CN111708868A - Text classification method, device and equipment for electric power operation and inspection events - Google Patents
Text classification method, device and equipment for electric power operation and inspection events Download PDFInfo
- Publication number
- CN111708868A CN111708868A CN202010039817.5A CN202010039817A CN111708868A CN 111708868 A CN111708868 A CN 111708868A CN 202010039817 A CN202010039817 A CN 202010039817A CN 111708868 A CN111708868 A CN 111708868A
- Authority
- CN
- China
- Prior art keywords
- text
- power
- electric power
- query text
- online operation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/332—Query formulation
- G06F16/3329—Natural language query formulation or dialogue systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/06—Electricity, gas or water supply
Abstract
A text classification method, device and equipment for electric power operation and inspection events are disclosed. In an embodiment of the present application, a text classification method for an electric power operation and inspection event may include: receiving an online operation and inspection event query text; performing text preprocessing on the online operation detection event query text to obtain power service words in the online operation detection event query text; determining a power business label of the power business word through a power business rule engine; processing the power service label through a pre-trained power language model to generate a feature vector of the online operation inspection event query text; processing the feature vectors of the online operation examination event query text through a pre-trained classification model, determining the category of the online operation examination event query text, and obtaining a text classification result; and outputting the text classification result. The method and the device can accurately classify the inquiry text facing to the inquiry and answer scene in the field of electric power operation and inspection.
Description
Technical Field
The invention relates to the technical field of electric power, in particular to a text classification method, a text classification device and text classification equipment for electric power operation and inspection events.
Background
At present, short text classification technologies in the fields of internet, news, medical treatment and the like, question classification based on text classification, intention identification and other technologies cannot be used for text classification of electric power operation inspection events because the texts and events related to the electric power operation inspection cannot be understood, and business logic and mechanism models related to electric power operation inspection services are not considered.
The classification counting of the defect texts of the existing power equipment can only classify the events of the defect classes respectively, and the classification problem of the query class texts facing the question and answer scene cannot be solved.
Disclosure of Invention
In order to solve the technical problems, it is desirable to provide a text classification method, device and equipment for an electric power operation inspection event, which can efficiently and accurately classify inquiry texts facing to an answer scene in electric power operation inspection.
According to one aspect of the application, a text classification method for an electric power operation detection event is provided, and comprises the following steps:
receiving an online operation and inspection event query text;
performing text preprocessing on the online operation detection event query text to obtain power service words in the online operation detection event query text;
determining a power business label of the power business word through a power business rule engine;
processing the power service label through a pre-trained power language model to generate a feature vector of the online operation inspection event query text;
processing the feature vectors of the online operation examination event query text through a pre-trained classification model, determining the category of the online operation examination event query text, and obtaining a text classification result;
and outputting the text classification result.
According to an aspect of the present application, there is provided a text classification apparatus for an electric power lucky event, including:
a receiving unit configured to receive an online operation detection event query text;
the preprocessing unit is configured to perform text preprocessing on the online operation inspection event query text to obtain power service words in the query text;
the power business rule engine is configured to determine a power business label of the power business word;
the feature vector generating unit is configured to process the power service label through a pre-trained power language model and generate a feature vector of the online operation inspection event query text;
and the classification unit is configured to process the feature vectors of the online operation detection event query text through a pre-trained classification model, determine the category of the online operation detection event query text, obtain a text classification result and output the text classification result.
According to an aspect of the present application, there is provided an electronic device including:
one or more processors;
a memory for storing the processor-executable instructions;
the one or more processors are used for reading the executable instructions from the memory and executing the instructions to realize the text classification method of the electric power operation detection event.
According to the embodiment of the application, the power language model, the classification model and the power business rule engine are combined, and the accurate classification of the query text facing to the query and answer scene in the field of power operation and inspection is realized.
Drawings
Fig. 1 is a flowchart illustrating a text classification method for an electric power inspection event according to an embodiment of the present application;
fig. 2 is a schematic structural diagram of a text classification device for an electric power operation and inspection event according to an embodiment of the present application;
FIG. 3 is a flowchart illustrating an exemplary implementation of text classification of an electrical inspection event according to an embodiment of the present application;
fig. 4 is a schematic structural diagram of an electronic device in an embodiment of the present application.
