Information processing methods

Abstract

Improve the technique for selecting queries. [Solution] The method executed by the information processing device 10 includes registering a corresponding domain keyword and target phrase for each candidate for the search target, obtaining a query containing the domain keyword corresponding to the search target identified from the input question, executing a search process multiple times which includes the steps of inputting the query into a search engine, identifying and counting each snippet containing the target phrase corresponding to the search target on a screen displaying the search results, and adding keywords contained in one or more identified snippets to the query, obtaining an answer to the question based on the search results of the multiple executed search processes in which the count of snippets containing the target phrase was the highest, and outputting the obtained answer.

Description

Technical Field

[0001] This disclosure relates to an information processing method.

Background Art

[0002] Conventionally, technologies related to dialogue systems that provide answers to user questions are known. For example, Patent Document 1 discloses a technology for generating a dialogue bot specialized for a given domain by using a large language model based on documents in the domain.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] In the above document, a technology for constructing a large language model using a machine learning method such as an autoregressive model to generate a dialogue bot is disclosed. However, there is room for improvement in the selection of queries input to the search engine in order to output appropriate answers in the dialogue system.

[0005] In view of such circumstances, an object of the present disclosure is to improve the technology for selecting queries used to output appropriate answers.

Means for Solving the Problems

[0006] A method executed by an information processing apparatus according to an embodiment of the present disclosure includes: registering corresponding domain keywords and target phrases for each candidate for a search target; obtaining a query including the domain keyword corresponding to the search target specified from the input question; Performing a search process multiple times, which includes the steps of: entering the query into a search engine; identifying and counting each snippet containing the target phrase corresponding to the search target on a screen displaying the search results; and adding keywords contained in one or more identified snippets to the query. In the search process which is performed multiple times, the answer to the question is obtained based on the search result from the time in which the count of the snippet containing the target phrase was the highest, and Output the obtained response. Includes. [Effects of the Invention]

[0007] According to one embodiment of the present disclosure, the technique for selecting queries used to produce appropriate answers is improved. [Brief explanation of the drawing]

[0008] [Figure 1] This block diagram shows a schematic configuration of a system according to one embodiment of the present disclosure. [Figure 2] This is a flowchart showing the operation of an information processing device. [Modes for carrying out the invention]

[0009] The embodiments of this disclosure will be described below.

[0010] (Summary of the embodiment) Referring to Figure 1, an overview of System 1 according to the embodiment of this disclosure will be described. System 1 comprises an information processing device 10, a domain database (domain DB) 20, and a terminal device 30. System 1 is connected to an external server 40 via a network 50, including, for example, the Internet and a mobile communication network. System 1 constructs a dialogue system that outputs answers to questions such as "Please tell me the appraisal price of a used car" from a car dealer or the like.

[0011] The information processing device 10 is, for example, a computer such as a server. The domain DB 20 is a database that stores information related to the input and output to the search engine for each search target. The domain DB 20 may be located on a computer such as a server installed in a cloud environment or on-premises environment, or it may be located on the information processing device 10. The terminal device 30 is, for example, a mobile device such as a smartphone, mobile phone, wearable device, or tablet, a navigation device installed in a vehicle, a PC (personal computer), or a general-purpose or dedicated device, but is not limited to these.

[0012] Furthermore, the external server 40 shown in Figure 1 is a server of an entity that provides an LLM (Large-Scale Language Model) 41. The LLM 41 is a language model built by machine learning on large-scale data. The LLM 41 outputs answers to questions entered by the user. The external server 40 also includes a RAG (Search Enhancement Generation) 42. The RAG 42 provides the LLM with data that the LLM has not yet learned, including data specific to each entity or data searched in real time, and assists the LLM in creating answers. The RAG 42 may be configured as part of the information processing device 10.

[0013] First, an overview of this embodiment will be described, and details will be described later. The method executed by the information processing device involves registering a corresponding domain keyword and target phrase for each candidate search target, and obtaining a query containing the domain keyword corresponding to the search target identified from the input question. This method then executes a search process multiple times, which includes the steps of inputting the query into a search engine, identifying and counting each snippet containing the target phrase corresponding to the search target on the screen displaying the search results, and adding keywords contained in one or more identified snippets to the query. Furthermore, this method obtains an answer to the question based on the search results from the multiple search processes that were executed, specifically the one with the highest count of snippets containing the target phrase, and outputs the obtained answer.

[0014] Thus, according to this embodiment, queries are automatically modified and expanded so that the number of snippets containing the target phrase increases on the search results screen using the search engine. This reduces the burden on the user compared to, for example, a method of pre-registering multiple appropriate queries selected manually, while improving the accuracy of answers to questions. Therefore, the technology for selecting appropriate queries is improved in terms of outputting highly accurate answers and reducing human costs.

