Device and method for carrying out blood relationship examination by using large-scale language model to locate abnormity

Abstract

The invention relates to a device and a method for carrying out blood relationship examination by using a large-scale language model to locate abnormity, and the method comprises the following steps of: analyzing an abnormity locating request of a natural language by using the large-scale language model to obtain field information of abnormal target data; obtaining consanguinity data of the target data according to the field information so as to generate difference link information representing abnormal upstream data and / or services, and outputting an abnormality positioning result containing the difference link information in a natural language through a large language model. The complexity of the inspection process and professional requirements of personnel can be reduced, the inspection time is shortened, and the technical effects of improving the inspection efficiency, accuracy and satisfaction are achieved.

Description

Technical Field

[0001] An abnormal data detection device and method thereof, particularly a device and method for locating abnormalities by performing lineage checks using a large language model. Background Technology

[0002] The electronics assembly industry's characteristics of low profit margins and high sales volume, coupled with fierce price competition, drive businesses to pursue more effective control and optimization of raw materials and production tools, maximizing the efficiency of factory production resources. To effectively manage production lines in the electronics assembly industry, production line engineers often monitor visual dashboards on the production line to understand order completion status and notify maintenance personnel when data anomalies are detected.

[0003] After receiving notification from the production line engineer, operations and maintenance personnel typically need to perform a series of complex checks, including data source checks, ETL process troubleshooting, data quality verification, upstream and downstream data flow analysis, and debugging and verification. Data source checks verify the proper functioning of the data source and identify any missing or incorrect data. ETL process troubleshooting involves sequentially checking the data extraction, transformation, and loading processes, and reviewing the data processing records and execution status at each step. Data quality verification checks the integrity, accuracy, and consistency of the data. Upstream and downstream data flow analysis traces the data flow path to pinpoint the actual point where the problem occurred. Debugging and verification involve manually executing relevant scripts and tasks to verify the location of the problem.

[0004] However, the aforementioned inspection work requires checking multiple systems and tools. Therefore, the inspection work is complicated and involves many steps, which requires a lot of time to check each link step by step. It takes a long time to locate the link where the problem occurs, resulting in low inspection efficiency, inability to quickly respond to user reactions and needs, and the inspection process is highly dependent on the experience and skills of maintenance personnel, which is also prone to human error.

[0005] In summary, it is evident that the existing technology has long suffered from problems such as high complexity, long time consumption, low efficiency, and reliance on personnel experience in the inspection process of existing production lines. Therefore, it is necessary to propose improved technical means to solve this problem. Summary of the Invention

[0006] In view of the problems of existing technologies in inspecting anomalies on the production line, such as high complexity, long time consumption, low efficiency, and reliance on human experience, this invention discloses an apparatus and method for locating anomalies by performing lineage checks using a large language model, wherein:

[0007] The apparatus disclosed in this invention for locating anomalies by performing lineage checks using a large language model includes at least: a request acquisition module for acquiring an anomaly location request for target data; a large language model for parsing the anomaly location request to acquire feature parameters, the feature parameters including field information of the target data; a lineage tracking module for acquiring lineage information of the target data based on the field information, and acquiring lineage data based on the lineage information, the lineage data including upstream data and service detection results; and an anomaly location module for comparing whether the upstream data matches the target data and determining whether the service that generated the upstream data is normal based on the service detection results to generate difference link information, thereby enabling the large language model to generate and output anomaly location results, the anomaly location results including difference link information.

[0008] The method for locating anomalies using a large language model, disclosed in this invention, includes at least the following steps: obtaining an anomaly location request for target data; parsing the anomaly location request using a large language model to obtain feature parameters, the feature parameters including field information of the target data; obtaining the lineage information of the target data based on the field information; obtaining lineage data based on the lineage information, the lineage data including upstream data and service detection results; comparing whether the upstream data matches the target data and determining whether the service that generated the upstream data is normal based on the service detection results to generate discrepancy information; and outputting the anomaly location result through the large language model, the anomaly location result including the discrepancy information.