Detailed Description
Hereinafter, embodiments of the present application will be described in detail with reference to the accompanying drawings. It should be noted that, in the present application, the embodiments and the features thereof may be arbitrarily combined with each other without conflict.
Fig. 1 shows an exemplary flow of a text classification method for an electric power operation inspection event in an embodiment of the present application. As shown in fig. 1, a text classification method for an electric power operation inspection event in the embodiment of the present application may include:
step S101, receiving an online operation inspection event inquiry text;
step S102, performing text preprocessing on the online operation and inspection event query text to obtain power service words in the online operation and inspection event query text;
step S103, determining a power business label of a power business word through a power business rule engine;
step S104, processing the power business label through a pre-trained power language model to generate a feature vector of an online operation inspection event query text;
step S105, processing the feature vector of the online operation inspection event query text through a pre-trained classification model, determining the category of the online operation inspection event query text, and obtaining a text classification result;
and step S106, outputting a text classification result.
In the embodiment of the application, the electric power language model and the business rule engine are combined, and the advantages of uncertainty and randomness of oral language description in the electric power language model solving query text and certainty of electric power logic mechanism solving by the electric power business rule engine are fully integrated to realize the classification of the query text of the electric power operation test event, so that the classification accuracy of the electric power operation test query text is improved, and the electric power operation test query text can be accurately classified and recognized according to electric power professional business, oral vocabularies, oral sentences and the like.
In step S102, the text preprocessing may include: and identifying the electric power service words in the online operation and detection event query text. In addition, one or more of the following may also be included: word segmentation; filtering stop words; carrying out standardization treatment; removing stop words; removing high-frequency words; low frequency words are removed. In some examples, the normalization process may include processes such as case unifying, space elimination, and the like. In some examples, the text preprocessing may perform word segmentation and recognition for power professional terms and power related spoken language.
In step S102, the power service word may include one or more of the following: words representing electrical equipment names; words representing electrical equipment component names; words describing electrical equipment faults; words describing electrical equipment defects; and describing the numerical value of the electric equipment operation and detection parameter. In addition, the electric power service words may also include any other words related to electric power operation and detection, and the specific type thereof is not limited in the embodiments of the present application.
In step S103, the power service words may be parsed according to at least one of a standard guideline related to power operation and inspection and an expert experience database, so as to generate a power service tag. For example, the power business rules engine may generate a new label with deterministic power business significance by parsing certain entities and values in the power operational survey query text according to standard guidelines, expert experience regarding power operational surveys, such as normal/abnormal status under certain data range for certain types of indicators in the status evaluation guidelines.
In some examples, the power service tag may include one or more of: the type of electrical device; the type of power equipment component; a fault type of the electrical device; a defect type of the electrical equipment; and indicating that the electric equipment is normal or abnormal in operation. In addition, the power service tag may further include other tags that can accurately represent the power operation and detection related information, and the specific type of the tags is not limited in the embodiments of the present application.
Since the power operation test query text is generally a short text and is sparse in the matrix space, to ensure that the classification result is accurate, before step S104, the method may further include: and performing data enhancement processing on the power business label. For example, the data enhancement process may include, but is not limited to: synonym Replacement (SR), Random Insertion (RI), Random Swap (RS), Random Deletion (RD).
Before step S101, the method may further include: and constructing the power language model and the classification model, and training the power language model and the classification model by using the historical inspection event texts of known classes.
Specifically, the process of training the power language model and classification model using historical shipping event texts of known classes may include: step a1, initializing parameters of the power language model and parameters of the classification model; a2, acquiring a historical operation inspection event text of a known type, preprocessing the historical operation inspection event text, determining a power service label through a power service rule engine, processing the power service label by using a power language model to generate a characteristic vector of the historical operation inspection event, and processing the characteristic vector of the historical operation inspection event through a classification model to obtain a prediction type of the historical operation inspection event; step a3, adjusting parameters of the power language model and/or parameters of the classification model according to the prediction category of the historical operation inspection event and the real category of the historical operation inspection event. The above training process may be an iterative process, that is, the steps a2 and a3 are iterated until a predetermined convergence condition is reached, so as to finally determine the parameter values of the electric power language model and the parameter values of the classification model.
