Information processing apparatus

Abstract

Techniques for retrieving products conforming to specific specifications from a question text in a natural language are improved. An information processing device (20) is provided with a control unit (21) that registers, in a database, one or more attributes and attribute values, which are attribute values, for a plurality of vehicles, and performs control that, when a question text expressed in a natural language is input, specifies search conditions relating to the attributes indicated in the question text using machine learning, and performs search for the search conditions relating to the attributes indicated in the question text. And generating a query for selecting a vehicle satisfying the determined search condition from the database, and outputting a search result of the database based on the query.

Description

Technical Field

[0001] This disclosure relates to information processing devices. Background Technology

[0002] Patent document 1 discloses a technique for selecting documents containing answer content and outputting answers to user questions input into a retrieval model.

[0003] Existing technical documents

[0004] Patent documents

[0005] Patent Document 1: Japanese Patent Application Publication No. 2023-076413

[0006] There is room for improvement in the technology of retrieving products that meet specific specifications from natural language query text. Summary of the Invention

[0007] The problem that the invention aims to solve

[0008] The purpose of this disclosure is to improve the technology for retrieving products that conform to specific specifications from natural language query text.

[0009] Methods for solving problems

[0010] An embodiment of the information processing apparatus disclosed herein includes a control unit that performs the following controls:

[0011] For multiple vehicles, one or more attributes are combined with their values ​​(i.e., attribute values) and registered in the database.

[0012] When a natural language-based query text is input, machine learning is used to determine the search criteria related to the attribute shown in the query text.

[0013] Generate a query to select vehicles from the database that meet the determined search criteria.

[0014] Output the search results of the database obtained based on the query.

[0015] Invention Effects

[0016] According to this disclosure, an improved technique for retrieving products conforming to specific specifications from natural language query text is provided. Attached Figure Description

[0017] Figure 1 This is a block diagram illustrating the system configuration involved in the embodiments of this disclosure.

[0018] Figure 2 This is a flowchart illustrating the operation of an information processing apparatus according to an embodiment of the present disclosure. Detailed Implementation

[0019] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings.

[0020] The same or equivalent parts in the various figures are labeled with the same reference numerals. In the description of this embodiment, the description of the same or equivalent parts is appropriately omitted or simplified.

[0021] Reference Figure 1 The configuration of the system 10 involved in this embodiment is explained.

[0022] The system 10 has at least one information processing device 20 and at least one external device 30.

[0023] The information processing device 20 is able to communicate with external devices 30 via network 40.

[0024] Information processing device 20 is a computer such as a server that is installed in a data center or other equipment and belongs to a cloud computing system or other computing system.

[0025] External device 30 is a terminal device installed at a vehicle sales dealership or similar location. External device 30 may be, for example, a mobile phone, smartphone, tablet computer, or PC. "PC" is an abbreviation for personal computer.

[0026] Network 40 includes the Internet, at least one WAN, at least one MAN, or any combination thereof. "WAN" is an abbreviation for Wide Area Network. "MAN" is an abbreviation for Metropolitan Area Network.

[0027] Reference Figure 1 This document provides an overview of the implementation method.

[0028] By using a dialogue system employing information retrieval technology such as RAG (Retrieval-Augmented Generation), vehicle parameter tables can be retrieved, thereby answering queries about the specifications of a specific vehicle. Using this dialogue system, for example during a negotiation, it's possible to retrieve vehicle models that match the customer's desired conditions. However, when a salesperson, acting as a user, uses natural language to query the customer's desired conditions, sometimes the ambiguity of the query text prevents the correct generation of search criteria. For example, when wanting to use specific values ​​for parameters such as "vehicle height," "tire width," and "track width" as search criteria, the units of these values ​​are sometimes omitted in the actual dialogue. As a result, the search criteria cannot be correctly determined.

