Data processing device, method, and program

Abstract

A data processing device according to one embodiment of the present invention includes: an input unit that receives input relating to the state of a service at the time of use thereof from a user using the service; a supplementation unit whereby, when the state at the time of use of the service includes an expression requiring supplementation, a state at the time of use of the service in which the expression requiring supplementation is supplemented is output using general knowledge that can be acquired from the exterior; and a calculation unit that uses the state at the time of use of the service after supplementation by the supplementation unit as a basis to calculate a requirement relating to provision of the service to the user.

Description

Data Processing Apparatus, Method, and Program

[0001] Embodiments of the present invention relate to a data processing apparatus, method, and program.

[0002] In the field of service automatic control of network services such as Autonomous NW (a network with autonomy) or application services such as SaaS (Software as a Service) (hereinafter sometimes collectively referred to as services), there is a technology for extracting a quantitative Intent, which is a service usage requirement by a user (sometimes referred to as a customer), for example, an Intent that satisfies the user's intention or purpose, in order to provide a service that meets the user's service usage requirements (for example, Patent Document 1, Non-Patent Documents 1 and 2).

[0003] International Publication No. 2023 / 157304

[0004] Hiroaki Kikushima, Nobuhiro Fukuda, Chao Wu, Shinobu Horiuchi, Kenichi Tayama, "Quantitative Intent Derivation Method Based on User's Service Usage Requirements", Proceedings of the 2022 National Convention of the Institute of Electronics, Information and Communication Engineers, B- (14-8), Mar.2022.Y. Xiao, W. Quan, H. Zhou, M. Liu, and K. Liu, "Lightweight Natural Language Driven Intent Translation Mechanism for Intent Based Networking," 2022 7th International Conference on Computer and Communication Systems (ICCCS)

[0005] However, currently, it is necessary to store a dataset used for quantitative intent extraction in a storage device and prepare it in advance. If the dataset does not contain data that satisfies the service usage requirements entered by the user, quantitative intent extraction becomes impossible. This invention was made in view of the above circumstances, and its purpose is to provide a data processing device, method, and program that enable the provision of appropriate services based on the service usage requirements of the user.

[0006] A data processing device according to one aspect of the present invention includes: an input unit that receives input from a user of the service regarding the circumstances of the service usage; a completion unit that outputs the circumstances of the service usage in which the expressions requiring completion have been completed using general knowledge obtainable from an external source when the circumstances of the service usage include expressions requiring completion; and a calculation unit that calculates requirements for providing the service to the user based on the circumstances of the service usage completed by the completion unit.

[0007] A data processing method according to one aspect of the present invention is a method performed by a data processing device, comprising: an input unit of the data processing device receiving input from a user of the service regarding the circumstances of the service usage; a supplementation unit of the data processing device outputting the circumstances of the service usage in which, when the circumstances of the service usage include expressions that require supplementation, the expressions requiring supplementation have been supplemented using general knowledge obtainable from an external source; and a calculation unit of the data processing device calculating requirements for providing the service to the user based on the circumstances of the service usage that have been supplemented by the supplementation unit.

[0008] According to the present invention, it is possible to provide appropriate network services based on the user's service usage requirements.

[0009] Figure 1 is a diagram showing an example of the application of a data processing device according to one embodiment of the present invention. Figure 2 is a diagram illustrating an example of the processing procedure by the data processing device. Figure 3 is a diagram showing an example of a conversational sentence including user input. Figure 4 is a diagram showing an example of an entity extracted from user input. Figure 5 is a diagram showing an example of an entity extracted from user input. Figure 6 is a block diagram showing an example of the hardware configuration of a data processing device according to one embodiment of the present invention.

[0010] Embodiments relating to this invention will be described below. Figure 1 is a diagram showing an example of application of a data processing device according to one embodiment of the present invention. As shown in Figure 1, the data processing device 100 according to one embodiment of the present invention has a service usage status input unit 10, an Intent acquisition unit 20, an information supplementation unit 30, and a quantitative Intent calculation unit 40. The information supplementation unit 30 has an information acquisition determination unit 31 and a general requirements information acquisition unit 32.

