A method and system for extracting entity relations from operation and maintenance work orders

Abstract

A method and system for extracting entity relationships from an operation and maintenance work order. The method includes obtaining text data from an operation and maintenance work order, annotating the text data according to a preset operation and maintenance knowledge base to obtain text annotation data; performing word segmentation processing on the text annotation data to obtain a word sequence, and determining an annotation word vector of an entity annotation word in the word sequence; determining a semantic vector of the entity annotation word according to the annotation word vector and an entity pair specified by the word sequence, and determining an entity relationship extraction result of the operation and maintenance work order according to the semantic vector and a preset entity relationship extraction model. The solution of the present invention improves the accuracy of entity extraction from an operation and maintenance work order.

Description

Technical Field

[0001] The present invention belongs to the technical field of natural language processing, and in particular relates to a method and system for extracting entity relationships of an operation and maintenance work order. Background Art

[0002] With the rapid development of the power industry and the acceleration of technological innovation, the amount of data on power grid operation and maintenance work orders has exploded. These work order data not only record key information such as the operating status of power grid equipment, fault reports, and maintenance records, but also contain rich power knowledge and experience, which plays a vital role in the stable operation of the power industry, refined equipment maintenance, and efficient management decisions. However, in the face of such a large amount of data, how to accurately and efficiently extract key entities, such as equipment name, fault type, maintenance time, etc., has become a key technical problem that must be overcome in the process of information and intelligent upgrading of the power industry.

[0003] Traditional methods for extracting entities from maintenance work orders mainly rely on manually set rules and templates to identify and extract key information by matching preset patterns. This method is still competent when the data volume is small and the rules are clear, but with the dramatic increase in data volume and the continuous updating of technology in the power industry, traditional methods have gradually exposed the disadvantages of high manpower and time costs, difficulty in adapting to changes, and limited ability to process complex data. Especially in the context of the professionalism and complexity of maintenance work orders in the power industry, traditional methods often find it difficult to accurately identify and extract key entities, resulting in a significant reduction in the accuracy and efficiency of information extraction. Summary of the invention

[0004] In order to solve the deficiencies in the prior art, the present invention provides a method and system for extracting entity relationships of an operation and maintenance work order, so as to improve the accuracy of entity extraction of the operation and maintenance work order.

[0005] In order to solve the above technical problems, the present invention adopts the following technical solutions.

[0006] The present invention first discloses a method for extracting entity relationships of an operation and maintenance work order, the method comprising the following steps:

[0007] Acquire text data of the operation and maintenance work order, and annotate the text data according to a preset operation and maintenance knowledge base to obtain text annotation data;

[0008] Performing word segmentation processing on the text annotation data to obtain a word sequence, and determining an annotation word vector of an entity annotation word in the word sequence;

[0009] According to the annotation word vector and the entity pair specified by the word sequence, the semantic vector of the entity annotation word is determined, and the entity relationship extraction result of the operation and maintenance work order is determined according to the semantic vector and a preset entity relationship extraction model.

[0010] The present invention further includes the following preferred embodiments:

[0011] The step of labeling the text data according to a preset operation and maintenance knowledge base to obtain text labeling data further includes:

[0012] Performing data preprocessing on the text data to obtain processed text data, wherein the data preprocessing at least includes data cleaning, text normalization processing, and stop word removal processing;

[0013] Determining a plurality of named entities in the processed text data, and constructing entity triples based on three different named entities;

[0014] An operation and maintenance triplet identical to the entity triplet is searched from a preset operation and maintenance knowledge base, and the entity triplet is annotated according to the association degree identifier of the operation and maintenance triplet to obtain text annotation data.

[0015] The determining the semantic vector of the entity annotation word according to the annotation word vector and the entity pair specified by the word sequence further comprises:

[0016] Determine the entity pair specified by the word sequence, and obtain the starting position and the ending position of the entity pair in the word sequence;

[0017] Determining a relative distance between the entity annotation word and the entity pair according to the starting position and the ending position, and converting the relative distance into a position vector of the entity annotation word according to a preset position vector encoding rule;

[0018] The annotation word vector and the position vector are superimposed to obtain a semantic vector of the entity annotation word.