[0009] The apparatus and method disclosed in this invention are as described above. The difference between this invention and the prior art lies in that this invention uses a large language model to parse the anomaly localization request of natural language to obtain the field information of the target data in which the anomaly occurred. Based on the field information, it obtains the lineage data of the target data to generate difference link information representing the upstream data and / or services in which the anomaly occurred. Then, it outputs the anomaly localization result containing the difference link information in natural language through a large language model. This solves the problems existing in the prior art and can achieve the technical effect of reducing the need for repeated inspections and improving inspection efficiency, accuracy and satisfaction. Attached Figure Description

[0010] Figure 1 This is a schematic diagram of the components of the device proposed in this invention for performing bloodline checks using a large language model to locate abnormalities.

[0011] Figure 2 This is a schematic diagram of the processor module proposed in this invention.

[0012] Figure 3A This is a flowchart of the method for locating abnormalities by performing lineage checks using a large language model, as proposed in this invention.

[0013] Figure 3B This is a flowchart of the method for obtaining lineage data of target data proposed in this invention.

[0014] Figure 3C This is a flowchart of the method for actively generating anomaly location requests proposed in this invention.

[0015] The annotations in the attached figures are explained as follows:

[0016] 100: Device

[0017] 110: Memory

[0018] 120: Input Unit

[0019] 130: Communication Interface

[0020] 140: Storage medium

[0021] 150: Output Unit

[0022] 170: Processor

[0023] 190: Bus

[0024] 210: Anomaly Detection Module

[0025] 220: Request Generation Module

[0026] 230: Request module

[0027] 240: Large Language Model

[0028] 250: Bloodline Tracking Module

[0029] 260: Query Tool

[0030] 270: Anomaly Location Module

[0031] Step 305: Obtain the production line's initial output and actual output.

[0032] Step 307: When the input output does not match the actual output, generate an anomaly location request with the actual output and / or input output as the target data.

[0033] Step 310: Obtain an anomaly location request for the target data

[0034] Step 330: Use a large language model to parse the anomaly localization request to obtain feature parameters, which contain field information of the target data.

[0035] Step 350: Obtain the lineage information of the target data based on the field information.

[0036] Step 360: Obtain kinship data based on kinship information. Kinship data includes upstream data and service test results.

[0037] Step 361: Generate node query statements based on bloodline information

[0038] Step 363: Call the query tool

[0039] Step 365: The query tool reads upstream data from the upstream data repository based on the node query statement. Step 367: The service that generates lineage data is detected to produce service detection results.

[0040] Step 370: Compare whether the upstream data matches the target data and determine whether the service that generated the upstream data is functioning normally based on the service test results to generate discrepancy information.

[0041] Step 390: Output anomaly localization results using a large language model. The anomaly localization results include information on the points of difference. Detailed Implementation

[0042] The features and implementation methods of the present invention will be described in detail below with reference to the accompanying drawings and embodiments. The content is sufficient to enable any person skilled in the art to easily and fully understand the technical means used by the present invention to solve the technical problem and to implement it accordingly, thereby achieving the effects that the present invention can achieve.

[0043] This invention utilizes a Large Language Model (LLM) to analyze user-input descriptions of anomalies to obtain field information of the anomalous data. It then traces the lineage of the anomalous data to pinpoint the point at which the anomaly occurred. Finally, the LLM responds to the user with the details of that point. The descriptions are typically sentences in natural language. The points at which the anomalies occur may be due to data storage errors (e.g., errors in the storage medium or system) or data processing errors (e.g., errors in data processing services performing conversions or calculations). However, this invention is not limited to these limitations.