[0011] The service usage status input unit 10 accepts natural language input indicating the service usage status related to a service through operations by the user on an input / output interface (not shown). The service usage status may be referred to as the service usage requirements. Examples of services include network services and application services.

[0012] The Intent Acquisition Unit 20 extracts entities from the natural language text shown in the input results from the Service Usage Status Input Unit 10 using LLM (Large Language Models). Based on the results of this extraction, it acquires information related to the user's request, such as the service used, the equipment used, and the purpose of use. After receiving additional information from the Information Supplementation Unit 30 as needed, it passes the information necessary for quantifying the Intent to the Quantitative Intent Calculation Unit 40.

[0013] The information supplementation unit 30 supplements ambiguous information included in the user's request that the user themselves does not fully understand. The information acquisition determination unit 31 of the information supplementation unit 30 determines whether the natural language shown in the input result from the service usage status input unit 10 contains ambiguous expressions, such as expressions with a relatively high level of abstraction, such as "general" or "recommended".

[0014] Based on this determination, if there is an ambiguous expression in the natural language text described above, the information acquisition determination unit 31 determines whether or not information that complements this ambiguous expression exists in the knowledge graph stored in the internal memory (not shown) of the information complementation unit 30. If there is no information in the knowledge graph that complements the ambiguous expression described above, the information acquisition determination unit 31 determines whether or not it is necessary to acquire information from an external source to use in calculating the quantitative Intent, as there is no information that can be used to calculate the quantitative Intent according to the input result from the service usage status input unit 10.

[0015] When the Information Acquisition Determination Unit 31 determines that there is no information in the Knowledge Graph to complement the ambiguous expressions mentioned above, and that the information used to calculate the quantitative Intent needs to be acquired from external information consisting of a large amount of general information, the General Requirements Information Acquisition Unit 32 acquires general requirements information, which is general information related to the service usage requirements shown in the input results from the Service Usage Status Input Unit 10, from external general knowledge, such as that stored as LLM. The Intent Acquisition Unit 20 uses this acquired information to output to the Quantitative Intent Calculation Unit 40 information that embodies the user's requests included in the input results from the Service Usage Status Input Unit 10 as information necessary for quantifying the Intent.

[0016] The quantitative intent calculation unit 40 calculates a quantitative intent based on the information output from the intent acquisition unit 20 and passes this calculation result to an external automatic control system. This calculated quantitative intent is a requirement related to the provision of services to the user.

[0017] This automatic control system can perform service-related control, such as controlling the service content and resources of network services, based on the quantitative intent.

[0018] The data processing device 100 according to this embodiment uses LLM to acquire information from general knowledge in order to supplement information for ambiguous requests that the user themselves does not fully understand, and based on this acquired information, it is possible to calculate a quantitative intent.

[0019] In this embodiment, when a user's input includes ambiguous non-technical information, such as the services to be used, location, and purpose, the system presents the user with candidate information to clarify the ambiguous request based on general service usage requirement information stored as LLM (Limited Liability Management). The user can then select the appropriate candidate based on this presentation, thereby clarifying the ambiguous request.

[0020] Whether a user's input contains ambiguous expressions can be determined based on the extraction of entities from natural language text. While information to clarify ambiguous requests could be recorded in advance as the knowledge graph mentioned above, it is desirable that general service usage requirements information be collected in real time by LLM to prevent the knowledge graph from becoming bloated.

[0021] Figure 2 illustrates an example of the processing procedure by the data processing device. First, the service usage status input unit 10 receives natural language input including the service usage status (S11). Next, the Intent acquisition unit 20 acquires information from the natural language input in S11 that indicates the user's request regarding the use of the service, such as the type of service to be used, the equipment used for using the service, the type of communication line, and the purpose of using the service (S12).