In some examples, the power language model of the embodiments of the present application may be a multi-semantic feature fused power language model. For example, the power language model may be, but is not limited to, SG, CBOW, GLOVE, FASTTEXT (fast text classification algorithm), BOW (one-hot, tf-idf, n-gram), BERT, and the like.
In some examples, the classification model of the embodiments of the present application may be, but is not limited to, SVM, XGBOOST, CNN, and the like.
According to the method and the device, a specific power language model is trained aiming at a power operation and inspection scene, the power language model is fused with a power business rule engine, a classification model supporting multiple event texts is trained, and different levels of classification of question sentences can be achieved. For example, classification of aspects of a business scenario may be performed, determining that an online power operational query text is a category such as an information query category, a fault handling category, and the like. For another example, the power health query text may be classified from a specific information perspective, and for example, it may be determined that the online power health query text is a category such as a device ledger, a hidden danger, a historical event, and the like. In other words, the text classification result in the embodiment of the present application may be information indicating a text category of the online power operation check query. For example, the text classification result may include a probability that the online power operation query text belongs to the information query class, a probability that the online power operation query text belongs to the fault handling class, and the like. As another example, the text classification result may include a probability that the online power operation check query text belongs to the equipment ledger, a probability that the online power operation check query text belongs to a hidden danger, a probability that the online power operation check query text belongs to a historical event, and the like. Of course, the power operation query text may be classified from other aspects, and the embodiments of the present application are not limited thereto.
Fig. 2 shows an exemplary structure of a text classification device of an electric power fortune detection event according to an embodiment of the present application. As shown in fig. 2, an exemplary structure of the text classification apparatus for an electric power lucky event may include:
a receiving unit 21 configured to receive an online operation examination event query text;
the preprocessing unit 22 is configured to perform text preprocessing on the online operation inspection event query text to obtain power service words in the query text;
a power business rule engine 23 configured to determine a power business label of the power business word;
a feature vector generating unit 24 configured to process the power service tag through a pre-trained power language model, and generate a feature vector of the online operation inspection event query text;
and the classification unit 25 is configured to process the feature vectors of the online inspection event query text through a pre-trained classification model, determine the category of the online inspection event query text, obtain a text classification result, and output the text classification result.
In some examples, the above exemplary structure of the text classification apparatus for an electric power fortune checking event may further include: and the data enhancement unit 26 is configured to perform data enhancement processing on the power business label obtained by the power business rule engine 23 and provide the feature vector generation unit 24.
For the specific technical details of each unit in the text classification device for the electric power operation inspection event, reference may be made to the above method part, and details are not repeated.
The following describes a specific implementation process of the present application in detail with reference to specific examples.
Fig. 3 shows an exemplary implementation flow of text classification of an electric power lucky inspection event in the embodiment of the present application. The process can include two phases of off-line training and on-line identification. In the off-line training stage, the historical inspection event text is used for obtaining an electric power language model and a classification model through text preprocessing, electric power business rule engine execution, data enhancement, electric power language model construction, training of an inspection event text classification model and the like. In the online identification process, text preprocessing, business rule engine processing and data enhancement can be performed on the online operation inspection event query text, and online identification is completed by fusing a feature vector with an electric power language model and a classification model to obtain a text classification result.
Embodiments of the present application also provide an electronic device, as shown in fig. 4, which may include one or more processors 41 and a memory 42 for storing executable instructions of the processors, where the processors 41 are configured to perform the steps of the text classification method for power lucky check events described above.
The processor 41 may be a CPU or other form of processing unit having instruction execution capabilities, and may control other components in the electronic device to perform desired functions.
Memory 42 may include one or more computer program products that may include various forms of computer-readable storage media, such as volatile memory and/or non-volatile memory. The volatile memory may include, for example, Random Access Memory (RAM), cache memory (cache), and/or the like. The non-volatile memory may include, for example, Read Only Memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium and executed by processor 41 to implement the text classification method for power biopsy events described above and/or other desired functions.
In one example, the electronic device may further include: an input device 43 and an output device 44, which are interconnected by a bus system and/or other form of connection mechanism (not shown). The input device 43 may also include, for example, a keyboard, mouse, etc., and may be configured to receive user-entered power question text. The output devices 44 may include, for example, a display, speakers, a printer, and a communication network and remote output devices connected thereto, among others, and may be configured to return answers to the user or devices used by the user in a predetermined form.