[0044] The lineage data proposed in this invention includes upstream data and the service detection results of the service that generated the upstream data. More specifically, in this invention, when source data is processed to produce result data, the source data is referred to as the upstream data of the result data. It should be noted that in this invention, upstream data may also be another type of result data. That is, if data A is generated by a service processing data B, and data B is generated by another service processing data C, then data B and data C are both upstream data of data A. Similarly, if another service processes data D to generate data C, then data D is also upstream data of data A.

[0045] The apparatus for implementing this invention can be a computing device. The computing device of this invention includes, but is not limited to, one or more processing modules, one or more memory modules, and buses connecting different hardware components (including memory modules and processing modules). Through the included hardware components, the computing device can load and execute an operating system, allowing the operating system to run on the computing device, and can also execute software or programs. The computing device also includes a housing, within which the aforementioned hardware components are disposed.

[0046] The bus of the computing device proposed in this invention can include one or more types, such as a data bus, address bus, control bus, expansion bus, and / or local bus. The bus of the computing device includes, but is not limited to, Industry Standard Architecture (ISA) bus, Peripheral Component Interconnect (PCI) bus, Video Electronics Standards Association (VESA) local bus, and serial Universal Serial Bus (USB), PCI Express (PCI-E / PCIe) bus, etc.

[0047] The processing module of the computing device proposed in this invention is coupled to a bus. The processing module includes a register set or register space, which may be entirely located on the processing chip of the processing module, or wholly or partially located outside the processing chip and coupled to the processing chip via dedicated electrical connections and / or via a bus. The processing module may be a central processing unit, a microprocessor, or any suitable processing element. If the computing device is a multiprocessor device, that is, the computing device contains multiple processing modules, then the processing modules contained in the computing device are identical or similar and are coupled and communicate via a bus. In some embodiments, the processing module may interpret a computer instruction or a series of multiple computer instructions to perform specific operations or calculations, such as mathematical operations, logical operations, data comparison, copying / moving data, etc., thereby driving other hardware components in the computing device or running an operating system or executing various programs and / or modules. Computer instructions can be assembly language instructions, instruction set architecture instructions, machine instructions, machine-dependent instructions, microinstructions, firmware instructions, or source code or object code written in any combination of one or more programming languages. Computer instructions can be executed entirely on a single computing device, partially on a single computing device, or partially on one computing device and partially on another connected computing device. The aforementioned programming languages ​​include object-oriented programming languages ​​such as Common Lisp, Python, C++, Objective-C, Smalltalk, Delphi, Java, Swift, C#, Perl, Ruby, etc., as well as conventional procedural programming languages ​​such as C or other similar programming languages.

[0048] Computing devices typically include one or more chipsets. The processing modules of the computing device can be coupled to or electrically connected to the chipset via a bus. A chipset consists of one or more integrated circuits (ICs), including a memory controller and peripheral input / output (I / O) controllers, etc. That is, the memory controller and I / O controllers can be contained within a single IC or implemented using two or more ICs. Chipsets typically provide I / O and memory management functions, as well as multiple general-purpose and / or special-purpose registers, timers, etc., which can be accessed or used by one or more processing modules coupled to or electrically connected to the chipset. In some embodiments, the chipset may also be part of the processing module.

[0049] The processing module of a computing device can also access data in memory modules and mass storage areas installed on the computing device through the memory controller. The aforementioned memory modules include any type of volatile memory and / or non-volatile memory (NVRAM), such as Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Read-Only Memory (ROM), Flash memory, etc. The aforementioned mass storage areas can contain any type of storage device or storage medium, such as hard disk drives, optical discs, flash drives, memory cards, solid-state drives (SSDs), or any other storage device. In other words, the memory controller can access data in static random access memory, dynamic random access memory, flash memory, hard disk drives, and solid-state drives.