[0019] The step of determining the entity relationship extraction result of the operation and maintenance work order according to the semantic vector and a preset entity relationship extraction model further includes:

[0020] Determine a plurality of operation and maintenance category layers in a preset entity relationship extraction model, and determine a feature weight of each of the operation and maintenance category layers according to the semantic vector;

[0021] The product data between the semantic vector and each of the feature weights is superimposed to obtain the predicted entity relationship result of the operation and maintenance work order, and the operation and maintenance verification work order identical to the operation and maintenance work order is found from the operation and maintenance knowledge base, and the semantic verification vector corresponding to the operation and maintenance verification work order is input into the entity relationship extraction model for model training to obtain the verification entity relationship result of the operation and maintenance verification work order;

[0022] The entity relationship extraction result of the operation and maintenance work order is determined according to the cosine similarity between the predicted entity relationship result and the verified entity relationship result.

[0023] The determining the feature weight of each operation and maintenance category layer according to the semantic vector further includes:

[0024] Determine the layer sequence of each of the operation and maintenance category layers, and sort the operation and maintenance category layers in order from low to high according to the layer sequence to obtain sorted operation and maintenance category layers;

[0025] The semantic vectors are sequentially input into the sorted operation and maintenance category layers in order of the layer numbers from low to high for feature extraction, so as to obtain the feature weights of the semantic vectors on each of the operation and maintenance category layers.

[0026] The step of determining the entity relationship extraction result of the operation and maintenance work order according to the cosine similarity between the predicted entity relationship result and the verified entity relationship result further includes:

[0027] Determining the cosine similarity between the predicted entity relationship result and the verified entity relationship result, and detecting whether the cosine similarity is within a preset cosine threshold range;

[0028] If the cosine similarity is within the cosine threshold range, the predicted entity relationship result is used as the entity relationship extraction result of the operation and maintenance work order.

[0029] After detecting whether the cosine similarity is within a preset cosine threshold range, the entity relationship extraction method of the operation and maintenance work order includes:

[0030] If the cosine similarity is not within the cosine threshold range, the entity relationship extraction model is updated according to the preset operation and maintenance model parameters to obtain an updated entity relationship extraction model, and the updated entity relationship extraction model is used as the next preset entity relationship extraction model, and the step of determining multiple operation and maintenance category layers in the preset entity relationship extraction model is returned to execute.

[0031] The present invention also discloses an entity relationship extraction system for operation and maintenance work orders using the entity relationship extraction method for operation and maintenance work orders, comprising:

[0032] An acquisition module is used to acquire text data of the operation and maintenance work order, and annotate the text data according to a preset operation and maintenance knowledge base to obtain text annotation data;

[0033] A word segmentation module is used to perform word segmentation processing on the text annotation data to obtain a word sequence and determine the annotation word vector of the entity annotation word in the word sequence;

[0034] The extraction module is used to determine the semantic vector of the entity annotation word according to the entity pair specified by the annotation word vector and the word sequence, and determine the entity relationship extraction result of the operation and maintenance work order according to the semantic vector and a preset entity relationship extraction model.

[0035] Correspondingly, the present invention also discloses a terminal, including a processor and a storage medium;

[0036] The storage medium is used to store instructions;

[0037] The processor is used to operate according to the instructions to execute the steps of the entity relationship extraction method according to the aforementioned operation and maintenance work order.

[0038] Correspondingly, the present invention also discloses a computer-readable storage medium on which a computer program is stored, and when the program is executed by a processor, the steps of the entity relationship extraction method of the above-mentioned operation and maintenance work order are implemented.