Of course, only a part of the components in the electronic device are shown in fig. 4 for simplicity, and components such as a bus, an input/output interface, and the like are omitted. In addition, the electronic device may include any other suitable components, depending on the particular application.
Furthermore, embodiments of the present application may further include a computer-readable storage medium having a computer program stored thereon, where the computer program, when executed by a processor, causes the processor to execute the steps in the text classification method for power health check events described above in this specification.
The above description is only for the preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present invention are also included in the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (10)
1. A text classification method for an electric power operation detection event comprises the following steps:
receiving an online operation and inspection event query text;
performing text preprocessing on the online operation detection event query text to obtain power service words in the online operation detection event query text;
determining a power business label of the power business word through a power business rule engine;
processing the power service label through a pre-trained power language model to generate a feature vector of the online operation inspection event query text;
processing the feature vectors of the online operation examination event query text through a pre-trained classification model, determining the category of the online operation examination event query text, and obtaining a text classification result;
and outputting the text classification result.
2. The method of claim 1, wherein the text pre-processing comprises at least: and identifying the electric power service words in the online operation and detection event query text.
3. The method of claim 2, wherein the text pre-processing further comprises at least one of:
word segmentation;
filtering stop words;
carrying out standardization treatment;
removing stop words;
removing high-frequency words;
low frequency words are removed.
4. The method of claim 1, wherein the power service term comprises one or more of:
words representing electrical equipment names;
words representing electrical equipment component names;
words describing electrical equipment faults;
words describing electrical equipment defects;
and describing the numerical value of the electric equipment operation and detection parameter.
5. The method of claim 1, wherein the power service tag comprises one or more of:
the type of electrical device;
the type of power equipment component;
a fault type of the electrical device;
a defect type of the electrical equipment;
and indicating that the electric equipment is normal or abnormal in operation.
6. The method of claim 1, further comprising: and before the electric power service label is processed through a pre-trained electric power language model, performing data enhancement processing on the electric power service label.
7. The method of claim 1, wherein determining, by a power business rules engine, a power business label for the power business term comprises:
and analyzing the electric power service words according to at least one of standard guide rules and expert experience databases related to electric power operation and inspection to generate electric power service labels.
8. The method of claim 1, wherein the method further comprises:
and constructing the power language model and the classification model, and training the power language model and the classification model by using the historical inspection event texts of known classes.
9. A text classification apparatus for an electric power lucky inspection event, comprising:
a receiving unit configured to receive an online operation detection event query text;
the preprocessing unit is configured to perform text preprocessing on the online operation inspection event query text to obtain power service words in the query text;
the power business rule engine is configured to determine a power business label of the power business word;
the feature vector generating unit is configured to process the power service label through a pre-trained power language model and generate a feature vector of the online operation inspection event query text;
and the classification unit is configured to process the feature vectors of the online operation detection event query text through a pre-trained classification model, determine the category of the online operation detection event query text, obtain a text classification result and output the text classification result.