[0050] The processing module of a computing device can also connect and communicate with peripheral devices or interfaces such as peripheral output devices, peripheral input devices, communication interfaces, and various data or signal receiving devices via a peripheral input / output controller and a peripheral input / output bus. Peripheral input devices can be any type of input device, such as a keyboard, mouse, trackball, touchpad, joystick, etc. Peripheral output devices can be any type of output device, such as a monitor, printer, etc. Peripheral input devices and peripheral output devices can also be the same device, such as a touchscreen. Communication interfaces can include wireless communication interfaces and / or wired communication interfaces. Wireless communication interfaces can include interfaces supporting wireless local area networks (such as Wi-Fi, Zigbee, etc.), Bluetooth, infrared, near-field communication (NFC), 3G / 4G / 5G and other mobile communication networks (cellular networks), or other wireless data transmission protocols. Wired communication interfaces can be Ethernet devices, DSL modems, cable modems, asynchronous transfer mode (ATM) devices, or fiber optic communication interfaces and / or components, etc. The data or signal receiving device may include a GPS receiver or a physiological signal receiver, wherein the physiological signals received by the physiological signal receiver include, but are not limited to, heartbeat, blood oxygenation, etc. The processing module may periodically poll various peripheral devices and interfaces, enabling the computing device to input and output data through various peripheral devices and interfaces, and also to communicate with another computing device having the hardware components described above.

[0051] The following is a preliminary step. Figure 1 The schematic diagram of the device for locating abnormalities by performing bloodline checks using a large language model, as proposed in this invention, illustrates the device for implementing this invention. Figure 1 As shown, the device 100 of the present invention includes a memory 110, a communication interface 130, a storage medium 140, a processor 170, and a bus 190. The processor 170 can be connected to the memory 110, the input unit 120, the communication interface 130, the storage medium 140, and the output unit 150 via the bus 190.

[0052] The memory 110 can store one or more sets of computer instructions.

[0053] Input unit 120 can provide input data through peripheral input devices of device 100. For example, input unit 120 can input data through keyboard, mouse, touchpad, or touch screen.

[0054] The communication interface 130 can be connected to external network storage devices or servers, and request and download data from the connected network devices.

[0055] Storage medium 140 can store data or signals downloaded from communication interface 130, data or signals provided to processor 170 or required for processor 170 to operate, and data or signals generated by processor 170.

[0056] The output unit 150 can also output the data generated by the processor 170 through the peripheral output device of the device 100. For example, the output unit 150 can display the data through a display or a touch screen.

[0057] Processor 170 can be like Figure 2 The schematic diagram of the modules proposed in this invention shows that they include modules such as a request retrieval module 230, a large language model 240, a lineage tracing module 250, and an anomaly localization module 270, as well as optional anomaly detection module 210, request generation module 220, and query tool 260. In some embodiments, the processor 170 can execute computer instructions stored in the memory 110, and can generate a request after executing the computer instructions. Figure 2 The various modules within; in another embodiment, Figure 2 The modules within can be generated from one or more circuits and / or complete or partial chips and other hardware components; that is, the processor 170 includes components... Figure 2 The hardware components of each module in the processor 170, that is, each module included in the processor 170 can be a software module or a hardware module, and there are no particular limitations in this invention.

[0058] The anomaly detection module 210 can obtain the input output and actual output of the production line. Generally, the anomaly detection module 210 can obtain the input output and actual output from the management device on the production line through the communication interface 130, but the present invention is not limited to this. For example, the input output and actual output can also be provided through the input unit 120.

[0059] The anomaly detection module 210 can also determine whether the input output matches the actual output. For example, the anomaly detection module 210 can obtain the output ratio of one or more components of a product produced on the production line to the actual output, and determine whether the input output matches the actual output based on the output ratio. However, the method by which the anomaly detection module 210 determines whether the input output matches the actual output is not limited to the above. The anomaly detection module 210 can read the output ratio from the storage medium 140, or it can download the output ratio from the production line management device through the communication interface 130. This invention has no particular limitations.