[0039] The beneficial effect of the present invention is that, compared with the prior art, the present invention provides an entity relationship extraction method and system for an operation and maintenance work order, optimizes the existing operation and maintenance work order entity extraction method, and significantly improves the accuracy of operation and maintenance work order entity extraction. The text data of the operation and maintenance work order can be obtained in a timely manner, and the text data can be accurately annotated according to the preset operation and maintenance knowledge base, ensuring the standardization and readability of the text data, so that high-quality text annotation data can be accurately obtained. The text data is converted into an ordered sequence of words through word segmentation processing, and the annotation word vector of the entity annotation word in the word sequence is determined, thereby providing strong data support for subsequent semantic analysis. Furthermore, according to the entity pairs specified by the annotation word vector and the word sequence, the semantic vector of the entity annotation word can be accurately obtained, so that the deep meaning of the text data can be deeply excavated, laying a solid foundation for the extraction of entity relationships. By automatically processing the semantic vector according to the preset entity relationship extraction model, the entity relationship extraction result of the operation and maintenance work order can be accurately determined, thereby achieving the goal of efficiently and accurately extracting valuable information from massive operation and maintenance data, significantly improving the accuracy of entity extraction of operation and maintenance work orders, and providing accurate and reliable technical support for the optimization of power grid operation and maintenance work. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a flowchart of the first embodiment of the entity relationship extraction method of the operation and maintenance work order of the present invention;

[0041] Figure 2 It is a flow chart of the second embodiment of the entity relationship extraction method of the operation and maintenance work order of the present invention;

[0042] Figure 3 A schematic diagram of the system structure of the entity relationship extraction system of the operation and maintenance work order of the present invention;

[0043] Figure 4 A schematic diagram of the device structure of the hardware operating environment involved in the device of the present invention;

[0044] Figure 5 A schematic diagram of the storage medium structure of the entity relationship extraction system for operation and maintenance work orders of the present invention. DETAILED DESCRIPTION

[0045] In order to make the purpose, technical solution and advantages of the present invention more clear, the technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0046] The embodiments described in the present invention are only some embodiments of the present invention, not all embodiments. Based on the spirit of the present invention, other embodiments obtained by ordinary technicians in this field without creative work are all within the protection scope of the present invention.

[0047] In view of the shortcomings of the prior art, the present invention proposes a method and system for extracting entity relationships from operation and maintenance work orders. Figure 1 As shown, the entity relationship extraction method of the operation and maintenance work order disclosed in the present invention includes the following steps:

[0048] See also Figure 1 The present invention provides a first embodiment of a method for extracting entity relationships of an operation and maintenance work order, the method comprising:

[0049] Step S10: Acquire text data of the operation and maintenance work order, and annotate the text data according to a preset operation and maintenance knowledge base to obtain text annotation data.

[0050] In this embodiment, according to the preset data collection frequency, a large number of operation and maintenance work orders in the field of power operation and maintenance are obtained from various power operation and maintenance websites, including power equipment full life cycle archives, fault handling reports, equipment inspection reports, and preventive maintenance reports.

[0051] Since these operation and maintenance work orders may exist in the form of pictures, videos, voices, and texts, after converting the operation and maintenance work orders in the form of pictures, videos, and voices into the same text format as the operation and maintenance work orders in the form of texts, the text data of a large number of operation and maintenance work orders can be accurately obtained. Then, the text data is accurately annotated according to the preset operation and maintenance knowledge base to ensure the standardization and readability of the text data, so that high-quality text annotation data can be accurately obtained.

[0052] Step S20: performing word segmentation processing on the text annotation data to obtain a word sequence, and determining the annotation word vector of the entity annotation word in the word sequence.

[0053] In this embodiment, according to the preset word segmentation rules, the continuous text annotation data is divided into multiple independent word units, and then each entity annotation word in the word sequence is encoded into an annotation word vector acceptable to the computer through the preset word embedding model, thereby providing strong data support for subsequent semantic analysis.

[0054] Step S30: Determine the semantic vector of the entity annotation word according to the annotation word vector and the entity pair specified by the word sequence, and determine the entity relationship extraction result of the operation and maintenance work order according to the semantic vector and a preset entity relationship extraction model.