10. An electronic device, comprising:
one or more processors;
a memory for storing the processor-executable instructions;
the one or more processors configured to read the executable instructions from the memory and execute the instructions to implement the method of any of claims 1-8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010039817.5A CN111708868A (en) | 2020-01-15 | 2020-01-15 | Text classification method, device and equipment for electric power operation and inspection events |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010039817.5A CN111708868A (en) | 2020-01-15 | 2020-01-15 | Text classification method, device and equipment for electric power operation and inspection events |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111708868A true CN111708868A (en) | 2020-09-25 |
Family
ID=72536543
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010039817.5A Pending CN111708868A (en) | 2020-01-15 | 2020-01-15 | Text classification method, device and equipment for electric power operation and inspection events |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111708868A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113515629A (en) * | 2021-06-02 | 2021-10-19 | 中国神华国际工程有限公司 | Document classification method and device, computer equipment and storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170086039A1 (en) * | 2015-09-21 | 2017-03-23 | Sap Se | Short text messaging in digital mobile telecommunication networks |
| CN107861942A (en) * | 2017-10-11 | 2018-03-30 | 国网浙江省电力公司电力科学研究院 | A kind of electric power based on deep learning is doubtful to complain work order recognition methods |
| CN107908716A (en) * | 2017-11-10 | 2018-04-13 | 国网山东省电力公司电力科学研究院 | 95598 work order text mining method and apparatus of word-based vector model |
| CN108170818A (en) * | 2017-12-29 | 2018-06-15 | 深圳市金立通信设备有限公司 | A kind of file classification method, server and computer-readable medium |
| CN108597519A (en) * | 2018-04-04 | 2018-09-28 | 百度在线网络技术（北京）有限公司 | A kind of bill classification method, apparatus, server and storage medium |
| CN108920473A (en) * | 2018-07-04 | 2018-11-30 | 中译语通科技股份有限公司 | A kind of data enhancing machine translation method based on similar word and synonym replacement |
-
2020
- 2020-01-15 CN CN202010039817.5A patent/CN111708868A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170086039A1 (en) * | 2015-09-21 | 2017-03-23 | Sap Se | Short text messaging in digital mobile telecommunication networks |
| CN107861942A (en) * | 2017-10-11 | 2018-03-30 | 国网浙江省电力公司电力科学研究院 | A kind of electric power based on deep learning is doubtful to complain work order recognition methods |
| CN107908716A (en) * | 2017-11-10 | 2018-04-13 | 国网山东省电力公司电力科学研究院 | 95598 work order text mining method and apparatus of word-based vector model |
| CN108170818A (en) * | 2017-12-29 | 2018-06-15 | 深圳市金立通信设备有限公司 | A kind of file classification method, server and computer-readable medium |
| CN108597519A (en) * | 2018-04-04 | 2018-09-28 | 百度在线网络技术（北京）有限公司 | A kind of bill classification method, apparatus, server and storage medium |
| CN108920473A (en) * | 2018-07-04 | 2018-11-30 | 中译语通科技股份有限公司 | A kind of data enhancing machine translation method based on similar word and synonym replacement |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113515629A (en) * | 2021-06-02 | 2021-10-19 | 中国神华国际工程有限公司 | Document classification method and device, computer equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108363790B (en) | Method, device, equipment and storage medium for evaluating comments | |
| CN110781276B (en) | Text extraction method, device, equipment and storage medium | |
| US10282468B2 (en) | Document-based requirement identification and extraction | |
| US10089296B2 (en) | System and method for sentiment lexicon expansion | |
| Fischbach et al. | Specmate: Automated creation of test cases from acceptance criteria | |
| CN111507088B (en) | Sentence completion method, equipment and readable storage medium | |
| CA3048356A1 (en) | Unstructured data parsing for structured information | |
| CN110377733B (en) | Text-based emotion recognition method, terminal equipment and medium | |
| CN111091004B (en) | Training method and training device for sentence entity annotation model and electronic equipment | |
| CN115328756A (en) | Test case generation method, device and equipment | |
| US20120078950A1 (en) | Techniques for Extracting Unstructured Data | |
| Krisch et al. | The myth of bad passive voice and weak words an empirical investigation in the automotive industry | |
| Schraagen et al. | Extraction of semantic relations in noisy user-generated law enforcement data | |
| US20100306143A1 (en) | Effort estimation using text analysis | |
| CN116882494B (en) | Method and device for establishing non-supervision knowledge graph oriented to professional text | |
| CN111708868A (en) | Text classification method, device and equipment for electric power operation and inspection events | |
| CN114969334B (en) | Abnormal log detection method and device, electronic equipment and readable storage medium | |
| Hänig et al. | Comparison of structured vs. unstructured data for industrial quality analysis | |
| CN115525750A (en) | Robot phonetics detection visualization method and device, electronic equipment and storage medium | |
| CN112735465B (en) | Invalid information determination method and device, computer equipment and storage medium | |
| Salman | Test Case Generation from Specifications Using Natural Language Processing | |
| CN114239602A (en) | Session method, apparatus and computer program product | |
| US11120381B2 (en) | Product declaration validation | |
| CN112183714A (en) | Automated data slicing based on artificial neural network | |
| Timoshchuk | Assessing the quality of the requirements specification by applying GQM approach and using NLP tools |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |