Information processing device, information processing method, program, and information processing system

Abstract

This invention provides an information processing device, information processing method, program, and information processing system that calculate an appropriate valuation price based on the vehicle's maintenance history. [Solution] The information processing system obtains vehicle information of the target vehicle from an external vehicle information database, extracts recommended maintenance items from an external recommended maintenance database, obtains maintenance history information from an external maintenance history database, compares the recommended maintenance items with the maintenance history information to calculate a maintenance evaluation score, calculates an evaluation price based on the maintenance evaluation score, applies a weighting coefficient to important maintenance items to perform an important maintenance evaluation, and adjusts the score based on the reliability of the maintenance factory.

Description

[Technical Field]

[0001] The present invention relates to an information processing apparatus, an information processing method, a program, and an information processing system that perform information processing based on vehicle maintenance history information. [Background Art]

[0002] In the used vehicle trading market, vehicle price evaluation has been performed mainly based on the model year, mileage, and the conditions of the exterior and interior as main evaluation criteria. These evaluation criteria are widely used as indicators that indirectly indicate the usage status of a vehicle.

[0003] However, conventional evaluation methods have a problem that the maintenance history of a vehicle is not appropriately reflected in the price. Specifically, when a vehicle that has regularly undergone scheduled inspection and maintenance and had important parts replaced appropriately and a vehicle that has been neglected in maintenance have the same model year, grade, equipment and mileage, they may be traded at the same price.

[0004] Such a situation is not only unfair to vehicle owners who have spent costs on appropriate maintenance, but also causes a problem that it is difficult for vehicle purchasers to objectively evaluate the safety and reliability of the vehicle.

[0005] Further, as a method for reflecting maintenance history in prices, a method of adding a fixed amount for a specific maintenance item is conceivable. For example, it is a method of adding a predetermined amount when a vehicle inspection is performed. However, such a simple addition rule has a weak correlation with actual transaction prices in the market, and can hardly be said to be an appropriate price evaluation. [Prior Art Documents] [Patent Documents]

[0006] [Patent Document 1] Japanese Unexamined Patent Application Publication No. 2002-259753 [Summary of the Invention] [Problem to be Solved by the Invention]

[0007] This Patent Document 1 describes storing maintenance status data in a maintenance status database for used vehicles subject to evaluation, which includes at least one of the following data: data on replacement of consumable parts, inspection and maintenance data, accident and failure data, and data from the on-board computer. It also describes calculating a basic evaluation value based on appraisal item data, which includes at least one of the following items other than maintenance status: manufacturer name, vehicle model, type, grade, year of manufacture, paint color, and exterior condition. Furthermore, Patent Document 1 describes calculating a maintenance status evaluation value based on the contents of a maintenance status database, and performing a price evaluation of the used vehicle based on the basic evaluation value and the maintenance status evaluation value. However, Patent Document 1 only describes calculating a basic evaluation value based on assessment item data that includes at least one of the following: manufacturer name, vehicle model, type, grade, year of manufacture, paint color, and exterior condition.

[0008] The present invention has been made in view of the problems of the prior art described above, and aims to provide an information processing device, an information processing method, a program, and an information processing system capable of performing information processing based on vehicle maintenance history information. [Means for solving the problem]

[0009] To solve the above problems, an information processing device according to a first aspect of the present invention is an information processing device for calculating the appraised value of a vehicle, comprising: a vehicle information acquisition unit that acquires vehicle information including the vehicle type, year of manufacture, and mileage of a target vehicle from an external vehicle information database; a recommended maintenance extraction unit that extracts recommended maintenance items corresponding to the vehicle type, year of manufacture, and mileage from an external recommended maintenance database; a maintenance history acquisition unit that acquires maintenance history information showing the maintenance history performed on the target vehicle from an external maintenance history database; a maintenance implementation evaluation unit that compares the extracted recommended maintenance items with the maintenance history information and calculates a maintenance evaluation score showing the status of maintenance performed for the recommended maintenance items; and a price calculation unit that calculates the appraised value of the target vehicle based on the maintenance evaluation score.

[0010] The information processing method according to the present invention is an information processing method for calculating the appraised value of a vehicle, which is performed by a computer, and includes the steps of: obtaining vehicle information including the vehicle type, year of manufacture, and mileage of a target vehicle from an external vehicle information database; extracting recommended maintenance items corresponding to the vehicle type, year of manufacture, and mileage from an external recommended maintenance database; obtaining maintenance history information showing the maintenance history performed on the target vehicle from an external maintenance history database; comparing the extracted recommended maintenance items with the maintenance history information and calculating a maintenance evaluation score showing the status of maintenance performed for the recommended maintenance items; and calculating the appraised value of the target vehicle based on the maintenance evaluation score.

[0011] The program according to the present invention is a program for causing a computer to function as an information processing device for calculating the appraised value of a vehicle, wherein the computer functions as: a vehicle information acquisition means for acquiring vehicle information including the vehicle type, year of manufacture, and mileage of a target vehicle from an external vehicle information database; a recommended maintenance extraction means for extracting recommended maintenance items corresponding to the vehicle type, year of manufacture, and mileage from an external recommended maintenance database; a maintenance history acquisition means for acquiring maintenance history information showing the maintenance history performed on the target vehicle from an external maintenance history database; a maintenance implementation evaluation means for comparing the extracted recommended maintenance items with the maintenance history information and calculating a maintenance evaluation score showing the status of maintenance performed on the recommended maintenance items; and a price calculation means for calculating the appraised value of the target vehicle based on the maintenance evaluation score.