[0015] Next, we will describe each component of System 1 in detail.

[0016] (Configuration of information processing device) As shown in Figure 1, the information processing device 10 comprises a communication unit 11, a storage unit 12, and a control unit 13.

[0017] The communication unit 11 includes one or more communication interfaces connected to the domain DB 20 and the network 50, respectively. These communication interfaces may support, for example, mobile communication standards, wired LAN (Local Area Network) standards, or wireless LAN standards, but are not limited to these, and may support any communication standard. In this embodiment, the information processing device 10 communicates with the domain DB 20, terminal devices 30, and external servers 40 via the communication unit 11 and the network 50.

[0018] The storage unit 12 includes one or more memories. Each memory included in the storage unit 12 may function as, for example, a main memory, an auxiliary memory, or a cache memory. The storage unit 12 stores any information used in the operation of the information processing device 10. The storage unit 12 may store, for example, system programs and application programs. In this embodiment, the storage unit 12 stores application programs for a web browser and any search engine, and queries to be entered into the search engine. The information stored in the storage unit 12 may be updateable with information obtained from the network 50 via, for example, the communication unit 11.

[0019] The control unit 13 includes one or more processors, one or more programmable circuits, one or more dedicated circuits, or a combination thereof. The processor is, for example, a general-purpose processor such as a CPU (Central Processing Unit) or a GPU (Graphics Processing Unit), or a dedicated processor specialized for specific processing, but is not limited thereto. The programmable circuit is, for example, but not limited to, an FPGA (Field-Programmable Gate Array). The dedicated circuit is, for example, but not limited to, an ASIC (Application Specific Integrated Circuit). The control unit 13 controls the operation of the entire information processing apparatus 10.

[0020] (Configuration of the Domain DB) The domain DB 20 is a database that stores information related to the input / output to the search engine for each search target. The domain DB 20 stores, for example, a search target, a domain keyword, and a target phrase. Details of the data structure of the domain DB 20 will be described later.

[0021] (Configuration of the Terminal Device 30) As shown in FIG. 1, the terminal device 30 includes a communication unit 31, a storage unit 32, a control unit 33, an output unit 34, and an input unit 35. The configurations of the communication unit 31, the storage unit 32, and the control unit 33 are basically the same as the configurations of the communication unit 11, the storage unit 12, and the control unit 13 of the information processing apparatus 10, so the description will be simplified.

[0022] The communication unit 31 includes one or more communication interfaces connected to the network 50.

[0023] The storage unit 32 includes one or more memories. In the present embodiment, the storage unit 32 stores an application program of an interactive system provided by an operator using the information processing apparatus 10, an application program of a web browser, and the like.

[0024] The control unit 33 includes one or more processors, one or more programmable circuits, one or more dedicated circuits, or a combination thereof. The control unit 33 is capable of executing application programs stored in the storage unit 32.

[0025] The output unit 34 includes one or more output devices that output information. These output devices are, for example, a display that outputs information as video, or a speaker that outputs information as sound, but are not limited to these.

[0026] The input unit 35 includes one or more input devices that accept operations from an operator. These input devices are, but are not limited to, physical keys, capacitive keys, capacitive panels, touchscreens integrated with displays, or microphones that accept voice input.

[0027] (Operation flow of the information processing device) Referring to Figure 2, the operation of the information processing device 10 according to this embodiment will be described.

[0028] Step S100: The control unit 13 registers the corresponding domain keyword and target phrase for each candidate search target. Specifically, the search target, along with the domain keyword and target phrase for each search target, are input by the business operator via the input unit 35 of the terminal device 30, transmitted by the control unit 33 to the information processing device 10 via the communication unit 31, and stored in the domain DB 20 by the control unit 13 of the information processing device 10. Here, the data structure of the domain DB 20 will be explained with reference to Table 1. The search target is a part of the question sentence or a main word (e.g., "appraisal amount") identified from the question entered by the business operator (e.g., "Tell me the appraisal amount of a used car"). The domain keyword is one or more words (e.g., one or more words such as "used," "appraisal," and "amount") that are entered as a query into the search engine and are stored in association with the search target. The target phrase is one or more words of information that the business operator wants to obtain as a result of searching for the search target (e.g., a word representing the appraisal amount such as "yen" or "¥," or a combination of words representing the range of the appraisal amount such as "¥*~¥*").

[0029] [Table 1]

[0030] Step S101: The control unit 13 obtains a query containing domain keywords corresponding to the search target identified from the input question. Specifically, the control unit 13 receives a question (for example, "Tell me the appraisal price of a used car") entered by the business operator via the input unit 35 of the terminal device 30 from the terminal device 30 via the communication unit 11, and identifies the search target (for example, "appraisal price") from the received question. The control unit 13 obtains a query from the domain DB 20 containing domain keywords (for example, "used appraisal price") corresponding to the search target (for example, "appraisal price").

