302 includes one or more processors, such as a CPU. Communication interface 1303 includes an interface for wired or wireless communication, enabling communication with, for example, an information processing terminal via the Internet or a specific network. Communication interface 1303 includes communication circuitry for communicating with the information processing terminal. Memory 1304 includes storage devices such as hard disk drives (HDDs) or solid-state drives (SSDs), for example, storing middleware such as operating systems (OS), database management systems (DBMSs), web server programs, and software recommended for automobiles. Input interface 1305 includes a keyboard, etc., for receiving operations from administrators managing the server device; however, if the server device is operated remotely via a network, an input interface may not be necessary. Bus 1306 is used for transmitting data between various modules of the server device.

[0068] The processor 1302 expands the computer program stored in the memory 1304 into the memory 1301 and executes it, thereby achieving the following: Figure 1 The functions of each part, including the vehicle characteristic calculation unit 11, the user-required vehicle characteristic calculation unit 12, the user-attribute-classified vehicle characteristic calculation unit 13, the user-classified value calculation unit 14, the current user-required vehicle characteristic calculation unit 15, and the recommended vehicle calculation unit 16.

[0069]

[0070] Combination Figure 14 This document describes a series of operations performed in the car recommendation system, specifically the car recommendation process (hereinafter referred to as "recommendation process"). The recommendation process can be implemented, for example, by having the processor 1302 of the server device 1300 expand a computer program stored in memory 1304 into memory 1301 and execute it.

[0071] In addition, Figure 14 Before the recommendation process begins, processor 1302 applies a machine learning model to the performance parameter data of each vehicle model stored in the vehicle model classification performance parameter 21 to calculate the characteristic value of each vehicle model. Processor 1302 stores the calculated characteristic value of each vehicle model (as vehicle model classification characteristic value 31) in memory 1304. Furthermore, the process of calculating the vehicle model characteristic value corresponds to the process of the vehicle model characteristic calculation unit 11 described above.

[0072] The characteristic values ​​of each vehicle model are obtained by reducing the performance parameter data of each model to a feature vector of a predetermined dimension. This feature vector is expressed in a dimension lower than that of the performance parameter data, but still retains a certain degree of the original information. This "certain degree of retention" refers to the extent to which characteristic differences are preserved from the trends in the performance parameter data of the vehicle models as a group. Characteristic differences refer to, for example, the degree to which the dimensionality reduction preserves the perspectives that users value when choosing a car. In other words, the vehicle performance parameter data is mapped to a predetermined-dimensional vector space (the characteristic values ​​of the vehicle model) through a machine learning model, allowing for precise comparison of the characteristics of each vehicle model. Furthermore, through mapping using the machine learning model, the vector space of the characteristic values ​​of each vehicle model is constructed as a space that reflects (weightedly represents) the performance aspects (such as horsepower, price, etc.) that are easily valued in the car purchase decision. In this embodiment, by using vectorized vehicle characteristic values, meaningful operations can be performed between multiple vehicle characteristic values. For example, if the average value (average vector) of multiple vehicle characteristic values ​​is calculated, this average value represents the average characteristic of the multiple vehicle models. Furthermore, the difference between a specific vehicle characteristic value and the average value of the multiple vehicle characteristic values ​​represents the difference between the specific vehicle characteristic and the average characteristic of the multiple vehicle models. Previously, directly using vehicle performance parameter data as feature values ​​easily resulted in high-dimensional vectors, causing small differences in performance parameter data to accumulate and making it difficult to compare vehicles based on characteristic differences. On the other hand, in this embodiment, by using data (vehicle feature values) obtained by mapping the performance parameter data of each vehicle to a vector of predetermined dimensions, high-precision recommendations based on data emphasizing characteristic differences between vehicle models can be achieved.

[0073] Furthermore, the following processing describes the process for a specific user. Therefore, if you want to recommend car models to multiple users, you can repeat the processing described below for each user.

[0074] In S1401, processor 1302 obtains the attributes of a specific user at a predetermined time, the vehicle model information held at the predetermined time, and the current attributes. For example, processor 1302 obtains the aforementioned information of a specific user input through the information processing terminal included in the car recommendation system 1 via communication interface 1303. Alternatively, processor 1302 obtains the user identifier input through the information processing terminal, and then may also obtain the aforementioned information of the specific user from the vehicle model held by the user 22, the user attribute 23 at the start of holding, and the current user attribute 24.