[0060] The request generation module 220 can generate an anomaly location request with the actual output and / or input output as target data when the anomaly detection module 210 determines that the input output does not match the actual output. For example, the request generation module 220 can add the names or identification data of the products produced on the production line, the names or identification data of the stations where the input output and actual output of components do not match, the names or identification data of the Kanban boards where the input output and actual output do not match, the names or identification data of the components where the input output and actual output do not match, and the inconsistent actual output and / or input output as target data to the corresponding positions in a predefined natural language template, thereby generating an anomaly location request using natural language. However, the present invention is not limited to this; for example, the request generation module 220 can also directly use the above-mentioned target data as an anomaly location request. The natural language in the natural language template can be correctly parsed by the large language model 240.

[0061] The request acquisition module 230 is responsible for acquiring abnormal location requests for target data. Generally, the request acquisition module 230 acquires abnormal location requests through the input unit 120 or the communication interface 130, but the present invention is not limited thereto. For example, it can also acquire abnormal location requests generated by the request generation module 220.

[0062] The large language model 240 is responsible for parsing the anomaly location request obtained by the request acquisition module 230, so as to obtain feature parameters after parsing. The feature parameters obtained by the large language model 240 include field information of the target data. The aforementioned field information includes, but is not limited to, the station name or identification data on the production line, the field name of the target data, etc. The large language model 240 can be trained to understand the key prompts in the anomaly location request, such as technical terms, encoding rules of various identification data, Kanban names of various Kanban boards, station names of various stations, field names of various data, etc., so as to identify and extract feature parameters from the anomaly location request. For example, if the anomaly location request is "SH0123456789's fixed asset number in Kanban A is inaccurate", the large language model 240 can identify from the anomaly location request that the product identification data is "SH0123456789", the Kanban name is "Kanban A", and the target data is the fixed asset number. Furthermore, the large language model 240 can look up the field information such as the fixed asset number's field name "assettag" through a pre-established correspondence table. In this way, the large language model 240 can obtain feature parameters such as product identification data, Kanban name, and target data's field information from the anomaly location request.

[0063] The large language model 240 is also responsible for generating anomaly localization results. The anomaly localization results generated by the large language model 240 include the difference information generated by the anomaly localization module 270. The large language model 240 can also output the generated anomaly localization results through the output unit 150, for example, by displaying or printing the anomaly localization results. Generally, the anomaly localization results are in natural language, that is, the large language model 240 can generate anomaly localization results that express the difference information in natural language, but the present invention is not limited thereto.

[0064] The lineage tracing module 250 is responsible for obtaining the lineage information of the target data based on the field information obtained by the large language model 240. The lineage information obtained by the lineage tracing module 250 includes data sources of one or more upstream data that generated the target data and one or more services that generated the target data or upstream data. The aforementioned data sources typically represent devices, databases, or files storing data, such as the name or identification data or network address of the device, the name or identification data of the database, or the file name, but this invention is not limited thereto. For example, the lineage tracing module 250 can look up the data source of the upstream data of the target data in a pre-established field mapping table using the field information (and the dashboard name) of the target data. It can also query the service name or service identification data of the service that generated the target data based on the upstream data or the service that records the relationship between two related data sources of source data and result data in a pre-established service mapping table.

[0065] The lineage tracing module 250 is also responsible for obtaining the lineage data associated with the target data based on the obtained lineage information. For example, the lineage tracing module 250 can generate a node query statement based on the data source of the upstream data of the target data in the obtained lineage information, and execute the generated node query statement to read the upstream data associated with the target data from one or more upstream data repositories. The lineage tracing module 250 can also query the system log file to check whether the service is operating normally based on the service name or service identification data of the service that generated the upstream data of the target data in the obtained lineage information.

[0066] In some embodiments, the bloodline tracking module 250 can obtain the bloodline information of the target data and the bloodline data associated with the target data through the query tool 260, or it can directly obtain the bloodline information and bloodline data in the same way as the query tool 260. There are no particular limitations in this invention.