[0031] Step S102: The control unit 13 inputs the query into the search engine as part of the search process. Specifically, the control unit 13 inputs the query obtained in step S101 (for example, "used goods appraisal price") into the search engine. Here, the search engine may be any search engine such as Google (registered trademark).

[0032] Step S103: As part of the search process, the control unit 13 identifies and counts each snippet (hereinafter also referred to as a good snippet) that contains the target phrase corresponding to the search target on the screen displaying the search results (hereinafter also referred to as the results screen). Specifically, the control unit 13 obtains the results screen in step S102. The control unit 13 obtains the target phrase (e.g., "yen ￥") corresponding to the search target (e.g., "assessment amount") from the domain DB 20. The control unit 13 identifies good snippets by determining for each snippet whether one or more snippets on the results screen (for example, snippet A: "Easily understand the market price of used cars from the used car market price table! You can find out the market price of used cars by the combination of price, mileage, and year of manufacture.", snippet B: "The estimated purchase price as of MM month, YYYY is ¥1,000,000 to ¥1,500,000.") contains the target phrase (for example, snippet A does not contain either "yen" or "¥", so it does not qualify as a good snippet. Snippet B contains "¥", so it qualifies as a good snippet). The control unit 13 further counts the number of good snippets (for example, snippet B). Here, a snippet is text information that is displayed along with a link or title of a web page on the screen that displays the results of a search performed by entering a word or sentence into a search engine, and which excerpts and explains the content of the web page. In this embodiment, the results screen may be the first page of the screen that displays the search results. Furthermore, the number of snippets displayed on the results screen can be set arbitrarily.

[0033] Step S104: As part of the search process, the control unit 13 adds keywords (hereinafter also referred to as "good keywords") contained in one or more identified snippets to the query. Specifically, the control unit 13 extracts good keywords (e.g., "YYYY / MM / as of") from a good snippet (e.g., snippet B). The control unit 13 adds the good keywords to the query. As a result, for example, the query becomes "used goods appraisal amount YYYY / MM / as of". The phrase "YYYY / MM / as of" may be a specific number (e.g., October 2024) or a wildcard for fuzzy searching (e.g., *year*month). In an additional embodiment, good keywords may be keywords that appear frequently in snippets containing the target phrase, and keywords that appear only in snippets containing the target phrase, or both. Specifically, the control unit 13 counts each keyword in the good snippet and determines which keywords to add to the query based on the ranking of the counts (e.g., from highest-ranking keywords to lowest, or randomly). In a further additional embodiment, the control unit 13 may determine whether the good keyword is included in each snippet that does not contain the target phrase, and if it determines that the good keyword is not included, it may add the good keyword to the query.

[0034] Step S105: The control unit 13 executes the search process multiple times. Specifically, the control unit 13 determines whether it has executed the steps including steps S102 to S104 as the search process N times. If the control unit 13 determines that it has not executed the search process N times, it returns to step S102 (step S105-No) and repeats the search process. If the control unit 13 determines that it has executed the search process N times, it proceeds to step S106 (step S105-Yes). Here, N is any integer value of 1 or more, and may have a predetermined upper limit. In an additional embodiment, the control unit 13 may terminate the execution of the search process when it obtains search results in which the occurrence rate of snippets containing the target phrase is equal to or greater than a threshold. For example, the control unit 13 may count the snippets on the results screen and terminate the search process when the number of snippets is 10 and the number of good snippets in step S103 is 9.

[0035] Step S106: The control unit 13 obtains an answer to the question based on the search results of the search with the highest count of snippets containing the target phrase in the multiple search processes executed (hereinafter also referred to as the best search). Specifically, the control unit 13 determines the best search based on the number of good snippets counted in step S103 as a result of repeating the search process, and obtains the result screen for the best search. This result screen may be stored in the storage unit 12 in association with the query in step S103 and retrieved from there, or it may be obtained by re-entering the query used in the best search into the search engine. The control unit 13 sends the result screen of the best search along with the question entered by the business operator to the external server 40 in any data format. For example, the control unit 13 may create an instruction statement such as "Please tell me the appraisal amount of a used car by referring to the given information," add the information of the result screen of the best search to the instruction statement, and send it to the external server 40. The LLM41 on the external server 40 references the best result screen received as RAG42 and creates an answer to the question (for example, "The appraisal price of a used car is between ¥1,000,000 and ¥1,500,000"). The LLM41 sends the created answer to the information processing device 10. The control unit 13 of the information processing device 10 receives the answer from the external server 40.