[0075] In S1402, processor 1302 determines a first group of users whose attributes are the same as or similar to those of a specific user at a predetermined time. For example, processor 1302 may identify users whose attributes (such as gender, residential area, and age classification) are consistent with those of a specific user at a predetermined time as the first group of users. Processor 1302 may, for example, search for the first group of users from the user attributes 23 at the time of initial possession. The attributes of a specific user at the predetermined time may also include family composition and hobbies, etc.

[0076] In S1403, processor 1302 calculates the characteristic values ​​of the vehicle models required by the first group of users. For example, processor 1302 obtains the vehicle models required by each user in the first group from the vehicle models 22 held by the users, and uses the vehicle model classification characteristic value 31 to obtain the characteristic values ​​of the vehicle models required by each user in the first group. Further, processor 1302 calculates, for example, the average of the obtained vehicle model characteristic values, and uses it as the characteristic value of the vehicle models required by the first group of users. The processing of S1402 and S1403 corresponds to the processing of the vehicle model characteristic calculation unit 13 classified by user attributes. Processor 1302 can store the vehicle model characteristic values ​​required by the first group of users in the vehicle model characteristic 33 classified by user attributes. Through the processing of S1402 and S1403, the vehicle model characteristics required by multiple users with specific attributes can be processed as a vectorized characteristic value.

[0077] In S1404, processor 1302 obtains the characteristic values ​​of the vehicle model required by a specific user at a predetermined time from the vehicle model information held at the predetermined time. This process corresponds to the processing of the user-required vehicle model characteristic calculation unit 12. Based on the aforementioned vehicle model information held by the specific user at the predetermined time and the vehicle model classification characteristic value 31, processor 1302 obtains the characteristic values ​​of the vehicle model required by the specific user at the predetermined time. Processor 1302 can store the vehicle model characteristic values ​​for the specific user in the user-classified required vehicle model characteristic 32.

[0078] In S1405, processor 1302 calculates the difference between the vehicle characteristic values ​​required by the first group of users and the vehicle characteristic values ​​required by a specific user at a predetermined time, and stores this difference as an indicator representing user values ​​(stored in user-categorized value 34). This process corresponds to the process of user-categorized value calculation unit 14. The calculated difference is a vectorized value of a predetermined dimension, and therefore can be used as a computable quantity to express the direction and magnitude of user values ​​within the vehicle characteristic-related space.

[0079] In S1406, processor 1302 determines a second group of users who belong to the same or similar attributes as the current attributes of a specific user. For example, processor 1302 can determine users whose current attributes (e.g., gender, residential area, and age classification) are consistent with those of the specific user as the second group of users. In S1407, the vehicle characteristics required by the second group of users are added to an indicator representing the values ​​of the specific user to calculate a combined vehicle characteristic value (equivalent to the vehicle characteristics required by the current specific user). For example, processor 1302 can obtain the vehicle characteristics required by the second group of users (from the vehicle characteristics 33 required by user attribute classification) and add them to the indicator representing the values ​​of the specific user. The value (characteristic value) obtained by addition is a vectorized value of a predetermined dimension, thus expressing the vehicle characteristics required by the current specific user within the vehicle characteristic related space. The processing of S1406 and S1407 corresponds to the processing of the vehicle characteristics calculation unit 15 required by the current user. Processor 1302 can store the added value (characteristic value) in the vehicle characteristics required by the current user 35.

[0080] In S1408, processor 1302 determines candidate vehicle models (e.g., those with similar characteristics) based on the merged vehicle characteristic values ​​and the actual characteristic values ​​of each vehicle model (stored in the vehicle model classification characteristic value 31). For example, processor 1302 can compare the merged vehicle characteristic values ​​with the characteristics of each vehicle model to extract the characteristic values ​​of one or more vehicle models within a predetermined Euclidean distance (actual vehicle characteristic values ​​similar to the merged vehicle characteristic values). Then, processor 1302 determines the candidate vehicle models corresponding to the extracted vehicle characteristic values ​​as recommended target vehicle models (candidate vehicle models). Furthermore, if there are multiple candidate vehicle models, they can be sorted according to their similarity to the merged vehicle characteristic values; the higher the similarity, the higher the priority, and higher-priority models can be displayed at the top.

[0081] In S1409, processor 1302 recommends candidate vehicle models to a specific user. The processes in S1408 and S1409 correspond to the processing of the vehicle model recommendation calculation unit 16. For example, processor 1302 can provide a list of candidate vehicle models to the information processing terminal and display the list on the information processing terminal's display. After executing the processing in S1409, processor 1302 terminates the series of operations of this process.