[0067] The query tool 260 can obtain the field information acquired by the lineage tracking module 250 and generate lineage information for the target data based on the acquired field information. For example, the query tool 260 can obtain a field lineage table and query the lineage information from the field lineage table based on data such as product name or identification data, site name or identification data, component name or identification data, and field name in the field information, but the present invention is not limited thereto. The query tool 260 can read the field lineage table from the storage medium 140 or download the field lineage table from a network device or a management device on the production line via the communication interface 130.

[0068] The query tool 260 can also generate node query statements corresponding to different data sources based on the generated lineage information, and can also read upstream data associated with the target data from the data source based on the generated node query statements, for example, by executing node query statements to obtain upstream data. The aforementioned data sources can be one or more upstream data repositories, which may include data warehouses (DW), data marketplaces (DM), operational data stores (ODS), and raw databases, but this invention is not limited thereto. For example, query tool 260 can generate node query statements corresponding to different data sources based on the upstream data source of the target data. If the field name of the target data's data field is "assettag", when querying upstream data from the data warehouse, query tool 260 can generate a node query statement of "Select assettag From dw.fact_product_sn Where sno='SH0123456789'". When querying upstream data from the operation data table, query tool 260 can generate a node query statement of "Select assettag From ods.sno Where sno='SH0123456789'". When querying upstream data from the data marketplace or the original database, query tool 260 can also generate similar node query statements. And when querying upstream data from the data production layer, query tool 260 can generate "Select data->custattribute->assettag As assettag Fromproduct_stream Where The node query statement for "data->sn='SH0123456789'" is used, but this invention is not limited thereto.

[0069] The query tool 260 can also detect services that generate upstream data to produce service detection results for each detected service. These services include, but are not limited to, data extraction, transformation, and loading (ETL) functions in databases, and cloud-based synchronous storage services.

[0070] The anomaly localization module 270 is responsible for comparing each upstream data obtained by the lineage tracing module 250 with the target data. It is also responsible for determining whether the service generating each upstream data is normal based on the service detection results obtained by the lineage tracing module 250. Furthermore, it is responsible for generating discrepancy information based on whether the upstream data matches the target data and whether the service generating each upstream data is normal. This discrepancy information includes the data source of the upstream data that does not match the target data, and the name of the service that generated the abnormal upstream data.

[0071] The system operation and method of the present invention will then be explained using an embodiment, and please refer to [reference needed]. Figure 3A The flowchart of the method for locating anomalies using a large language model proposed in this invention is shown. In this embodiment, it is assumed that device 100 also operates on a production line.

[0072] First, the request acquisition module 230 of device 100 can continuously detect whether an abnormal location request occurs. If so, the request acquisition module 230 can obtain the abnormal location request (step 310). In this embodiment, it is assumed that the production line engineer on the production line can continuously monitor the visual dashboard on the production line to understand the order completion status. When the production line engineer finds that the actual output of a certain production line is abnormal from the input output, he can contact the maintenance personnel through SMS or telephone. After receiving the notification from the production line engineer, the maintenance personnel can use natural language to write an abnormal location request containing the name or identification data of the dashboard of the abnormal production line and the abnormal data (i.e., target data) on the dashboard of the abnormal production line, and transmit the written abnormal location request to device 100, so that the request acquisition module 230 of device 100 can obtain the abnormal location request received by the communication interface 130.

[0073] After the request acquisition module 230 of device 100 obtains the anomaly location request (step 310), the large language model 240 of device 100 can parse the anomaly location request to obtain feature parameters (step 330). In this embodiment, it is assumed that the feature parameters include the dashboard name and the site where the anomaly occurred.

[0074] After the large language model 240 of device 100 obtains the feature parameters, the lineage tracking module 250 of device 100 can obtain the lineage information of the target data based on the field information in the feature parameters (step 350). In this embodiment, it is assumed that the large language model 240 can call the lineage tracking module 250 and transmit the field information as a parameter to the lineage tracking module 250 during the call, so that the lineage tracking module 250 can query the lineage information of the target data based on the field information.