[0055] In this embodiment, according to the entity pairs specified by the annotation word vector and the word sequence, the semantic vector of the entity annotation word can be accurately obtained, so that the deep meaning of the text data can be deeply excavated, laying a solid foundation for the extraction of entity relationships. Finally, the semantic vector is automatically processed according to the preset entity relationship extraction model, and the entity relationship extraction result of the operation and maintenance work order can be accurately determined, that is, the goal of efficiently and accurately extracting valuable information from massive operation and maintenance data is achieved, and the accuracy of the operation and maintenance work order entity extraction is significantly improved, providing accurate and reliable technical support for the optimization of power grid operation and maintenance work.

[0056] In a further embodiment, based on the first embodiment of the present invention, a second embodiment of the entity relationship extraction method of the operation and maintenance work order of the present invention is proposed, see Figure 2 The above step S10 may further include the following steps:

[0057] Step S101: performing data preprocessing according to the text data to obtain processed text data, wherein the data preprocessing at least includes data cleaning, text normalization processing, and stop word removal processing;

[0058] Step S102: determining a plurality of named entities in the processed text data, and constructing entity triples according to three different named entities;

[0059] Step S103: searching a preset operation and maintenance knowledge base for an operation and maintenance triplet identical to the entity triplet, and annotating the entity triplet according to the association degree identifier of the operation and maintenance triplet to obtain text annotation data.

[0060] In this embodiment, the data processing flow of steps S101 to S103 can significantly improve the utilization efficiency and accuracy of operation and maintenance work order data. First, in data preprocessing, noise and irrelevant information in text data are removed by data cleaning, text normalization processing ensures the consistency of data format, and stop word removal further refines the text content, providing a purer data basis for subsequent analysis. Then, named entity recognition technology can accurately identify key entities in the text, such as equipment model, fault type, treatment measures, etc., and by constructing entity triples, it reveals the potential relationship between these entities. Finally, using the preset operation and maintenance knowledge base, the entity triples are accurately labeled, which not only realizes the in-depth understanding and interpretation of the operation and maintenance work order data, but also provides strong support for subsequent fault prediction, diagnosis and operation and maintenance decisions through the degree of association identification. This set of data processing processes not only improves the level of intelligence of power operation and maintenance work, but also provides solid technical support for power companies to achieve efficient and accurate operation and maintenance management.

[0061] Further, based on the first embodiment and / or the second embodiment of the present invention, a third embodiment of the entity relationship extraction method of the operation and maintenance work order of the present invention is proposed. The above step S30 may also include the following steps:

[0062] Step S301: determining an entity pair specified by the word sequence, and obtaining a starting position and an ending position of the entity pair in the word sequence;

[0063] Step S302: determining the relative distance between the entity annotation word and the entity pair according to the starting position and the ending position, and converting the relative distance into a position vector of the entity annotation word according to a preset position vector encoding rule;

[0064] Step S303: superimposing the annotation word vector and the position vector to obtain a semantic vector of the entity annotation word.

[0065] In this embodiment, by determining the entity pairs specified in the word sequence and accurately locating the starting position and ending position of the entity pairs in the word sequence, a data basis is provided for the subsequent relative distance calculation. According to the starting position and the ending position, the relative distance between the entity annotation word and the entity pair can be accurately calculated, and the relative distance can be converted into the position vector of the entity annotation word according to the preset position vector encoding rule, so that the spatial position information of the word can be effectively encoded into the vector representation. By superimposing the annotation word vector and the position vector, the semantic vector of the entity annotation word containing rich semantic information is obtained. This technical process not only improves the representation accuracy of entity annotation words in the text, but also enhances the model's ability to understand the relationship between entities in the text, and provides strong technical support for text mining and intelligent analysis in the field of power operation and maintenance.

[0066] Further, based on the first embodiment, the second embodiment and / or the third embodiment of the present invention, a fourth embodiment of the entity relationship extraction method of the operation and maintenance work order of the present invention is proposed, and the above step S30 may also include the following steps:

[0067] Step A10: determining a plurality of operation and maintenance category layers in a preset entity relationship extraction model, and determining a feature weight of each of the operation and maintenance category layers according to the semantic vector;

[0068] Step A20: superimpose the product data between the semantic vector and each of the feature weights to obtain the predicted entity relationship result of the operation and maintenance work order, and find the operation and maintenance verification work order that is the same as the operation and maintenance work order from the operation and maintenance knowledge base, input the semantic verification vector corresponding to the operation and maintenance verification work order into the entity relationship extraction model for model training, and obtain the verification entity relationship result of the operation and maintenance verification work order;

[0069] Step A30: Determine the entity relationship extraction result of the operation and maintenance work order according to the cosine similarity between the predicted entity relationship result and the verified entity relationship result.

[0070] In this embodiment, by determining multiple operation and maintenance category layers in the preset entity relationship extraction model, and calculating the feature weight of each operation and maintenance category layer based on the semantic vector, it is ensured that the model can accurately extract features for different types of operation and maintenance work orders. By multiplying and superimposing the semantic vector and the feature weight, the predicted entity relationship result of the operation and maintenance work order is generated. In order to further improve the accuracy of the model, the semantic verification vector corresponding to the actual operation and maintenance verification work order is input into the model for training, and the verification entity relationship result is obtained. By calculating the cosine similarity between the predicted entity relationship result and the verified entity relationship result, the entity relationship extraction effect of the operation and maintenance work order is accurately evaluated. This process not only improves the automation level of operation and maintenance work order processing, but also significantly enhances the robustness and accuracy of the entity relationship extraction model by introducing a training mechanism for verification work orders, providing strong technical support for the intelligent and efficient operation and maintenance of power.

[0071] Furthermore, in some other feasible embodiments, step A10 may also include the following steps:

[0072] Step A101: Determine the layer sequence of each operation and maintenance category layer, and sort the operation and maintenance category layers in order from low to high according to the layer sequence to obtain sorted operation and maintenance category layers.

[0073] Step A102: input the semantic vectors into the sorted operation and maintenance category layers in order from low to high in the order of the layer numbers to extract features, and obtain feature weights of the semantic vectors on each of the operation and maintenance category layers.

[0074] In this embodiment, a more orderly and refined feature extraction method is provided for the model. By determining the layer order of multiple operation and maintenance category layers in the preset entity relationship extraction model, and sorting the operation and maintenance category layers according to this order, it is ensured that the model can follow a certain logical order when processing operation and maintenance information at different levels. According to the sorted layer order, the semantic vector is input into each operation and maintenance category layer in turn for feature extraction. This not only ensures that the model can deeply mine the operation and maintenance information in the semantic vector layer by layer, but also can assign reasonable feature weights to each operation and maintenance category layer to reflect the importance of the layer in the entity relationship extraction process. This orderly feature extraction method not only improves the model's processing efficiency of operation and maintenance information, but also enhances the model's ability to recognize complex entity relationships in operation and maintenance work orders, providing strong technical support for the intelligent and precise operation and maintenance of power.

[0075] Furthermore, in some feasible embodiments, step A30 may also include the following steps:

[0076] Step A301: determining the cosine similarity between the predicted entity relationship result and the verified entity relationship result, and detecting whether the cosine similarity is within a preset cosine threshold range;

[0077] Step A302: If the cosine similarity is within the cosine threshold range, the predicted entity relationship result is used as the entity relationship extraction result of the operation and maintenance work order.

[0078] In this embodiment, the accuracy and reliability of the extraction results are ensured. By calculating the cosine similarity between the predicted entity relationship results and the verified entity relationship results, the similarity between the two is quantitatively evaluated. Then, by detecting whether this cosine similarity is within the preset cosine threshold range, it can be determined whether the accuracy of the prediction result meets the preset requirements. If the cosine similarity is within the threshold range, it means that the prediction result is highly similar to the verification result and has a high accuracy. The predicted entity relationship result is directly used as the entity relationship extraction result of the operation and maintenance work order. This method based on cosine similarity comparison and threshold judgment not only improves the degree of automation of entity relationship extraction, but also ensures the accuracy and consistency of the extraction results, provides reliable data support for power operation and maintenance, and helps to improve the efficiency and accuracy of entity extraction of operation and maintenance work orders.

[0079] Furthermore, in some other feasible embodiments, step A301 may also include the following steps:

[0080] Step B10: If the cosine similarity is not within the cosine threshold range, the entity relationship extraction model is updated according to the preset operation and maintenance model parameters to obtain an updated entity relationship extraction model, and the updated entity relationship extraction model is used as the next preset entity relationship extraction model, and the step of determining multiple operation and maintenance category layers in the preset entity relationship extraction model is returned to execute.

[0081] In this embodiment, when the cosine similarity between the predicted entity relationship result and the verified entity relationship result does not meet the preset threshold requirement, an efficient model optimization and iteration strategy is adopted. The entity relationship extraction model is updated according to the preset operation and maintenance model parameters to ensure that the model can continue to learn and improve. The updated entity relationship extraction model will participate in the subsequent processing flow as the next preset model, and will be re-executed from the operation and maintenance category layer to ensure that the model can have higher accuracy and robustness when processing new operation and maintenance work orders. This iterative update method not only improves the self-correction ability of the model, but also helps the model adapt to the ever-changing operation and maintenance environment and needs, providing more reliable and intelligent technical support for power operation and maintenance.

[0082] The beneficial effect of the present invention is that, compared with the prior art, the present invention provides an entity relationship extraction method and system for an operation and maintenance work order, optimizes the existing operation and maintenance work order entity extraction method, and significantly improves the accuracy of operation and maintenance work order entity extraction. The text data of the operation and maintenance work order can be obtained in a timely manner, and the text data can be accurately annotated according to the preset operation and maintenance knowledge base, ensuring the standardization and readability of the text data, so that high-quality text annotation data can be accurately obtained. The text data is converted into an ordered sequence of words through word segmentation processing, and the annotation word vector of the entity annotation word in the word sequence is determined, thereby providing strong data support for subsequent semantic analysis. Furthermore, according to the entity pairs specified by the annotation word vector and the word sequence, the semantic vector of the entity annotation word can be accurately obtained, so that the deep meaning of the text data can be deeply excavated, laying a solid foundation for the extraction of entity relationships. By automatically processing the semantic vector according to the preset entity relationship extraction model, the entity relationship extraction result of the operation and maintenance work order can be accurately determined, thereby achieving the goal of efficiently and accurately extracting valuable information from massive operation and maintenance data, significantly improving the accuracy of entity extraction of operation and maintenance work orders, and providing accurate and reliable technical support for the optimization of power grid operation and maintenance work.

[0083] The present invention may be a system, a method and / or a computer program product. The present invention also discloses an entity relationship extraction system for operation and maintenance work orders based on the above-mentioned entity relationship extraction method for operation and maintenance work orders, comprising:

[0084] The acquisition module H01 is used to acquire text data of the operation and maintenance work order, and annotate the text data according to a preset operation and maintenance knowledge base to obtain text annotation data;

[0085] The word segmentation module H02 is used to perform word segmentation processing on the text annotation data to obtain a word sequence and determine the annotation word vector of the entity annotation word in the word sequence;

[0086] The extraction module H03 is used to determine the semantic vector of the entity annotation word according to the annotation word vector and the entity pair specified by the word sequence, and determine the entity relationship extraction result of the operation and maintenance work order according to the semantic vector and a preset entity relationship extraction model.

[0087] refer to Figure 4 , which shows a schematic diagram of the structure of an entity relationship extraction device for an operation and maintenance work order suitable for implementing an embodiment of the present invention. The entity relationship extraction device for an operation and maintenance work order in the embodiment of the present invention may include, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 4 The entity relationship extraction device for the operation and maintenance work order shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0088] like Figure 4As shown, the entity relationship extraction device for the operation and maintenance work order may include a processing device 1001 (such as a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM: Read Only Memory) 1002 or the program loaded from the storage device 1003 to the random access memory (RAM: Random Access Memory) 1004. Various programs and data required for the operation of the entity relationship extraction device for the operation and maintenance work order are also stored in RAM1004. The processing device 1001, ROM1002 and RAM1004 are connected to each other through a bus 1005. The input / output (I / O) interface 1006 is also connected to the bus. Typically, the following devices can be connected to the I / O interface 1006: input devices 1007 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; storage devices 1003 including, for example, a magnetic tape, a hard disk, etc.; and communication devices 1009. The communication device 1009 can allow the entity relationship extraction device of the operation and maintenance work order to communicate wirelessly or wired with other devices to exchange data. Although the figure shows the entity relationship extraction device of the operation and maintenance work order with various devices, it should be understood that it is not required to implement or have all the devices shown. More or fewer devices may be implemented or have instead.

[0089] In particular, according to the embodiments disclosed in the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present invention include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program contains program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through a communication device, or installed from the storage device 1003, or installed from the ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present invention are executed.

[0090] Based on the spirit of the present invention, those skilled in the art can easily think that a computer program product can be obtained based on the entity relationship extraction method of the above operation and maintenance work order. Figure 5 The computer program product may include a computer-readable storage medium on which are loaded computer-readable program instructions for causing a processor to implement various aspects of the present invention. That is, the present invention also includes a terminal, including a processor and a storage medium; the storage medium is used to store instructions; the processor is used to operate according to the instructions to execute the steps of the entity relationship extraction method according to the aforementioned operation and maintenance work order.

[0091] Computer readable storage medium can be a tangible device that can keep and store the instructions used by the instruction execution device. Computer readable storage medium can be, for example, - but not limited to - electrical storage device, magnetic storage device, optical storage device, electromagnetic storage device, semiconductor storage device or any suitable combination of the above. More specific examples (non-exhaustive list) of computer readable storage medium include: portable computer disk, hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), portable compact disk read-only memory (CD-ROM), digital versatile disk (DVD), memory stick, floppy disk, mechanical encoding device, such as punch card or groove protrusion structure with instructions stored thereon and any suitable combination of the above. Computer readable storage medium used here is not interpreted as instantaneous signal itself, such as radio wave or other free propagating electromagnetic wave, electromagnetic wave propagated by waveguide or other transmission medium (for example, light pulse by optical fiber cable) or electrical signal transmitted by wire.

[0092] The computer-readable program instructions described herein can be downloaded from a computer-readable storage medium to each computing / processing device, or downloaded to an external computer or external storage device via a network, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network can include copper transmission cables, optical fiber transmissions, wireless transmissions, routers, firewalls, switches, gateway computers, and / or edge servers. The network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in the computer-readable storage medium in each computing / processing device.

[0093] The computer program instructions for performing the operation of the present invention may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, etc., and conventional procedural programming languages ​​such as "C" language or similar programming languages. The computer-readable program instructions may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., using an Internet service provider to connect through the Internet). In some embodiments, the electronic circuit, such as a programmable logic circuit, a field programmable gate array (FPGA) or a programmable logic array (PLA), may be executed by using the state information of the computer-readable program instructions to personalize and customize the electronic circuit, thereby implementing various aspects of the present invention.