[0022] The information acquisition determination unit 31 of the information supplementation unit 30 determines whether the user request acquired in S12 contains ambiguous expressions (S13). If the user request contains ambiguous expressions (Yes in S13), the information acquisition determination unit 31 determines whether information that clarifies these ambiguous expressions exists in the knowledge graph described above (S14).

[0023] If the information acquisition determination unit 31 determines that there is no information in the knowledge graph that clarifies the ambiguous expression (No. in S14), the general requirements information acquisition unit 32 uses the LLM (Large-Scale Language Model) to acquire general service usage requirements information related to the service usage requirements indicated by the input result from the service usage status input unit 10 from general external knowledge (S15).

[0024] Based on the results obtained in S15, the Intent acquisition unit 20 outputs the results of the user's requests, which are included in the input results from the service usage status input unit 10, to the quantitative Intent calculation unit 40 as information necessary for quantifying the Intent. On the other hand, if the answer in S13 is "No" or in S14 is "Yes", the Intent acquisition unit 20 passes the information related to the user's requests, extracted from the input results from the service usage status input unit 10, to the quantitative Intent calculation unit 40 as information necessary for quantifying the Intent, without going through processing by the information completion unit 30.

[0025] Here, it is assumed that the following (1-1) to (1-4) regarding the use of network services are passed from the Intent acquisition unit 20 to the quantitative Intent calculation unit 40 as information necessary for quantifying the Intent. (1-1) Service used: Web conferencing (1-2) Purpose of use: Private (1-3) Usage environment (device): Smartphone (1-4) Usage environment (line): Mobile line (4G)

[0026] Next, the quantitative intent calculation unit 40 uses the information necessary for quantifying the intent from the intent acquisition unit 20 to identify the service category to which the name of the service being used belongs (S16). Here, it is assumed that the service category to which the web conference belongs, "video and audio two-way distribution service," has been identified.

[0027] Next, the quantitative intent calculation unit 40 uses the information necessary for quantifying the intent from the intent acquisition unit 20 to identify quality indicators of the service provided to the user in the service category "video and audio two-way distribution service" identified in S16 (S17).

[0028] Here, the following (2-1) to (2-4) are identified as quality indicators for the above-specified service category, "Two-way video and audio distribution service." (2-1) Quality indicator (QoE) (2-2) Stability indicator (Jitter) (2-3) Stability indicator (Packet loss) (2-4) Latency indicator (RTT)

[0029] Next, the quantitative Intent calculation unit 40 uses the information necessary for quantifying the Intent from the Intent acquisition unit 20 to identify the required value range for each quality indicator, which is the range of values ​​for the quality indicators identified in S17 (S21). This required value range for each indicator may include the identified service category, the identification information of this category, the number of indicators, the indicator name, the identification information (ID (Identifier)) of the indicator name, the required value range, and a vector v. This vector shows the relationship between the magnitude of the identified quality indicator value and the quality. For example, if the vector for a certain quality indicator value is "1", it means that the larger the value of that quality indicator, the better the quality, and if the vector is "-1", it means that the larger the value of that quality indicator, the worse the quality.

[0030] Here, the following (3-1) to (3-4) are identified as the setting ranges for the quality indicators in the specified service category, "Video and Audio Two-Way Distribution Service": (3-1) Quality Index (QoE) = 1.0 to 5.0, v = 1 (3-2) Stability Index (Jitter) = 10 to 40 [ms], v = -1 (3-3) Stability Index (Packet Loss) = 0.5 to 1 [%], v = -1 (3-4) Latency Index (RTT) = 200 to 400 [ms], v = -1

[0031] Next, as processing related to the service-independent area, the quantitative Intent calculation unit 40 uses the information necessary for quantifying the Intent from the Intent acquisition unit 20 to identify the environment-specific requirement indicators, which are related indicators for each usage environment that were input in S11 and are related to the quality indicators identified in S17 (S22).

[0032] This environment-specific requirements indicator may include the name of the environment, identification information for this name, the name of the requirements indicator, identification information for this name, the degree of impact on higher-level identified quality indicators, and a cost flag. This cost flag indicates whether or not the cost in the environment specified by the environment-specific requirements indicator has an impact on the value of the identified quality indicators.