[0029] In this embodiment, the information processing device 20 registers one or more attributes and their values ​​(i.e., attribute values) for multiple vehicles in a database. When a natural language-based query text is input, the information processing device 20 uses machine learning to determine the attribute-related search criteria shown in the query text. The machine learning can employ any model; for example, any natural language processing technique selected individually or in combination, such as large-scale language models, NLP, intent recognition models, NER, and semantic retrieval, can be used. "NLP" is an abbreviation for Natural Language Processing. "NER" is an abbreviation for Named Entity Recognition. The information processing device 20 generates a query from the database that selects vehicles matching the determined search criteria. The information processing device 20 outputs the search results from the database based on the query.

[0030] According to this embodiment, when a natural language-based query text is input, the information processing apparatus 20 uses machine learning to determine the attribute-related search conditions shown in the query text. For example, if the attribute determined from the query text is "vehicle height," and the reference value (the baseline value referenced for retrieving attribute values) used as the determined search condition related to "vehicle height" is "1.55," this reference value is corrected to "1.55m." Therefore, even if numerical units contained in the query text are omitted, or if the query text contains ambiguous expressions, the search conditions can be easily and correctly determined. This improves the technology for retrieving products conforming to specific specifications from natural language query text.

[0031] Reference Figure 1 This section explains the configuration of the information processing apparatus 20 involved in this embodiment.

[0032] The information processing device 20 includes a control unit 21, a storage unit 22, and a communication unit 23. The control unit 21 includes at least one processor, at least one programmable circuit, at least one dedicated circuit, or any combination thereof, and performs prescribed processing.

[0033] The storage unit 22 includes at least one semiconductor memory, at least one magnetic memory, at least one optical memory, or a combination of at least two of these. The storage unit 22 stores data used in the operation of the information processing device 20, as well as data obtained through the operation of the information processing device 20. In this embodiment, the storage unit 22 stores a database DB. The database DB will be explained later.

[0034] The communication unit 23 includes at least one communication interface. The communication unit 23 receives data used in the operation of the information processing device 20 and transmits data obtained through the operation of the information processing device 20. In this embodiment, the communication unit 23 communicates with the external device 30.

[0035] The functions of the information processing apparatus 20 are realized by executing the program according to this embodiment using the processor, which serves as the control unit 21. That is, the functions of the information processing apparatus 20 are implemented through software. The program causes the computer to perform the actions of the information processing apparatus 20, thereby enabling the computer to function as the information processing apparatus 20. In other words, the computer functions as the information processing apparatus 20 by executing the actions of the information processing apparatus 20 according to the program.

[0036] Reference Figure 2 This section describes the operation of the information processing apparatus 20 according to this embodiment. This operation corresponds to the information processing method according to this embodiment. That is, the information processing method according to this embodiment includes... Figure 2 The steps S1 to S7 are shown below. Hereinafter, S and numbers will be used to denote each step in the flowchart.

[0037] In S1, the control unit 21 of the information processing device 20 registers one or more attributes and their values ​​(i.e., attribute values) to the database DB for multiple vehicles. In this embodiment, the one or more attributes include parameters of the motor vehicle. The control unit 21 can read parameter tables for multiple vehicles or register the parameters of each motor vehicle by referring to publicly available official information. For example, the control unit 21 registers attributes such as "vehicle height," "tire width," and "wheelbase" along with their attribute values ​​to the database DB. The control unit 21 can also calculate attributes and their values ​​that are not in the parameter table and register them to the database DB. For example, one of the important constraints for multi-level parking garages is "tire width." However, "tire width" is generally not recorded in the parameter table of the vehicle catalog. "Tire width" is the outer width of the left and right tires of the vehicle. In this case, the control unit 21 can also calculate "tire width" based on "wheelbase" and "tire width (front / rear)" in the parameter table. "Tire width" can be calculated as the maximum value of wheelbase + front / rear tire width.

[0038] In S2, the control unit 21 of the information processing device 20 receives a question text based on natural language. In this embodiment, the question text is input to the external device 30. The question text can be input via voice or text. For example, the question text can be voice recordings of a negotiation recorded via a microphone or other input unit of the external device 30, or text input by a salesperson during a negotiation via a keyboard or touchscreen, which is also an input unit of the external device 30. The external device 30 sends the input question text to the information processing device 20. The control unit 21 of the information processing device 20 receives the question text sent from the external device 30 via the communication unit 23. The control unit 21 analyzes the received question text. In the analysis of the question text, any method can be used, such as morpheme analysis, syntactic analysis, or semantic analysis.

[0039] In step S3, the control unit 21 of the information processing device 20 uses machine learning to determine the search criteria shown in the query text. Specifically, the control unit 21 extracts one or more keywords used to define parameters of the vehicle as search criteria. The control unit 21 extracts vehicle attributes, reference values ​​corresponding to the attributes, and comparison conditions compared with the reference values ​​as one or more keywords. Keywords extracted as "vehicle attributes" may include, for example, "vehicle height," "tire width," and "wheelbase." Keywords extracted as "reference values" are any numerical values. Keywords extracted as "comparison conditions" may include, for example, "equal to," "not equal to," "above," "below," and "lower than."

[0040] In S4, the control unit 21 of the information processing device 20 adjusts the reference value according to the attribute corresponding to the extracted reference value. Specifically, the control unit 21 determines whether the extracted reference value has a unit. If it determines that there is no unit, it infers the unit of the reference value based on the number of digits in the attribute and the reference value. For example, if the attribute is "vehicle height", the control unit 21 infers that the unit is "mm" when the reference value has 4 digits and "m" when the reference value has 1 digit. Here, the number of digits does not include the number of digits after the decimal point. As an example, suppose the question text "Please inform me of vehicles with a vehicle height of 1.55 or less" is entered in S2. Here, "example" is not used to limit this disclosure, but to help understand this embodiment. In this case, in S3, the control unit 21 of the information processing device 20 uses keywords such as "vehicle height", "1.55", and "less" as search conditions to extract vehicle attributes, reference values ​​corresponding to the attributes, and comparison conditions for comparison with the reference values. Here, "1.55" is extracted as the reference value corresponding to the attribute "vehicle height". In this case, the control unit 21 discards the decimal places from the reference value "1.55", determines that it has 1 digit, and infers that the unit of the reference value "1.55" is "m". Alternatively, suppose the input in S2 is "Please inform me of vehicles with a height of 1550 mm or less". In this case, the control unit 21 of the information processing device 20 uses keywords such as "vehicle height", "1550", and "less" as search conditions in S3, extracting vehicle attributes, reference values ​​corresponding to the attributes, and comparison conditions for comparing with the reference values. Here, "1550" is extracted as the reference value corresponding to the attribute "vehicle height". In this case, the control unit 21 determines that the reference value has 4 digits, and infers that the unit of the reference value "1550" is "mm".

[0041] As a variation of this embodiment, the control unit 21 of the information processing device 20 may also convert the unit of the reference value to the unit registered in the database DB. Specifically, assuming that the attribute value for "vehicle height" is registered in the database DB in meters, while the reference value deduced in S4 is in mm, the control unit 21 may also convert the reference value to meters. For example, the control unit 21 may convert the reference value "1550mm" to "1.55m".

[0042] In step S5, the control unit 21 of the information processing device 20 generates a query to select vehicles that meet the determined search criteria from the database DB. Specifically, the control unit 21 generates an SQL query for the database DB based on the search criteria determined in step S3 and the adjusted reference value in step S4. "SQL" is an abbreviation for Structured Query Language. For example, in step S3, keywords such as "vehicle height" (vehicle attribute), "1.55" (reference value), and "below" (comparison condition with the reference value) are extracted as search criteria. In step S4, the reference value "1.55" is adjusted to "1.55m". In this case, the control unit 21 generates a query indicating "select vehicles with a height of 1.55m or less from the database DB".

[0043] In step S6, the control unit 21 of the information processing device 20 retrieves data from the database DB using the query generated in step S5. Specifically, the control unit 21 executes the query generated in step S5 on the database DB to obtain data D1 of vehicles that match the query. Data D1 contains information showing the model of one or more vehicles that match the search criteria. Data D1 may also contain information showing the year, price, mileage, etc., of each vehicle.