[0075] After the bloodline tracking module 250 of device 100 obtains the bloodline information of the target data, it can obtain the bloodline data associated with the target data based on the bloodline information (step 360). In this embodiment, it is assumed that... Figure 3B As shown in the process, if the lineage data contains the upstream data source of the target data (i.e., the name of the upstream data repository) and the name of the data extraction and transformation loading function between the upstream data sources, then the lineage tracking module 250 can generate a node query statement corresponding to one or more upstream data repositories based on the upstream data source (step 361), and can call the query tool 260 (step 363). The query tool 260 can query the corresponding upstream data from the corresponding upstream data repository based on the node query statement (step 365) and can detect the service that generated the upstream data (i.e., the data extraction and transformation loading function) to generate the corresponding service detection result (step 367).

[0076] Back Figure 3A After the lineage tracking module 250 of device 100 obtains the lineage data associated with the target data, the anomaly location module 270 of device 100 can compare whether each upstream data in the lineage data matches the target data and determine whether the service that generated each upstream data is normal based on the service detection results in the lineage data to generate difference link information (step 370). In this embodiment, it is assumed that the lineage tracking module 250 can transmit the obtained lineage data back to the large language model 240 of device 100, so that the large language model 240 calls the anomaly location module 270, and can provide the lineage data as a parameter to the anomaly location module 270 during the call, so that the anomaly location module 270 can compare whether each upstream data in the lineage data matches the target data, and the anomaly location module 270 can determine whether the service that generated each upstream data is normal based on the service detection results in the lineage data. In this way, the anomaly location module 270 can generate difference link information.

[0077] After the anomaly localization module 270 of device 100 generates the difference link information, the large language model of device 100 can output the anomaly localization result containing the difference link information (step 390). In this embodiment, it is assumed that the anomaly localization module 270 can provide the generated difference link information to the large language model 240, so that the large language model uses a natural language model to generate a semantic anomaly localization result containing the difference link information, and the generated anomaly localization result can be displayed to maintenance personnel through the output unit 150 of device 100.

[0078] Thus, through this invention, maintenance personnel can generate anomaly localization requests to a large language model using natural language, and the large language model can generate anomaly localization results in natural language, enabling maintenance personnel to intuitively understand the process in which the anomaly occurred.

[0079] In the above embodiments, if the device 100 further includes an anomaly detection module 210 and a request generation module 220, then as follows: Figure 3C In the process shown, before the request acquisition module 230 obtains the anomaly location request (step 310), the anomaly detection module 210 can connect to the management device on the production line through the communication interface 130 of the device 100 to download the data displayed on the visual dashboard on the production line, thereby obtaining the input output and actual output on the production line (step 305), and can determine whether the obtained input output and actual output are consistent. If the anomaly detection module 210 determines that the input output and actual output are inconsistent, the request generation module 220 can generate an anomaly location request with the actual output and / or input output as the target data (step 307), and can provide the generated anomaly location request to the request acquisition module 230, so that the request acquisition module 230 obtains the anomaly location request (step 310).

[0080] In summary, the difference between this invention and existing technologies lies in its technical means of using a large language model to parse natural language anomaly location requests to obtain field information of the target data where the anomaly occurred, obtaining lineage data of the target data based on the field information to generate difference information of upstream data and / or services where the anomaly occurred, and outputting anomaly location results containing difference information in natural language through a large language model. This technical means can solve the problems of high complexity, long time consumption, low efficiency, and reliance on human experience in the process of checking anomalies on the production line in existing technologies, thereby achieving the technical effect of reducing the need for repeated checks and improving inspection efficiency, accuracy, and satisfaction.

[0081] Furthermore, the method of using a large language model to perform lineage checks to locate abnormalities according to the present invention can be implemented in hardware, software, or a combination of hardware and software. It can also be implemented in a centralized manner in a computer system or in a decentralized manner with different components distributed among several interconnected computer systems.