[0033] For example, if the cost flag for a particular usage environment is "1", it means that the costs incurred in that usage environment have an impact on the quality indicator value, while if the cost flag is "0", it means that the costs do not have an impact on the quality indicator value.

[0034] Here, the following (4-1) is identified as the usage environment-specific requirement indicator, which is the relevant indicator for the input usage environment (terminal), and the following (4-2) is identified as the usage environment-specific requirement indicator, which is the relevant indicator for the input usage environment (line). (4-1) Usage terminal (terminal): QoE, jitter (4-2) Usage environment (line): packet loss, RTT

[0035] Next, as processing related to the service-independent area, the quantitative Intent calculation unit 40 uses the information necessary for quantifying the Intent from the Intent acquisition unit 20 to identify the purpose-specific cost impact, which is a related indicator for each purpose of use entered in S11 and is related to the quality indicator identified in S17 (S23).

[0036] This purpose-specific cost impact may include the name of the purpose of use, identification information for this name, the name of the required indicator, identification information for this name, and a cost cap (cap). This cost cap indicates the limit, or upper limit, on the value of the quality indicator identified above, due to the costs incurred in the purpose of use in the purpose-specific cost impact. Here, the purpose-specific cost impact for "Purpose of Use: Private" is identified as "Cost Cap: QoE (Maximum 60%)".

[0037] Next, as processing related to the service-independent area, the quantitative intent calculation unit 40 calculates a cost-inducing requirement adjustment value based on the usage environment-specific requirement indicators identified in S22 and the purpose-specific cost impact identified in S23 (S24).

[0038] Here, based on the purpose-specific cost impact identified in S23, the following (5-1) is calculated as a cost-driven requirement adjustment value, and based on the usage environment-specific requirement indicator identified in S22, the following (5-2) is calculated as a cost-driven requirement adjustment value. (5-1) Purpose of use: Private = Cost cap: QoE (maximum 60[%]) (5-2) Usage environment (line): Mobile line (4G) = Cost flag "1"

[0039] Next, as processing related to the service-independent area, the quantitative intent calculation unit 40 calculates the environment-specific requirement value impact (S25) based on the environment-specific requirement indicators identified in S22, the purpose-specific cost impact identified in S23, and the cost-inducing requirement adjustment value calculated in S24.

[0040] Here, as the environmental requirement value influence degree, the following (6-1) and (6-2) are calculated. (6-1) Usage environment (terminal): Smartphone → QoE: 65 [%], Jitter: 30 [%] (6-2) Usage environment (network): Mobile network (4G) → Packet loss: 0 [%], RTT: 50 [%]

[0041] Next, the quantitative Intent calculation unit 40 multiplies the environmental requirement value influence degree calculated in S25 by the requirement value range for each index specified in S21 for each index to calculate each required index value (S26), outputs this calculation result to the above automatic control system, and notifies the user of this calculation result using a display device not shown in the figure (S27).

[0042] Here, as each required index value, the following (7-1) to (7-4) are calculated. (7-1) Quality index (QoE): 3.0 or more (7-2) Stability index (Jitter): 20 [ms] or less (7-3) Stability index (Packet loss): 1.0 [%] or less (7-5) Delay index (RTT): 300 [ms] or less

[0043] FIG. 3 is a diagram showing an example of a conversation sentence including an input sentence by a user. In FIG. 3, an example of a conversation sentence including an input sentence by a user using the service usage status input unit 10 during the execution of an automatic conversation program that automatically generates a natural language conversation sentence with the user is shown.

[0044] The messages related to "(A1)", "(A2)", "(A3)" and "(A4)" in FIG. 3 are conversation sentences automatically generated regarding the use of the service. Also, the messages related to "(B1)", "(B2)", "(B3)" and "(B4)" in FIG. 3 are input sentences indicating the user's requests regarding the use of the service.