[0044] In S7, the control unit 21 of the information processing device 20 outputs the search results from the database DB obtained from the query. Specifically, the control unit 21 controls the display of the data D1 obtained in S6 on the external device 30. The data D1 can be displayed in any order, for example, in the following order: The control unit 21 obtains the dataset consistent with the query from the database DB as data D1. The control unit 21 sends the data D1 to the external device 30 via the communication unit 23. The external device 30 receives the data D1 sent from the information processing device 20 and displays the search results shown by the data D1 on a display or the like, which serves as an output unit. It should be noted that the control unit 21 can also input the query text received in S2 and the data D1 obtained in S5 into a large-scale language model to obtain a natural language response text and output the response text.

[0045] According to this embodiment, when querying for products that meet specific specifications using natural language query text, even if the unit of the numerical value used as a reference in the query text is unclear, a correct search condition can be automatically generated. This improves the technology for retrieving products that meet specific specifications from natural language query text.

[0046] It should be noted that in this embodiment, the external device 30 is described as being located outside the information processing device 20, and the information processing device 20 receives and sends various data between the external device 30 and the information processing device 20 via the network 40 while processing the data. However, it is also possible that some or all of the functions of the information processing device 20 are carried on the external device 30.

[0047] In this embodiment, when the search criteria are determined in S3, if the unit of the reference value is clear, the control unit 21 of the information processing device 20 can omit the processing in S4. Alternatively, when the search criteria are determined in S3, if the unit of the reference value is clear, the control unit 21 can also refer to the database DB in S4 to confirm the registered values ​​and / or units for the attribute corresponding to the reference value, thereby correcting the value or unit of the reference value contained in the query text. For example, if a user makes a slip of the tongue and says "1.55 centimeters" instead of "1.55 meters" for the reference value of the attribute "vehicle height," the control unit 21 can detect unnatural unit changes by referring to the database DB to confirm the registered values ​​for "vehicle height." According to this configuration, numerical unit errors caused by slips of the tongue or mis-input can be corrected. The control unit 21 can also control the display of a prompt seeking confirmation from the user when a numerical unit error is detected. For example, the control unit 21 can also control the external device 30 to display a message saying "It's not 1.55 centimeters, but 1.55 meters?"

[0048] This disclosure is not limited to the embodiments described above. For example, the multiple blocks shown in the block diagram can be integrated, or a single block can be divided. Instead of executing the multiple steps shown in the flowchart in chronological order as described, the steps can be executed in parallel or in a different order, depending on the processing capability of the device executing each step or as needed. Furthermore, modifications can be made without departing from the spirit of this disclosure.

[0049] Explanation of reference numerals in the attached figures

[0050] 10 system

[0051] 20 Information Processing Devices

[0052] 21 Control Department

[0053] 22 Storage Department

[0054] 23 Ministry of Communications

[0055] 30 external devices

[0056] 40 networks.

Claims

1. An information processing device, It has a control unit that performs the following controls: For multiple vehicles, one or more attributes are combined with their values ​​(i.e., attribute values) and registered in the database. When a natural language-based query text is input, machine learning is used to determine the search criteria related to the attribute shown in the query text. Generate a query to select vehicles from the database that meet the determined search criteria. Output the search results of the database obtained based on the query.

2. The information processing apparatus according to claim 1, wherein, The one or more attributes mentioned include parameters of the motor vehicle. The control unit extracts one or more keywords used to define the parameters as the search criteria.

3. The information processing apparatus according to claim 2, wherein, The control unit extracts vehicle attributes, reference values ​​corresponding to the attributes, and comparison conditions for comparing with the reference values ​​as one or more keywords. The control unit adjusts the reference value according to the attribute corresponding to the extracted reference value.

4. The information processing apparatus according to claim 3, wherein, The control unit determines whether the extracted reference value has a unit. If it determines that there is no unit, it infers the unit of the reference value based on the attribute and the number of digits in the reference value.

5. The information processing apparatus according to claim 3 or 4, wherein, The control unit converts the unit of the reference value to the unit registered in the database.

Images