[0082] While the embodiments disclosed in this invention are as described above, the content is not intended to directly limit the scope of patent protection for this invention. Any modifications made by those skilled in the art to the form and details of the implementation of this invention without departing from the spirit and scope disclosed herein shall fall within the scope of patent protection for this invention. The scope of patent protection for this invention shall still be determined by the scope defined in the appended claims.

Claims

1. A method for pedigree testing using a large language model to locate abnormalities, applied to a device or system, characterized in that, The method includes at least the following steps: An anomaly location request to obtain target data; The anomaly localization request is parsed using a large language model to obtain feature parameters, which contain field information of the target data. Based on the field information, obtain the bloodline information of the target data; Based on the bloodline information, bloodline data is obtained, and the bloodline data includes at least one upstream data and at least one service detection result; Compare whether the at least one upstream data matches the target data and determine whether the service that generated the at least one upstream data is normal based on the service detection result to generate difference link information; and The large language model outputs anomaly localization results, which include the information about the differences.

2. The method for pedigree testing using a large language model to locate abnormalities as described in claim 1, characterized in that, The step of obtaining the lineage information of the target data based on the field information is to use the field information as input parameters of the query tool and call the query tool to generate the lineage information of the target data.

3. The method for pedigree testing using a large language model to locate abnormalities as described in claim 1, characterized in that, The step of obtaining the lineage data from the lineage information further includes generating a node query statement based on the lineage information, and calling a query tool using the large language model based on the node query statement, so that the query tool reads the at least one upstream data from at least one upstream data repository and detects the service that generated the lineage data to generate the at least one service detection result.

4. The method for pedigree testing using a large language model to locate abnormalities as described in claim 1, characterized in that, The step of outputting the anomaly localization result through the large language model further includes the step of generating the anomaly localization result in natural language to express the information of the difference links.

5. The method for pedigree testing using a large language model to locate abnormalities as described in claim 1, characterized in that, Before the step of obtaining the anomaly location request for the target data, the method further includes the step of obtaining the production line's input output and actual output, and when it is determined that the input output and the actual output do not match, generating the anomaly location request with the actual output and / or the input output as the target data.

6. A device for kinship testing using a large-scale language model to locate abnormalities, characterized in that, The device comprises at least: The request retrieval module is used to retrieve abnormal location requests for target data; A large language model is used to parse the anomaly location request to obtain feature parameters, which include field information of the target data; A lineage tracing module is used to obtain the lineage information of the target data based on the field information, and to obtain the lineage data based on the lineage information. The lineage data includes at least one upstream data and at least one service detection result. and An anomaly localization module is used to compare whether the at least one upstream data matches the target data and to determine whether the service that generated the at least one upstream data is normal based on the service detection result, so as to generate difference link information, and to enable the large language model to generate and output anomaly localization results, the anomaly localization results including the difference link information.

7. The apparatus for locating abnormalities by performing bloodline checks using a large language model as described in claim 6, characterized in that, The device further includes a query tool for generating the lineage information of the target data based on the field information provided by the lineage tracing module.

8. The apparatus for locating abnormalities by performing kinship testing using a large language model as described in claim 6, characterized in that, The device further includes a query tool for reading at least one upstream data from at least one upstream data repository based on a node query statement generated by the lineage tracking module based on the lineage information, and detecting the service that generated the at least one upstream data to generate the at least one service detection result.

9. The apparatus for locating abnormalities by performing bloodline checks using a large language model as described in claim 6, characterized in that, The large language model generates the anomaly localization results by expressing the information of the differences in natural language.

10. The apparatus for locating abnormalities by performing kinship testing using a large language model as described in claim 6, characterized in that, The device further includes an anomaly detection module and a request generation module. The anomaly detection module is used to obtain the production output and actual output of the production line, and to determine whether the production output and the actual output are consistent. The request generation module is used to generate an anomaly location request with the actual output and / or the production output as the target data when the production output and the actual output are inconsistent.

Images