[0045] In FIG. 3, the message related to “(B3)” includes “<Services used in general exhibitions...>”, and this message includes the ambiguous expression “general”. The message related to “(A4)” automatically generated according to this message includes “<For the application service of the service used in general exhibitions> <Data viewing 80%, SNS mail 15%, video viewing 5%>, design resources accordingly.”

[0046] The message related to this “(A4)” is a message that complements the ambiguous expression included in the message related to “(B3)”, and is a message generated using the information acquired by the general requirement information acquisition unit 32 of the information complementation unit 30. That is, the messages related to “(B3)” and “(A4)” (reference symbol a in FIG. 3) refer to the conversation sentence when the ambiguous expression in the input sentence by the user is complemented.

[0047] FIGS. 4 and 5 are diagrams showing an example of an entity extracted by an LLM from an input sentence by a user. In FIGS. 4 and 5, examples of entities extracted by the Intent acquisition unit 20 for the input sentence by the user to the service usage status input unit 10 are shown.

[0048] In FIG. 4, an example is shown in which the entity “ABC Research Institute” related to “location” and the entity “remote operation of the robot” related to “service” are extracted from the input sentence “I want to conduct a demonstration of remote operation by a robot within the ABC Research Institute.” (reference symbol a in FIG. 4) by the user.

[0049] In FIG. 5, an example is shown in which the entity “exhibition” related to “location” and the entity “general service” (reference symbol b in FIG. 5) related to “service” are extracted from the input sentence “I want to provide services used in general exhibitions to visitors throughout the research and development center.” (reference symbol a in FIG. 5) by the user.

[0050] Next, an example of the acquisition result of general requirements information by the general requirements information acquisition unit 32 of the information supplementation unit 30 will be explained. Here, the message of the general requirements information acquisition result is generated using LLM as follows.

[0051] (Results of obtaining general requirements information) "Communication ratio Company A survey results: According to a survey of the communication ratio of free Wi-Fi users at exhibition venues, approximately 80% use smartphones and approximately 20% use PCs. Smartphone users: Approximately 60% browse SNS, approximately 30% browse the web, and approximately 10% watch videos. PC users: Approximately 50% browse the web, approximately 30% send and receive emails, and approximately 20% download documents. Company B survey results: In a survey conducted in 2020, the communication ratio of free Wi-Fi users at exhibition venues was as follows: Information gathering: Approximately 50% SNS: Approximately 30% Watching videos: Approximately 10% Voice calls: Approximately 5% Other: Approximately 5%"

[0052] In the above embodiment, an example was shown in which general requirements information is obtained from external information stored as LLM, but the invention is not limited to this, and general requirements information may also be obtained from, for example, search results on the internet.

[0053] Figure 6 is a block diagram showing an example of the hardware configuration of a data processing device according to one embodiment of the present invention. In the example shown in Figure 6, the data processing device 100 according to the above embodiment is composed of, for example, a server computer or a personal computer, and has a hardware processor 111A such as a CPU (Central Processing Unit). A program memory 111B, a data memory 112, an input / output interface 113, and a communication interface 114 are connected to this hardware processor 111A via a bus 115.

[0054] The communication interface 114 includes, for example, one or more wireless communication interface units, enabling the transmission and reception of information with the communication network. As the wireless interface, for example, an interface employing a low-power wireless data communication standard such as a wireless LAN (Local Area Network) is used.

[0055] Input / output interface 113 is connected to an input device 200 and an output device 300 used by users, etc., which are attached to the data processing device 100. The input / output interface 113 can take in operation data entered by users, etc., through an input device 200 such as a keyboard, touch panel, or touchpad, and can also output output data to an output device 300, including a display device using liquid crystal or organic EL (electroluminescence), etc., for display. Note that the input device 200 and output device 300 may be devices built into the data processing device 100, or input devices and output devices of other information terminals that can communicate with the data processing device 100 via a network may be used.