[0082] As described above, the car recommendation system 1 according to one embodiment of the present invention can infer the differences and changes in needs caused by differences in user attributes, as well as the values ​​of each user, based on the performance indicators of the car models owned by the user, thereby recommending cars. Moreover, in this way, it is possible to distinguish the tendencies categorized by user attributes and the values ​​of each user, and to recommend different cars as life stages change.

[0083] The invention described above is based on specific embodiments, but the invention is not limited to the above embodiments, and various modifications can be made without departing from its spirit. Furthermore, the above embodiments are detailed to facilitate understanding of the invention, but are not limited to having all the described components. Additionally, regarding a portion of the above embodiments, other components can be added, deleted, or substituted.

[0084] Furthermore, some or all of the aforementioned components, functions, processing units, and processing methods can be implemented in hardware, such as through design in integrated circuits. Alternatively, the aforementioned components and functions can be implemented in software by a processor interpreting and executing programs that perform their respective functions. The programs, tables, files, and other information implementing each function can be stored in storage devices such as memory, hard disks, and SSDs, or in recording media such as IC cards, SD cards, and DVDs.

[0085] Furthermore, the control lines and information lines shown in the above diagrams are for illustrative purposes only and may not represent all control lines and information lines in the installation. In practice, it can be assumed that almost all components are interconnected.

[0086] [Industrial practicality]

[0087] This invention can be applied to car recommendation systems that recommend cars.

[0088] [Symbol Explanation]

[0089] 1. Car Recommendation System

[0090] 11 Vehicle Characteristic Calculation Department

[0091] 12 User-Required Vehicle Model Characteristics Calculation Department

[0092] 13. Calculation of vehicle characteristics required for user attribute classification

[0093] 14. User-categorized value calculation department

[0094] 15. Calculation of Vehicle Characteristics Required by Current Users

[0095] 16 Recommended Model Calculation Department

[0096] 21 Performance parameters categorized by vehicle type

[0097] 22 user-owned vehicle models

[0098] 23 User attributes at the start of ownership

[0099] 24 Current User Attributes

[0100] 31. Characteristic values ​​categorized by vehicle type

[0101] 32. Vehicle characteristics required for user classification

[0102] 33. Classify required vehicle characteristics by user attributes

[0103] 34. User-categorized values

[0104] 35. Current user-required vehicle characteristics

[0105] 36 Recommended car models.

Claims

1. A car recommendation system, which is a system that recommends cars to users. It includes memory for storing instructions and one or more processors. When the instruction is executed, the one or more processors cause the car recommendation system to perform the following operations: Identify the first group of users who belong to the same or similar attributes as the specific user at the scheduled time; An index representing the user's values ​​is calculated based on the difference between the vehicle characteristic values ​​required by the first group of users and the vehicle characteristic values ​​required by the specific user at the predetermined time. Identify a second group of users who belong to the same or similar attributes as the current attributes of the specific user; The combined vehicle characteristic value is calculated based on the vehicle characteristic value required by the second group of users and the index representing the values ​​of the specific user. Candidate models are determined based on the combined and calculated model characteristic values ​​and the characteristic values ​​of each model. The candidate vehicle model is recommended to the specific user.

2. The car recommendation system according to claim 1, wherein, Determining the candidate vehicle model includes identifying the vehicle characteristic values ​​that are within a predetermined distance range after distance merging and calculation as the candidate vehicle characteristic values, and identifying the vehicle models with the candidate vehicle characteristic values ​​as the candidate vehicle models.

3. The car recommendation system according to claim 1, wherein, Determining the candidate vehicle models includes determining multiple candidate vehicle models to be displayed on the display based on the combined and calculated vehicle characteristics and the characteristics of each vehicle model; The candidate models are sorted according to their similarity to the merged model characteristic values. The higher the similarity, the higher the priority, and the higher the priority model is displayed in a more prominent position.

4. The car recommendation system according to claim 1, wherein, The characteristic values ​​of a vehicle model are data whose performance parameters are mapped to a vector space that allows for comparison of the characteristics of different vehicle models.

5. In the car recommendation system according to claim 1, When the instruction is executed, the car recommendation system performs the following operations: By using a machine learning model, the performance parameter data of the vehicle model is mapped to a feature vector that is expressed in a lower dimension than the performance parameter data, and the characteristic value of the vehicle model is calculated.

Images