[0036] Step S107: The control unit 13 outputs the acquired response. Specifically, the control unit 13 sends the response received from the external server 40 to the terminal device 30 via the communication unit 11. The control unit 33 of the terminal device 30 outputs the response to the business operator via the output unit 34.

[0037] In an additional embodiment, the search process may remove some of the domain keywords and added keywords (hereinafter referred to as extended keywords) from the query. Specifically, the control unit 13 removes some of the extended keywords (e.g., "used") from the extended keywords in the query (e.g., "used appraisal amount as of YYYY / MM") and generates a query that includes a new extended keyword ("appraisal amount as of YYYY / MM"). The addition to the query in step S104 and the removal from the query in this embodiment may be performed simultaneously or in any order. Furthermore, the control unit 13 can freely generate combinations of extended keywords by combining additions and deletions to the query. For example, the control unit 13 may determine combinations of extended keywords based on keyword ranking or randomly, using a history of past extended keywords. Alternatively, for example, a genetic algorithm may be used to determine multiple combinations of extended keywords. Multiple offspring genes (e.g., "used amount 10,000 yen", "appraisal market price", "used market price 10,000 yen", and "used market price") may be generated from multiple parent genes (e.g., "used appraisal amount" and "market price 10,000 yen"). This allows for more appropriate query selection.

[0038] In an additional embodiment, the control unit 13 updates the domain keyword in the query used in the instance with the highest count of snippets containing the target phrase. Specifically, the control unit 13 may overwrite the domain keyword in the domain DB 20 with the extended keyword in the query entered into the search engine in the best instance. For example, if "used appraisal amount as of YYYY / MM" is the extended keyword in the query used in the best instance, this extended keyword will be overwritten in the domain keyword associated with the search target ("appraisal amount") in the domain DB 20. This allows for the immediate retrieval of the appropriate query from the domain DB 20.

[0039] In an additional embodiment, the query may be configured to further include vehicle names. In this case, the domain DB20 may be configured for each vehicle name. In a further additional embodiment, the search processes from step S102 to step S105 may be executed in parallel for each vehicle name, and the query containing the extended keyword with the highest average occurrence rate of good snippets may be adopted as the best query across the search processes for each vehicle name. This allows for efficient selection of appropriate queries.

[0040] In an additional embodiment, the present disclosure (in particular step S100) may be performed by any user. That is, a user utilizing any interactive system built by System 1 can perform the present disclosure. The method can be used more generally.

[0041] While this disclosure has been described based on the drawings and embodiments, it should be noted that those skilled in the art may make various modifications and alterations based on this disclosure. Therefore, it should be noted that these modifications and alterations are within the scope of this disclosure. For example, the functions, etc., included in each component or step can be rearranged in a logically consistent manner, and multiple components or steps can be combined into one or divided into two.

[0042] For example, in the embodiment described above, it is also possible to distribute the configuration and operation of the information processing device 10 to multiple computers that can communicate with each other. Alternatively, for example, it is also possible to provide some or all of the components of the information processing device 10 in a terminal device 30. The number of terminal devices 30 provided in System 1 may be determined arbitrarily.

[0043] Furthermore, it is also possible to implement an embodiment in which a general-purpose computer functions as the information processing device 10 according to the above embodiment. Specifically, a program describing the processing content that realizes each function of the information processing device 10 according to the above embodiment is stored in the memory of the general-purpose computer, and the processor reads and executes the program. Therefore, this disclosure can also be implemented as a program that can be executed by a processor, or as a non-temporary computer-readable medium that stores the program. [Explanation of Symbols]

[0044] 1 System 10 Information Processing Devices 20. Domain Database (DB) 30 Terminal devices 40 External Servers 50 Networks

Claims

1. A method executed by an information processing device, For each potential search target, register the corresponding domain keyword and target phrase. To obtain a query containing the domain keyword corresponding to the search target identified from the input question, Performing a search process multiple times, which includes the steps of: entering the query into a search engine; identifying and counting each snippet containing the target phrase corresponding to the search target on a screen displaying the search results; and adding keywords included in one or more identified snippets to the query. In the search process which is performed multiple times, the answer to the question is obtained based on the search result from the time in which the count of the snippet containing the target phrase was the highest, and Output the obtained response. Methods that include...

2. In the method according to claim 1, A method further comprising updating the domain keyword in the query used in the instance where the snippet containing the target phrase had the highest count.

3. In the method according to claim 1, A method wherein the keyword added in the search process is either a keyword that appears frequently in the snippet containing the target phrase, or a keyword that appears only in the snippet containing the target phrase, or both.

4. In the method according to claim 1, The search process further includes the step of removing from the query some of the domain keywords and added keywords in the query.

5. In the method according to claim 1, A method further comprising terminating the execution of the search process when a search result is obtained in which the occurrence rate of the snippet containing the target phrase is equal to or greater than a threshold.

Images