[0056] The program memory 111B is a non-temporary tangible storage medium in which a non-volatile memory that can be written to and read at any time, such as an HDD (Hard Disk Drive) or SSD (Solid State Drive), is used in combination with another non-volatile memory such as ROM (Read Only Memory), and can store programs necessary for executing various control processes, etc., according to one embodiment.

[0057] The data memory 112 is a tangible storage medium that, for example, uses a combination of the above-mentioned non-volatile memory and volatile memory such as RAM (Random Access Memory), and can be used to store various data or information acquired and created during the process of various operations.

[0058] A data processing device 100 according to one embodiment of the present invention may be configured as a data processing device having the parts shown in Figure 1 as the software processing function unit.

[0059] The information storage unit used as work memory by each part of the data processing device 100 may be configured using the data memory 112 shown in Figure 6. However, these storage areas are not essential components within the data processing device 100, and may be, for example, areas provided in an external storage medium such as a USB (Universal Serial Bus) memory, or in a storage device such as a database server located in the cloud.

[0060] Each of the processing functions in the above-described section can be implemented by having the hardware processor 111A read and execute a program stored in the program memory 111B. Some or all of these processing functions may be implemented in various other forms, including application-specific integrated circuits (ASICs) or field-programmable gate arrays (FPGAs).

[0061] Furthermore, the methods described in each embodiment can be stored as programs (software means) that can be executed by a computer on recording media such as magnetic disks (floppy disks, hard disks, etc.), optical disks (CD-ROMs, DVDs, MOs, etc.), and semiconductor memories (ROMs, RAMs, flash memories, etc.), and can also be transmitted and distributed via communication media. The programs stored on the media also include configuration programs that configure the computer to run software means (including not only the execution program but also tables or data structures). The computer implementing this device reads the program recorded on the recording media and, if necessary, constructs the software means using the configuration program, and executes the above-described processes by controlling the operation of this software means. Note that the recording media referred to in this specification are not limited to those for distribution, but also include storage media such as magnetic disks or semiconductor memories provided inside the computer or in devices connected via a network.

[0062] It should be noted that the present invention is not limited to the embodiments described above, and can be modified in various ways during implementation without departing from its essence. Furthermore, each embodiment may be combined as appropriate, and in that case, the combined effects can be obtained. Moreover, the above embodiments include various inventions, and various inventions can be extracted by selecting combinations from the multiple constituent elements disclosed. For example, if the problem can be solved and effects obtained even if some constituent elements are deleted from all the constituent elements shown in the embodiment, then the configuration with these deleted constituent elements can be extracted as an invention.

[0063] 100...Data processing unit 10...Service usage status input unit 20...Intent acquisition unit 30...Information supplementation unit 31...Information acquisition determination unit 32...General requirements information acquisition unit 40...Quantitative Intent calculation unit

Claims

1. A data processing device comprising: an input unit that receives input from a user of the service regarding the circumstances of the service usage; a completion unit that, when the circumstances of the service usage include expressions that require completion, outputs the circumstances of the service usage in which such expressions have been completed using general knowledge obtainable from an external source; and a calculation unit that calculates requirements for providing the service to the user based on the circumstances of the service usage completed by the completion unit.

2. The data processing device according to claim 1, further comprising a storage device that stores information for supplementing expressions that require supplementation, wherein the supplementation unit outputs the service usage status in which the expressions requiring supplementation have been supplemented, when the service usage status includes expressions that require supplementation and information for supplementing the expressions that require supplementation is not stored in the storage device, by utilizing general knowledge obtainable from an external source.

3. A data processing method performed by a data processing device, comprising: an input unit of the data processing device receiving input from a user of the service regarding the circumstances of the service usage; a supplementation unit of the data processing device outputting the circumstances of the service usage in which, when the circumstances of the service usage include expressions that require supplementation, the expressions that require supplementation have been supplemented using general knowledge obtainable from an external source; and a calculation unit of the data processing device calculating requirements for providing the service to the user based on the circumstances of the service usage that have been supplemented by the supplementation unit.

4. A data processing program that causes a processor to function as a component of the data processing apparatus described in claim 1 or 2.

Images