[0094] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents, and any modifications or equivalent replacements that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for extracting entity relationships from operation and maintenance work orders, characterized in that: The entity relationship extraction method of the operation and maintenance work order includes: Acquire text data of the operation and maintenance work order, and annotate the text data according to a preset operation and maintenance knowledge base to obtain text annotation data; Performing word segmentation processing on the text annotation data to obtain a word sequence, and determining an annotation word vector of an entity annotation word in the word sequence; Determine the semantic vector of the entity annotation word according to the annotation word vector and the entity pair specified by the word sequence, and determine the entity relationship extraction result of the operation and maintenance work order according to the semantic vector and a preset entity relationship extraction model; The step of labeling the text data according to a preset operation and maintenance knowledge base to obtain text labeling data further includes: Performing data preprocessing on the text data to obtain processed text data, wherein the data preprocessing at least includes data cleaning, text normalization processing, and stop word removal processing; Determining a plurality of named entities in the processed text data, and constructing entity triples based on three different named entities; Searching for an operation and maintenance triplet identical to the entity triplet from a preset operation and maintenance knowledge base, and annotating the entity triplet according to the association degree identifier of the operation and maintenance triplet to obtain text annotation data; The determining the semantic vector of the entity annotation word according to the annotation word vector and the entity pair specified by the word sequence further comprises: Determine the entity pair specified by the word sequence, and obtain the starting position and the ending position of the entity pair in the word sequence; Determining a relative distance between the entity annotation word and the entity pair according to the starting position and the ending position, and converting the relative distance into a position vector of the entity annotation word according to a preset position vector encoding rule; Superimposing the annotation word vector and the position vector to obtain a semantic vector of the entity annotation word; The step of determining the entity relationship extraction result of the operation and maintenance work order according to the semantic vector and a preset entity relationship extraction model further includes: Determine a plurality of operation and maintenance category layers in a preset entity relationship extraction model, and determine a feature weight of each of the operation and maintenance category layers according to the semantic vector; The product data between the semantic vector and each of the feature weights is superimposed to obtain the predicted entity relationship result of the operation and maintenance work order, and the operation and maintenance verification work order identical to the operation and maintenance work order is searched from the operation and maintenance knowledge base, and the semantic verification vector corresponding to the operation and maintenance verification work order is input into the entity relationship extraction model for model training to obtain the verification entity relationship result of the operation and maintenance verification work order; Determine the entity relationship extraction result of the operation and maintenance work order according to the cosine similarity between the predicted entity relationship result and the verified entity relationship result; The determining the feature weight of each operation and maintenance category layer according to the semantic vector further includes: Determine the layer sequence of each of the operation and maintenance category layers, and sort the operation and maintenance category layers in order from low to high according to the layer sequence to obtain sorted operation and maintenance category layers; Inputting the semantic vectors into the sorted operation and maintenance category layers in order from low to high in the order of the layer numbers to extract features, and obtaining feature weights of the semantic vectors on each of the operation and maintenance category layers; The step of determining the entity relationship extraction result of the operation and maintenance work order according to the cosine similarity between the predicted entity relationship result and the verified entity relationship result further includes: Determining the cosine similarity between the predicted entity relationship result and the verified entity relationship result, and detecting whether the cosine similarity is within a preset cosine threshold range; If the cosine similarity is within the cosine threshold range, the predicted entity relationship result is used as the entity relationship extraction result of the operation and maintenance work order; After detecting whether the cosine similarity is within a preset cosine threshold range, the entity relationship extraction method of the operation and maintenance work order includes: If the cosine similarity is not within the cosine threshold range, the entity relationship extraction model is updated according to the preset operation and maintenance model parameters to obtain an updated entity relationship extraction model, and the updated entity relationship extraction model is used as the next preset entity relationship extraction model, and the step of determining multiple operation and maintenance category layers in the preset entity relationship extraction model is returned to execute.

2. An entity relationship extraction system for operation and maintenance work orders based on the entity relationship extraction method for operation and maintenance work orders according to claim 1, characterized in that: The entity relationship extraction system of the operation and maintenance work order includes: An acquisition module is used to acquire text data of the operation and maintenance work order, and annotate the text data according to a preset operation and maintenance knowledge base to obtain text annotation data; A word segmentation module is used to perform word segmentation processing on the text annotation data to obtain a word sequence and determine the annotation word vector of the entity annotation word in the word sequence; The extraction module is used to determine the semantic vector of the entity annotation word according to the entity pair specified by the annotation word vector and the word sequence, and determine the entity relationship extraction result of the operation and maintenance work order according to the semantic vector and a preset entity relationship extraction model.

3. An entity relationship extraction device for operation and maintenance work orders, characterized in that: The entity relationship extraction device for the operation and maintenance work order includes a processor, a memory, and an entity relationship extraction program for the operation and maintenance work order stored in the memory and executable by the processor, wherein when the entity relationship extraction program for the operation and maintenance work order is executed by the processor, the steps of the entity relationship extraction method for the operation and maintenance work order according to claim 1 are implemented.

4. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores an entity relationship extraction program for an operation and maintenance work order, wherein when the entity relationship extraction program for an operation and maintenance work order is executed by a processor, the steps of the entity relationship extraction method for an operation and maintenance work order according to claim 1 are implemented.

Images