[0012] The information processing system according to the present invention is an information processing system for calculating the appraised value of a vehicle, comprising: a vehicle information database that stores vehicle information including vehicle type, year of manufacture, and mileage; a recommended maintenance database that stores recommended maintenance data including recommended maintenance items corresponding to the vehicle type, year of manufacture, and mileage; a maintenance history database that stores maintenance history information showing the history of maintenance performed on the vehicle; a sales history database that stores the past sales history of the vehicle; and a market price information database that stores market price information showing the market transaction price range of the vehicle, and the vehicle information database, the recommended maintenance database, the maintenance history database, the sales history database, and the market price information database are all connected in a manner that allows communication between them. The information processing device comprises: a vehicle information acquisition unit that acquires the vehicle information of a target vehicle from the vehicle information database; a recommended maintenance extraction unit that extracts recommended maintenance items corresponding to the vehicle type, year of manufacture, and mileage of the target vehicle from the recommended maintenance database; a maintenance history acquisition unit that acquires the maintenance history information of the target vehicle from the maintenance history database; a maintenance implementation evaluation unit that compares the extracted recommended maintenance items with the maintenance history information and calculates a maintenance evaluation score indicating the status of maintenance implementation for the recommended maintenance items; and a price calculation unit that calculates the evaluated price of the target vehicle within a price range obtained from the market price information database based on the maintenance evaluation score. [Effects of the Invention]

[0013] According to the present invention, by calculating a maintenance evaluation score using the vehicle's maintenance history, properly maintained vehicles are fairly evaluated, and fair pricing is achieved. Furthermore, according to the present invention, by applying weighting coefficients to important maintenance items, adjusting scores based on the reliability of the maintenance shop, and conducting correlation analysis with market prices using a learned model, it becomes possible to calculate a fair price that reflects actual market conditions. Specifically, by considering the maintenance history compared to conventional price evaluations based only on year of manufacture, grade, equipment, and mileage, the accuracy of price evaluation, i.e., the correlation with actual transaction prices, can be improved. Furthermore, according to the present invention, the generation of a maintenance history guarantee certificate allows a vehicle purchaser to objectively grasp the safety and reliability of a vehicle, which improves the transparency of transactions. This promotes the distribution of vehicles with good maintenance history, and contributes to improving the quality of the entire used vehicle market. Brief Description of the Drawings

[0014] [Figure 1] Figure 1 is a block diagram showing the overall configuration of an information processing system according to an embodiment of the present invention. [Figure 2] Figure 2 is a flowchart of an evaluated price calculation process according to an embodiment of the present invention. [Figure 3] Figure 3 is a diagram for explaining a comparison process between recommended maintenance items and maintenance history information. [Figure 4] Figure 4 is a diagram showing an example of a maintenance history guarantee certificate and a price presentation screen. [Figure 5] Figure 5 is a diagram showing an example of important maintenance items categorized by vehicle model. [Figure 6] Figure 6 is a diagram showing a configuration example of a maintenance shop reliability table. [Figure 7] Figure 7 is a diagram showing the configuration and processing flow of a trained model. Mode for Carrying Out the Invention

[0015] Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. It should be noted that the following embodiments do not limit the present invention, and various modifications are possible within the scope of the technical idea of the present invention.

[0016] Figure 1 is a block diagram showing the overall configuration of an information processing system 1 according to an embodiment of the present invention. The information processing system 1 includes an information processing apparatus 10, a plurality of databases, and a dealer terminal 2.

[0017] The information processing device 10 is configured by a computer including a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), a storage device, and a communication interface. The information processing device 10 may be implemented as a server computer, or may be implemented in a cloud computing environment.

[0018] The information processing device 10 is communicably connected via a communication network to a vehicle information database 3, a sales history database 4, a market price information database 5, a maintenance history database 6, a recommended maintenance database 7, and a maintenance shop reliability database 8. These databases are stored in a storage device provided outside the information processing device 10. Of course, at least one of these databases may be stored inside the information processing device 10.

[0019] In order to improve access efficiency to databases, the information processing device 10 may be provided with a cache mechanism for data with low update frequency such as market price information. In addition, in order to process a large number of vehicle evaluation requests, the information processing device 10 may be horizontally scaled across a plurality of servers, or may be configured to perform load distribution via a load balancer.

[0020] The information processing device 10 has an error handling function that records error information in a log and transmits an error notification to the dealer terminal 2 when a database connection error, data inconsistency, or timeout occurs. In addition, when maintenance history information cannot be acquired, the information processing device 10 continues processing by treating the maintenance history as unknown, and calculates a maintenance evaluation score as a predetermined initial value. The predetermined initial value is, for example, 0 points or a median value.

[0021] The vehicle information database 3 is a storage unit that stores vehicle information for each vehicle, including vehicle ID, vehicle type, year of manufacture, mileage, engine displacement, drive system, transmission type, body color, and optional equipment. The vehicle ID is set as the primary key and is a unique identifier. A unique identifier is, for example, the chassis number or management number. The vehicle type is stored as a string type, the year of manufacture is stored as a four-digit integer type representing the year, and the mileage is stored as an integer type representing kilometers. The vehicle information database 3 has a composite index set for combinations of vehicle type, year of manufacture, and mileage to speed up search processing. The sales history database 4 is a storage unit that stores transaction information of vehicles that have been bought and sold in the past. The transaction information includes the vehicle ID, transaction date, transaction price, transaction type, and the condition of the vehicle at the time of the transaction. The transaction type is whether it was an auction or a retail sale. The market price database 5 is a memory unit that stores market price ranges corresponding to combinations of vehicle type, year of manufacture, and mileage. The market price range includes an upper price limit, a lower price limit, and a median price. The market price information is updated periodically based on retail price information from the internet, auction winning bids, and inter-dealer transaction prices. The maintenance history database 6 is a storage unit that stores the maintenance history performed on each vehicle. The maintenance history information includes the maintenance history ID, vehicle ID, maintenance date, maintenance details, repair shop ID, type of repair shop, type of parts used, and maintenance cost. The maintenance history ID is set as the primary key. The vehicle ID is set as a foreign key and refers to the vehicle information database 3. The repair shop ID is set as a foreign key and refers to the repair shop reliability database 8. The maintenance date is stored as a date type, the maintenance details are stored as a string type, and the maintenance cost is stored as an integer type in yen. The type of repair shop is a classification code indicating a dealer, certified repair shop, designated repair shop, or specialty shop. The type of parts used is a classification code indicating genuine, OEM, aftermarket, or remanufactured parts. The maintenance history database 6 has an index set up for the vehicle ID, allowing for high-speed retrieval of the maintenance history of a specific vehicle. The recommended maintenance database 7 is a storage unit that stores recommended maintenance data, including recommended maintenance items corresponding to combinations of vehicle type, year, and mileage. Recommended maintenance items include periodic inspections such as semi-annual inspections, annual inspections, and vehicle inspections, as well as oil changes, filter changes, brake-related maintenance, and timing belt replacements. Furthermore, each recommended maintenance item is categorized into routine maintenance and critical maintenance. The repair shop reliability database 8 is a memory unit that stores reliability points set for each repair shop. The reliability points are set on a scale of 10 from 0 to 9, with 5 being the median or baseline value.

[0022] Dealer terminal 2 is a terminal device installed in a used car dealership and is connected to information processing device 10 via a communication network. Dealer terminal 2 receives and displays the vehicle ID of the target vehicle and the calculated appraisal price and maintenance history warranty from information processing device 10.

[0023] The information processing device 10 includes, as functional blocks, a vehicle information acquisition unit 101, a recommended maintenance extraction unit 102, a maintenance history acquisition unit 103, a maintenance implementation evaluation unit 104, a critical maintenance evaluation unit 105, a maintenance factory reliability acquisition unit 106, a market price information acquisition unit 107, a price calculation unit 108, and a maintenance history guarantee certificate generation unit 109.

[0024] These functional blocks are realized by the CPU of the information processing device 10 reading a program stored in ROM or a storage device into RAM and executing it. In other words, the program according to this embodiment is a program that causes the computer to function as each of the above functional blocks.

[0025] The vehicle information acquisition unit 101 acquires vehicle information for the target vehicle from the vehicle information database 3 based on the vehicle ID entered from the dealer terminal 2. The vehicle information includes the make, model year, and mileage.

[0026] The recommended maintenance extraction unit 102 extracts the appropriate recommended maintenance items from the recommended maintenance database 7 based on the vehicle type, year, and mileage acquired by the vehicle information acquisition unit 101. The recommended maintenance extraction unit 102 has a function to dynamically switch the table it references according to the mileage. Specifically, it refers to the first recommended maintenance table when the mileage is below a predetermined threshold, and to the second recommended maintenance table when the mileage is above the predetermined threshold. The predetermined threshold is, for example, "100,000 kilometers," and the second recommended maintenance table includes maintenance items specific to high-mileage vehicles, such as timing belt replacement, engine overhaul, and transmission overhaul, in addition to the items included in the first recommended maintenance table.

[0027] The maintenance history acquisition unit 103 acquires maintenance history information for the target vehicle from the maintenance history database 6 based on the vehicle ID.

[0028] The maintenance implementation evaluation unit 104 compares the recommended maintenance items extracted by the recommended maintenance extraction unit 102 with the maintenance history information acquired by the maintenance history acquisition unit 103, and calculates a maintenance evaluation score indicating the status of maintenance implementation for the recommended maintenance items. The maintenance implementation evaluation unit 104 assigns a score to each recommended maintenance item based on the following evaluation criteria: If maintenance is performed at the recommended time, a base score is assigned. The base score is, for example, "10 points". If maintenance is performed at a time other than the recommended time, 50 percent of the base score is assigned. 50 percent is, for example, 5 points. If maintenance has not been performed, "0 points" is assigned.

[0029] The calculation logic for the maintenance evaluation score will be explained in detail. The maintenance evaluation score is a numerical value calculated by aggregating the scores assigned to each recommended maintenance item, and is expressed in the range of 0 to 100. First, let N be the total number of recommended maintenance items extracted by the recommended maintenance extraction unit 102, and let Si be the individual score for each recommended maintenance item i. The individual score Si is assigned as follows: a base score (e.g., 10 points) if maintenance is performed during the recommended period; 50% of the base score (e.g., 5 points) if maintenance is performed outside of the recommended period; and 0 points if maintenance is not performed. Next, let Wi be the weighting factor for each recommended maintenance item i. The weighting factor Wi can be set, for example, to "1.0" for normal maintenance items and to a range of "1.5 to 2.0" for critical maintenance items. The base value of the maintenance evaluation score is calculated by dividing the sum of the individual scores for each recommended maintenance item multiplied by a weighting factor, by the maximum possible score, and multiplying by 100. This maximum possible score is the sum of the weighted scores assuming all recommended maintenance items are performed at the recommended time. In other words, the base value of the maintenance evaluation score is calculated by multiplying the sum of the individual scores by a weighting factor, dividing that value by the sum of the base scores multiplied by a weighting factor, and then multiplying the result by 100.

[0030] The critical maintenance evaluation unit 105 identifies maintenance items related to vehicle safety or reliability from among the recommended maintenance items as critical maintenance items. Critical maintenance items include timing belt replacement, brake fluid replacement, and cooling system maintenance. The critical maintenance evaluation unit 105 calculates a maintenance evaluation score by applying a predetermined weighting factor to the critical maintenance items. The predetermined weighting factor is, for example, 1.5 to 2.0 times. Furthermore, the critical maintenance evaluation unit 105 considers critical maintenance items specific to each vehicle model. Critical maintenance items specific to each vehicle model are maintenance items related to weak points or parts prone to failure that are unique to that vehicle model, and are mapped to each vehicle model in the recommended maintenance database 7. For example, for certain vehicle models, critical maintenance items specific to each vehicle model include periodic inspections of the suspension system targeting tie rod ends and bushings, preventive replacement of the cooling system targeting water pumps and thermostats, early detection and repair of oil leaks, and rustproofing of the undercarriage.

[0031] The repair shop reliability acquisition unit 106 acquires repair shop reliability information from the repair shop reliability database 8 based on the identification information of the repair shop included in the maintenance history information.

[0032] The maintenance implementation evaluation unit 104 adjusts the maintenance evaluation score based on the reliability information of the maintenance factory. Specifically, it adds a predetermined percentage when the reliability points exceed the standard value and subtracts a predetermined percentage when the reliability points fall below the standard value. The standard value is 5, and the predetermined percentage is, for example, "1% to 4%".

[0033] This section explains how to calculate the adjusted maintenance evaluation score based on the reliability of the repair shop. An adjustment coefficient based on the reliability of the repair shop is applied to the base value of the maintenance evaluation score. The adjustment coefficient is determined for each maintenance item included in the maintenance history information, based on the reliability points of the maintenance shop that performed the maintenance. If multiple maintenance items are performed at different maintenance shops, a weighted average of the reliability points of each maintenance shop is calculated, and the adjustment coefficient is determined based on this weighted average. In calculating the weighted average, the importance or cost of each maintenance item can be used as a weight. The final maintenance evaluation score is the base value of the maintenance evaluation score, adjusted by adding or subtracting based on an adjustment coefficient. For example, if the weighted average of the maintenance shop reliability exceeds the standard value of 5, an addition of 1% to 4% is made depending on the excess. If it falls below the standard value, a subtraction of 1% to 4% is made depending on the deficit.

[0034] Next, we will explain an example of setting confidence points. Authorized dealerships are assigned a reliability score of "8 to 9". Designated workshops are assigned a reliability score of "7". Certified workshops are assigned a reliability score of "6". Highly-rated specialty car shops are assigned a reliability score of "8 to 9". General repair shops are assigned a reliability score of "5". Workshops with unclear service records are assigned a reliability score of "3 or less".

[0035] Furthermore, if there is a maintenance history at a specialist shop that is familiar with the specific weak points of the vehicle model, the Important Maintenance Evaluation Section 105 will award additional points. A specialist shop that is familiar with the specific weak points of the vehicle model would be, for example, a repair shop that specializes in handling certain European cars.

[0036] The market information acquisition unit 107 acquires the market transaction price range for vehicles of the same or similar type as the target vehicle from the market information database 5, based on the vehicle type, year of manufacture, and mileage acquired by the vehicle information acquisition unit 101. The market transaction price range includes the upper limit price, lower limit price, and median price.

[0037] The "market transaction price range" refers to the range of market transaction prices for vehicles of the same or similar make, model year, and mileage as the vehicle in question, and includes a statistically calculated upper limit price, lower limit price, and median price. The upper limit price is the upper predetermined value of the transaction price within that price range, the lower limit price is the lower predetermined value, and the median price is the median or average value. "Transaction price" refers to the price at which a sale was actually completed in the past, and is the actual price recorded in the sales history database 4. The "base valuation price" is the benchmark price before considering maintenance valuation, and corresponds to the median price in the market transaction price range. In this specification, the term "base selling cost A" has the same meaning. "Evaluation price" refers to the recommended transaction price of the subject vehicle calculated according to this invention, and is the price after applying a correction based on the maintenance evaluation score to the base evaluation price. The term "post-update sales cost B" is synonymous. Furthermore, in this specification, the term "cost price" refers to either the purchase price in the used car sales business or the price that forms the basis for calculating the selling price.

[0038] The price calculation unit 108 calculates the appraised value of the target vehicle based on the maintenance evaluation score. The price calculation unit 108 uses one of the two calculation methods described below. The first calculation method is a table-based evaluation, in which the price calculation unit 108 uses the median price of the market transaction price range acquired by the market information acquisition unit 107 as the base cost of goods sold A, and applies a correction coefficient based on the maintenance evaluation score to calculate the updated cost of goods sold B.

[0039] The updated sales cost B is the value obtained by multiplying A by the sum of 1 and the maintenance evaluation score correction coefficient, and is expressed as "B = A × (1 + maintenance evaluation score correction coefficient)". This maintenance evaluation score correction factor is a value obtained by converting the maintenance evaluation score to a correction factor based on a predetermined conversion formula or table. For example, if the maintenance evaluation score is "90% or higher," the correction factor is increased by 0.08, or "8%." If the maintenance evaluation score is "70% or higher but less than 90%," the correction factor is increased by 0.04, or "4%." If the maintenance evaluation score is "50% or higher but less than 70%," the correction factor is set to "0 (zero)," or "no addition." If the maintenance evaluation score is "less than 50%," the correction factor is decreased by 0.04, or "4%."

[0040] The method for determining the maintenance evaluation score correction coefficient in the first calculation method will be explained in detail. The maintenance evaluation score correction coefficient is set in stages according to the value of the maintenance evaluation score. If the maintenance evaluation score is 90 or higher, the maintenance evaluation score correction factor is +0.08. In this case, the evaluation price will be 108 percent of the base evaluation price. If the maintenance evaluation score is between 70 and 90, the maintenance evaluation score correction factor is +0.04. In this case, the evaluation price will be 104 percent of the base evaluation price. If the maintenance evaluation score is between 50 and 70, the maintenance evaluation score correction factor is 0. In this case, the evaluation price will be the same as the base evaluation price. If the maintenance evaluation score is less than 50, the maintenance evaluation score correction factor is -0.04. In this case, the evaluation price will be 96 percent of the base evaluation price. Alternatively, the maintenance evaluation score correction coefficient may be calculated continuously from the maintenance evaluation score instead of using the stepped setting described above. For example, the maintenance evaluation score correction coefficient may be calculated by subtracting 50 from the maintenance evaluation score and dividing the result by 1000. In this case, when the maintenance evaluation score is 100, the correction coefficient will be +0.05, and when the maintenance evaluation score is 0, the correction coefficient will be -0.05.

[0041] The second calculation method is evaluation using a trained model. In the second calculation method, the price calculation unit 108 calculates the evaluation price using a trained model that has been trained using past vehicle sales history and maintenance history associated with the vehicle as training data.

[0042] The trained model consists of a neural network, a gradient boosting decision tree (GBDT), or an ensemble model of these. The trained model accepts the input data described below and outputs a position coefficient that indicates the appropriate position within the market price range. The position coefficient is a value in the range of "0.0 to 1.0".

[0043] The "position coefficient" is a normalized numerical value that indicates the relative value position of a subject vehicle within a given price range. The position coefficient is an index that shows where the evaluation result of the subject vehicle falls within a continuous numerical range defined by its minimum and maximum values. The minimum value of the position coefficient corresponds to the lower limit of the price range, and the maximum value of the position coefficient corresponds to the upper limit of the price range. The position coefficient is an index that aggregates multiple evaluation elements into a single numerical value, and it reflects the result of a comprehensive evaluation of the vehicle's condition. By using the position coefficient, it becomes possible to compare multiple vehicles with different evaluation elements using a common scale.

[0044] In the method using position coefficients, a trained model learns the correlation between multiple evaluation factors and market prices, and outputs a single position coefficient based on these correlations. This makes it possible to represent the complex price formation mechanism, including the interactions between evaluation items, as a single numerical value.

[0045] The input data for the trained model is described below. The input data includes basic vehicle information and maintenance history-related information. Basic vehicle information includes vehicle type, year of manufacture, mileage, engine displacement, drive system, and transmission type. Maintenance history-related information includes the maintenance evaluation score calculated by the maintenance implementation evaluation unit 104 and the critical maintenance evaluation unit 105, the critical maintenance implementation rate, the average or weighted average of the reliability points acquired by the maintenance factory reliability acquisition unit 106, whether or not critical maintenance items were performed, and the types of parts used.

[0046] The output from the trained model is a position coefficient within the market price range. A position coefficient closer to 1.0 indicates good condition and that a price closer to the upper limit of the market price range is reasonable. A position coefficient closer to 0.0 indicates poor condition and that a price closer to the lower limit of the market price range is reasonable.

[0047] This section explains the relationship between the position coefficient and the maintenance status. Vehicles in extremely good maintenance status—that is, vehicles where all recommended maintenance items have been performed at the recommended time, all critical maintenance items have been performed, and which have been serviced at a highly reliable repair shop—will have a position coefficient close to 1.0. Such vehicles will be traded at the highest price among a group of vehicles of the same make, model year, and mileage. On the other hand, vehicles in poor condition, meaning those that have not undergone many recommended maintenance items and have not had any critical maintenance items performed, will have a position coefficient close to 0.0. Such vehicles have a high risk of breakdowns and are likely to incur maintenance costs after purchase, so they will be traded at the lowest price within a group of vehicles of the same make and model.

[0048] The valuation price is calculated by adding the lower limit price to the difference between the upper limit price and the lower limit price multiplied by the position coefficient, and is expressed as "Valuation Price = Lower Limit Price + (Upper Limit Price - Lower Limit Price) × Position Coefficient".

[0049] This section explains the relationship between the first and second calculation methods, as well as the selection criteria. The first calculation method, table-based evaluation, is applied when a conversion table from maintenance evaluation scores to correction coefficients is predefined. This method is effective when a pre-trained model is not available or when rapid price calculation is required. The second calculation method, evaluation of a trained model, is applied when a trained model based on past transaction data stored in the sales history database 4 is available. This method enables highly accurate price calculation that reflects market trends. The price calculation unit 108 preferentially uses the second calculation method if a pre-trained model is available, and uses the first calculation method if a pre-trained model is unavailable or if the prediction accuracy of the pre-trained model falls below a predetermined threshold. Alternatively, both the first and second calculation methods may be executed, and the weighted average of the calculation results from both methods may be used as the final valuation price. In either method, the calculated price is output to the retailer terminal 2 as the valuation price.

[0050] Next, we will explain the training process of the pre-trained model. This pre-trained model is generated by supervised learning using past vehicle sales history stored in the sales history database 4 as training data. The training data consists of a combination of basic vehicle information, maintenance history information, and actual transaction prices. The learning algorithm learns the correlation between the input data, "basic vehicle information and maintenance history information," and the output data, "the position (price) of the actual transaction price within the market price range."

[0051] Prior to the training process, preprocessing is performed. Preprocessing includes handling missing values, normalization, encoding of categorical variables, and outlier removal.

[0052] In handling missing values, records without maintenance history will have their maintenance evaluation score filled in with "0" or the median. Additionally, missing values ​​in numerical fields will be filled in with the median or mean of that field.

[0053] In normalization, numerical variables such as mileage, year of manufacture, and maintenance costs are normalized to a range of 0 (min-max normalization) and 1 (maximum), or standardized to a mean of 0 and a standard deviation of 1 (Z-score standardization).

[0054] In categorical variable encoding, categorical variables such as vehicle type and repair shop type are quantified using one-hot encoding or label encoding.

[0055] In outlier handling, abnormal values ​​where the trading price exceeds twice the upper limit of the market price range are excluded from the training data.

[0056] This section explains hyperparameters when using neural networks. The number of neurons in the input layer is the number of dimensions of the input features, for example, from "50" to "200". The number of hidden layers is from 2 to 5. The number of neurons in the hidden layers is from 64 to 256. The activation function is RELU. The activation function of the output layer is the sigmoid function to obtain an output from 0 to 1. The loss function is the mean squared error MSE or mean absolute error MAE. The learning rate is from 0.001 to 0.01. The batch size is from 32 to 128. The number of epochs is from 100 to 500. The dropout rate to prevent overfitting is from 0.2 to 0.5. The weight decay coefficient in L2 regularization is from 0.0001 to 0.001.

[0057] There is a correlation between maintenance history information and transaction price, based on so-called common technical knowledge among those in the industry. In other words, vehicles that have undergone regular inspections and maintenance and replacement of important parts have high mechanical reliability and a low risk of failure, and therefore tend to be of higher value to buyers. Therefore, vehicles with a good maintenance history tend to be traded at a relatively higher price compared to other vehicles of the same make, model year, and mileage. A trained model learns these correlations from a large amount of transaction data (such as buying and selling data), enabling the estimation (prediction) of a fair price based on maintenance history information.

[0058] As a common technical understanding of the existence of the above correlation, for example, Firstly, based on the principles of preventive maintenance in automotive engineering, regular inspections and maintenance allow for the early detection and correction of wear, deterioration, and malfunctions of parts, thus increasing the likelihood of maintaining the vehicle's mechanical reliability. Secondly, maintenance at dealerships and certified workshops tends to be of higher quality compared to general repair shops because they use more genuine parts and standardize their maintenance procedures. Thirdly, in the used car market, vehicles with a clear service history and dealer service records are more likely to gain the trust of buyers and are more likely to be traded at a higher price than vehicles with service records from other repair shops. Fourth, vehicles that have undergone proper maintenance, such as timing belt replacement and brake-related servicing, have a reduced risk of serious breakdowns, thus lowering post-purchase maintenance costs and reducing the risk of breakdowns. These are some examples.

[0059] The training of the pre-trained model begins by extracting training data from the sales history database 4, including basic vehicle information, maintenance history information, and transaction prices. Next, for each training data point, a position coefficient is calculated for the transaction price within its market price range. This position coefficient is calculated by dividing the value obtained by subtracting the lower limit price from the transaction price by the value obtained by subtracting the lower limit price from the upper limit price, and is expressed as "Position Coefficient = (Transaction Price - Lower Limit Price) ÷ (Upper Limit Price - Lower Limit Price)". Furthermore, supervised learning is performed with basic vehicle information and maintenance history information as input and the position coefficient as output. Finally, the weight coefficients obtained as a result of the training are saved as the pre-trained model.

[0060] The trained model is stored in the memory of the information processing device 10 and is referenced when the price calculation unit 108 calculates the valuation price. The trained model is retrained and updated when new transaction data is accumulated. In the case of periodic updates, retraining is performed monthly or weekly, including the most recent transaction data. In the case of on-demand updates, retraining is performed when a predetermined number of new transaction data are accumulated or when the prediction accuracy falls below a predetermined threshold. In online learning, the weighting coefficients of the existing model are updated sequentially using the new transaction data.

[0061] The processing time for inference, i.e., the processing for calculating the valuation price, is targeted to be within 100 milliseconds per vehicle valuation. This will enable real-time responses to valuation requests from dealer terminal 2. If the processing time exceeds the target, the system will address this by reducing the model size or using dedicated inference hardware. Model reduction involves reducing the number of layers or quantization. Dedicated inference hardware includes GPUs and TPUs.

[0062] The maintenance history guarantee certificate generation unit 109 generates maintenance history guarantee certificate data that includes the maintenance evaluation score calculated by the maintenance implementation evaluation unit 104 and the critical maintenance evaluation unit 105, the maintenance implementation rate for recommended maintenance items, the implementation status of critical maintenance items, the breakdown of maintenance shops, and the maintenance evaluation rank.

[0063] The maintenance evaluation rank is determined based on the maintenance evaluation score. For example, a maintenance evaluation score of "90% or higher" results in rank A, "70% or higher but less than 90%" results in rank B, "50% or higher but less than 70%" results in rank C, and "less than 50%" results in rank D.

[0064] The maintenance history guarantee data is transmitted to the dealer terminal 2 and displayed on the screen or printed out as a maintenance history guarantee certificate. This "maintenance history guarantee certificate" is an electronic or paper document that certifies the maintenance history information of the target vehicle in a format that can be verified by a third party. The maintenance history guarantee certificate is generated by the information processing device 10 based on information obtained from the vehicle information database 3, maintenance history database 6, recommended maintenance database 7, and repair shop reliability database 8, and is a document that describes the results of an objective evaluation of the maintenance status of the target vehicle. The maintenance history guarantee certificate is used in vehicle sales transactions by the seller to prove the maintenance status of the vehicle to the buyer. The information described in the maintenance history guarantee certificate is calculated by the information processing device 10 based on information obtained from external databases, and differs from conventional maintenance records in that it is an evaluation result based on objective data rather than the seller's self-declaration.

[0065] Referring to Figure 2, the flow of the valuation price calculation process according to this embodiment will be explained. The vehicle information acquisition unit 101 receives the vehicle ID of the target vehicle from the dealer terminal 2 and acquires the vehicle information of the target vehicle from the vehicle information database 3 (S201). The vehicle information includes the vehicle type, year of manufacture, and mileage. The price calculation unit 108 calculates the base sales cost A using the information from the sales history database 4 and the market price information database 5. The market price information acquisition unit 107 also acquires the market transaction price range from the market price information database 5 (S202). The market transaction price range includes the upper and lower limits of the price. The maintenance history acquisition unit 103 acquires the maintenance history information of the target vehicle from the maintenance history database 6 (S203).

[0066] Next, the recommended maintenance extraction unit 102 extracts recommended maintenance items from the recommended maintenance database 7 according to the year of manufacture and mileage. The recommended maintenance extraction unit 102 determines whether the mileage is above a predetermined threshold and selects the corresponding recommended maintenance table (S204). The predetermined threshold is, for example, a mileage specified by the vehicle manufacturer, such as 100,000 kilometers.

[0067] The maintenance implementation evaluation unit 104 compares the extracted recommended maintenance items with the maintenance history information and evaluates the maintenance implementation status for each recommended maintenance item. The critical maintenance evaluation unit 105 identifies critical maintenance items and applies weighting coefficients. The maintenance factory reliability acquisition unit 106 acquires maintenance factory reliability information from the maintenance factory reliability database 8. The maintenance implementation evaluation unit 104 adjusts the score based on the maintenance factory reliability information and calculates the final maintenance evaluation score (S205).

[0068] The price calculation unit 108 calculates the updated cost of goods sold B based on the maintenance evaluation score (S206). The price calculation unit 108 uses either a table-based evaluation method or a pre-trained model evaluation method. The maintenance history warranty generation unit 109 generates maintenance history warranty data. The price calculation unit 108 transmits the updated sales cost B and maintenance history warranty data to the dealer terminal 2. The dealer terminal 2 displays the valuation price and maintenance history warranty on its screen (S207).

[0069] Next, we will explain the comparison process between recommended maintenance items and maintenance history information, referring to Figure 3. Figure 3 shows the maintenance status for each vehicle.

[0070] Figure 3 shows the maintenance status, implementation rate, and score for recommended maintenance items corresponding to the age of each vehicle. Maintenance status is categorized as maintenance performed at the recommended time (○), maintenance performed outside the recommended time (△), and no maintenance performed (occasion). Important maintenance items are marked with an identification mark (★). The implementation rate indicates the completion rate for recommended maintenance items, and the score is the maintenance evaluation score.

[0071] The maintenance implementation evaluation unit 104 counts the number of items that were maintained at the recommended time, the number of items that were maintained outside the recommended time, and the number of items that were not maintained for each vehicle, and calculates the maintenance implementation rate. The maintenance implementation rate is calculated by adding 0.5 times the number of items that were maintained outside the recommended time to the number of items that were maintained at the recommended time, dividing the result by the total number of recommended maintenance items, and multiplying by 100 percent.

[0072] The critical maintenance evaluation unit 105 performs separate scoring for critical maintenance items. The implementation status of critical maintenance items is weighted by a weighting factor applied to the maintenance evaluation score.

[0073] Next, we will explain examples of the service history warranty and price display screens, referring to Figure 4. Figure 4 is an example of a screen displayed on the dealer terminal 2.

[0074] A vehicle information display area 21 is provided at the top of the screen. The vehicle information display area 21 displays the vehicle name, grade, year of manufacture, mileage, and vehicle ID. A price information display area is provided in the center of the screen. The price information display area includes a base sales cost A display section 22, a maintenance evaluation score display section 23, and an updated sales cost B display section 24. In the example in Figure 4, the base sales cost A is displayed as "1,200,000 yen", the maintenance evaluation score is displayed as +0.08, or "8% added (+8%)", and the updated sales cost B is displayed as "1,296,000 yen". A maintenance history warranty certificate 25 is displayed on the right side of the screen.

[0075] The service history warranty certificate 25 includes the recommended maintenance completion rate, the status of important maintenance performed, the breakdown of maintenance facilities, and the maintenance evaluation rank. In the example in Figure 4, the recommended maintenance completion rate is listed as "90%", the status of important maintenance performed is listed as 5 out of 5 completed, the breakdown of maintenance facilities is listed as "3 times" at dealerships and "2 times" at certified workshops, and the maintenance evaluation rank is listed as "A".

[0076] At the bottom of the screen, the maintenance history list 26 is displayed. The maintenance history list 26 displays the date of each maintenance, the maintenance details, the type of maintenance shop, and the cost in a list format. Important maintenance items are marked to indicate their importance.

[0077] Next, we will explain examples of important maintenance items by vehicle type, referring to Figure 5. Figure 5 is a diagram showing important maintenance items, recommended repair shops, and types of parts used for a specific vehicle type, according to mileage categories. Mileage categories are divided into less than 100,000 kilometers and 100,000 kilometers or more. This 100,000 kilometers is just an example and can be set arbitrarily.

[0078] For example, for a WAAA model vehicle from manufacturer M, manufactured between 1990 and 2018, the following important maintenance items are associated with the vehicle at a mileage of less than 100,000 kilometers. Item 1 corresponds to periodic inspection of the suspension system, including tie rod ends and bushings. Item 2 corresponds to preventative replacement of the cooling system, including the water pump and thermostat. Item 3 corresponds to early detection and repair of oil leaks. Item 4 corresponds to rustproofing treatment of the undercarriage.

[0079] For these critical maintenance items, additional points will be awarded if there is a service history at a repair shop specializing in the vehicle in question. Genuine or OEM parts are recommended.

[0080] For vehicles with a mileage of 100,000 kilometers or more, in addition to the above, engine top-end overhaul, transmission overhaul, and suspension replacement are added as important maintenance items. The parts used may be genuine parts, OEM parts, or remanufactured parts.

[0081] Next, with reference to Figure 6, an example of the structure of the repair shop reliability table will be explained. The repair shop reliability table is a table that associates a reliability score of 10 levels, from "0 to 9", with the type of repair shop or individual repair shops.

[0082] A confidence score of "5" is set as the median, or baseline. If the confidence score exceeds the median, i.e., "6" or higher, the maintenance implementation evaluation unit 104 adds a predetermined percentage to the maintenance evaluation score. The predetermined percentage is "1% to 4%". If the confidence score is below the median, i.e., "4" or lower, the maintenance implementation evaluation unit 104 subtracts a predetermined percentage from the maintenance evaluation score. If the confidence score is the median, i.e., "5", no addition or subtraction is performed.

[0083] Let's explain the specific evaluation adjustment rates. For confidence points of "9", a plus "4%" adjustment is made. For confidence points of "8", a plus "3%" adjustment is made. For confidence points of "7" and "6", a positive adjustment is made as with the others. For confidence points of "5", a plus or minus "0%" adjustment is made (no adjustment is made). For confidence points of "4", a minus "1%" adjustment is made. For confidence points of "3", a minus "2%" adjustment is made. For confidence points of "2", "1", and "0", a negative adjustment is made as with the others.

[0084] Next, we will explain the structure and processing flow of the trained model with reference to Figure 7.

[0085] The trained model consists of a neural network comprising an input layer, an intermediate layer (or hidden layer), and an output layer. The input layer receives basic vehicle information and maintenance history information. Basic vehicle information includes categorical variables indicating the vehicle type, as well as numerical variables such as year of manufacture, mileage, and engine displacement. Maintenance history information includes the average maintenance evaluation score, critical maintenance completion rate, and repair shop reliability score. The output layer outputs a position coefficient within the market price range. The position coefficient is a value in the range of 0.0 to 1.0.

[0086] The training data used to train the pre-trained model consists of past vehicle sales history stored in the sales history database 4. Each record in the training data consists of a combination of input data and a correct label. The input data includes basic vehicle information and maintenance history-related information. The correct label is the position coefficient of the actual transaction price within the market price range.

[0087] The learning process is performed using a known learning algorithm such as backpropagation. Through learning, the weighting coefficients representing the correlation between the input data and the output data are optimized. The learned weighting coefficients are stored in the memory of the information processing device 10 as a learned model.

[0088] In the inference process, or the process for calculating the valuation price, the price calculation unit 108 inputs the vehicle's basic information and maintenance history-related information into the trained model. The trained model performs calculations based on the trained weighting coefficients and outputs a position coefficient. The price calculation unit 108 calculates the valuation price based on the position coefficient and the market price range, i.e., the upper and lower price limits.

[0089] In the above embodiment, an example was described in which the information processing device 10 is implemented as a server computer. However, the information processing device 10 may also be configured to be integrated with the sales terminal 2. In this case, each database is stored in an external storage device connected to the information processing device 10 via a communication network.

[0090] In the above embodiment, an example using a neural network as the pre-trained model was described, but the pre-trained model may be a gradient boosting decision tree (GBDT), a random forest, a support vector machine (SVM), or an ensemble model of these.

[0091] In the above embodiment, the calculation of the appraised value of a vehicle was described as an example, but the present invention is also applicable to the calculation of the appraised value of assets other than vehicles. Assets other than vehicles include, for example, real estate, machinery and equipment, ships, or aircraft. In this case, maintenance history information is replaced with maintenance and inspection history information of the asset.

[0092] As used in the embodiments and claims described above, the terms “part,” “means,” “apparatus,” and “system” do not merely refer to physical means, but also include cases where the functions of these are realized by software or software services.

[0093] Furthermore, the functions of a single "part," "means," "apparatus," or "system" may not only be realized by a single physical means, software, software module, or apparatus, but may also be realized by multiple physical means, software, software modules, apparatus, or combinations thereof.

[0094] The terms used in the embodiments and claims described above should be interpreted as non-limiting terms. For example, the term "includes" should be interpreted as "not limited to those described as including." The term "contains" should be interpreted as "not limited to those described as containing." The term "equips" should be interpreted as "not limited to those described as equipped." The term "possesses" should be interpreted as "not limited to those described as possessing." The term "complements" should be interpreted as "not limited to those described as possessing." [Explanation of symbols]

[0095] 1. Information Processing System 2. Retailer terminals 3. Vehicle Information Database 4. Sales History Database 5 Market Information Database 6. Maintenance History Database 7. Recommended Maintenance Database 8. Repair Shop Reliability Database 10 Information Processing Devices 101 Vehicle Information Acquisition Unit 102 Recommended maintenance extraction unit 103 Maintenance History Acquisition Department 104 Maintenance Implementation Evaluation Department 105 Critical Maintenance Evaluation Department 106 Maintenance Factory Reliability Acquisition Department 107 Market Information Acquisition Department 108 Price Calculation Section 109 Maintenance History Warranty Generation Department 21 Vehicle information display area 22 Base Sales Cost A Display Section 23. Maintenance evaluation score display section 24 Updated Sales Cost B Display Section 25. Service History Warranty Certificate 26. List of Maintenance History

Claims

1. A recommended maintenance extraction unit extracts recommended maintenance items from the recommended maintenance database that correspond to the vehicle's make, model year, and mileage. A maintenance history acquisition unit acquires maintenance history information from a maintenance history database that shows the maintenance history performed on the target vehicle, A maintenance implementation evaluation unit compares the extracted recommended maintenance items with the maintenance history information and calculates a maintenance evaluation score indicating the status of maintenance implementation for the recommended maintenance items. A repair shop reliability acquisition unit obtains repair shop reliability information, which indicates the reliability level set for each repair shop, from a repair shop reliability database. A market information acquisition unit that obtains market transaction price ranges corresponding to the combination of the aforementioned vehicle type, year of manufacture, and mileage from a market information database, A price calculation unit that calculates the assessed price of the target vehicle based on the maintenance evaluation score and the market transaction price range. It is equipped with, The maintenance implementation evaluation unit is an information processing device that adjusts the maintenance evaluation score based on the maintenance history information and the reliability information of the maintenance factory included in the maintenance history information.

2. The aforementioned repair shop reliability information is information that associates multiple levels of reliability points with each repair shop. The information processing device according to claim 1, wherein the maintenance implementation evaluation unit adds a predetermined percentage to the maintenance evaluation score when the reliability points exceed a predetermined standard value.

3. The aforementioned repair shop reliability information is information that associates multiple levels of reliability points with each repair shop. The information processing device according to claim 1, wherein the maintenance implementation evaluation unit subtracts a predetermined percentage from the maintenance evaluation score when the reliability points are below a predetermined standard value.

4. The information processing device according to claim 1, wherein the price calculation unit uses a trained model that has been trained using past vehicle sales history and maintenance history associated with the vehicle as training data, inputs vehicle information including the vehicle type, year of manufacture and mileage of the target vehicle and the maintenance evaluation score into the trained model, and calculates the evaluation price based on the information output from the trained model and the market transaction price range.

5. A method of information processing performed by a computer, The aforementioned computer, The recommended maintenance database includes a step to extract recommended maintenance items corresponding to the vehicle's make, model year, and mileage, and The steps include obtaining maintenance history information from the maintenance history database that shows the maintenance history performed on the target vehicle, The steps include: comparing the extracted recommended maintenance items with the maintenance history information and calculating a maintenance evaluation score that indicates the status of maintenance performed for the recommended maintenance items; The steps include obtaining repair shop reliability information, which indicates the reliability level set for each repair shop, from a repair shop reliability database, and The steps include obtaining a market transaction price range corresponding to the combination of the vehicle type, year of manufacture, and mileage from a market information database, A step of calculating the appraised value of the subject vehicle based on the aforementioned maintenance evaluation score and the aforementioned market transaction price range. Execute, The step of calculating the maintenance evaluation score includes adjusting the maintenance evaluation score based on the maintenance history information and the maintenance factory reliability information.

6. On the computer, A process to extract recommended maintenance items from the recommended maintenance database that correspond to the vehicle's make, model year, and mileage. A process to obtain maintenance history information from the maintenance history database, which shows the maintenance history performed on the aforementioned vehicle. A process to compare the extracted recommended maintenance items with the maintenance history information and calculate a maintenance evaluation score indicating the status of maintenance performed for the recommended maintenance items. A process to retrieve repair shop reliability information, which indicates the reliability level set for each repair shop, from a repair shop reliability database. A process to obtain market transaction price ranges corresponding to the combination of the aforementioned vehicle type, year of manufacture, and mileage from a market information database, and A process for calculating the valuation price of the target vehicle based on the maintenance evaluation score and the market transaction price range. Make it run, The process for calculating the maintenance evaluation score is a program that includes adjusting the maintenance evaluation score based on the maintenance history information and the reliability information of the maintenance factory.

7. A recommended maintenance database that stores recommended maintenance data, including recommended maintenance items corresponding to the vehicle's make, model year, and mileage, A maintenance history database that stores maintenance history information showing the maintenance history performed on a vehicle, A repair shop reliability database that stores repair shop reliability information indicating the reliability level set for each repair shop, A market price information database that stores market price information indicating the price range of vehicle transactions, An information processing device that is communicatively connected to the aforementioned recommended maintenance database, the maintenance history database, the maintenance factory reliability database, and the market price information database, Equipped with, The aforementioned information processing device is A recommended maintenance extraction unit extracts recommended maintenance items corresponding to the vehicle's make, model year, and mileage from the aforementioned recommended maintenance database. A maintenance history acquisition unit acquires maintenance history information from the maintenance history database that shows the maintenance history performed on the target vehicle, A maintenance implementation evaluation unit compares the extracted recommended maintenance items with the maintenance history information and calculates a maintenance evaluation score indicating the status of maintenance implementation for the recommended maintenance items. A repair shop reliability acquisition unit acquires the repair shop reliability information from the aforementioned repair shop reliability database, A market information acquisition unit that acquires market transaction price ranges corresponding to the combination of vehicle type, year of manufacture, and mileage from the market information database, A price calculation unit that calculates the assessed price of the target vehicle based on the maintenance evaluation score and the market transaction price range. It is equipped with, The maintenance implementation evaluation unit is an information processing system that adjusts the maintenance evaluation score based on the maintenance history information and the reliability information of the maintenance factory included in the maintenance history information.

Images