CN107861942A - A kind of electric power based on deep learning is doubtful to complain work order recognition methods - Google Patents
A kind of electric power based on deep learning is doubtful to complain work order recognition methods Download PDFInfo
- Publication number
- CN107861942A CN107861942A CN201710938687.7A CN201710938687A CN107861942A CN 107861942 A CN107861942 A CN 107861942A CN 201710938687 A CN201710938687 A CN 201710938687A CN 107861942 A CN107861942 A CN 107861942A
- Authority
- CN
- China
- Prior art keywords
- work order
- doubtful
- complaint
- learning
- model
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 42
- 238000013135 deep learning Methods 0.000 title claims abstract description 30
- 238000013136 deep learning model Methods 0.000 claims abstract description 8
- 238000010801 machine learning Methods 0.000 claims abstract description 6
- 239000013598 vector Substances 0.000 claims description 49
- 230000008569 process Effects 0.000 claims description 26
- 238000012545 processing Methods 0.000 claims description 20
- 238000012549 training Methods 0.000 claims description 13
- 230000005540 biological transmission Effects 0.000 claims description 12
- 230000007935 neutral effect Effects 0.000 claims description 8
- 238000010276 construction Methods 0.000 claims description 6
- 230000006378 damage Effects 0.000 claims description 6
- 230000005611 electricity Effects 0.000 claims description 6
- 238000007726 management method Methods 0.000 claims description 6
- 239000011159 matrix material Substances 0.000 claims description 6
- 230000011218 segmentation Effects 0.000 claims description 6
- 241001122315 Polites Species 0.000 claims description 5
- 230000008859 change Effects 0.000 claims description 5
- 238000012544 monitoring process Methods 0.000 claims description 5
- 101000911390 Homo sapiens Coagulation factor VIII Proteins 0.000 claims description 3
- 208000027418 Wounds and injury Diseases 0.000 claims description 3
- 102000057593 human F8 Human genes 0.000 claims description 3
- 230000008676 import Effects 0.000 claims description 3
- 208000014674 injury Diseases 0.000 claims description 3
- 229940047431 recombinate Drugs 0.000 claims description 3
- 238000013519 translation Methods 0.000 claims description 3
- 238000013024 troubleshooting Methods 0.000 claims description 3
- 230000002159 abnormal effect Effects 0.000 claims description 2
- 230000006399 behavior Effects 0.000 claims description 2
- 238000001914 filtration Methods 0.000 claims description 2
- 230000009191 jumping Effects 0.000 claims description 2
- 238000005516 engineering process Methods 0.000 abstract description 15
- 230000000694 effects Effects 0.000 abstract description 3
- 238000005201 scrubbing Methods 0.000 abstract description 2
- 239000002699 waste material Substances 0.000 abstract description 2
- 238000007689 inspection Methods 0.000 description 9
- 230000007246 mechanism Effects 0.000 description 7
- 230000006870 function Effects 0.000 description 5
- 238000004458 analytical method Methods 0.000 description 4
- 241001269238 Data Species 0.000 description 2
- 230000002776 aggregation Effects 0.000 description 2
- 238000004220 aggregation Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000004069 differentiation Effects 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 230000009849 deactivation Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000005036 nerve Anatomy 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000003825 pressing Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000001953 sensory effect Effects 0.000 description 1
- 238000004513 sizing Methods 0.000 description 1
- 230000001502 supplementing effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
-
- 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
- G06F16/3331—Query processing
- G06F16/3332—Query translation
- G06F16/3335—Syntactic pre-processing, e.g. stopword elimination, stemming
-
- 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
- G06F16/3331—Query processing
- G06F16/334—Query execution
- G06F16/3344—Query execution using natural language analysis
-
- 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
- G06F16/353—Clustering; Classification into predefined classes
-
- 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
- G06Q30/00—Commerce
- G06Q30/01—Customer relationship services
- G06Q30/015—Providing customer assistance, e.g. assisting a customer within a business location or via helpdesk
- G06Q30/016—After-sales
-
- 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—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Economics (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- General Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- General Health & Medical Sciences (AREA)
- Public Health (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Water Supply & Treatment (AREA)
- Finance (AREA)
- Human Resources & Organizations (AREA)
- Primary Health Care (AREA)
- Tourism & Hospitality (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
A kind of electric power based on deep learning is doubtful to complain work order recognition methods, is related to the complaint recognition methods of power customer service.Electric power is doubtful to complain work order to use manual identified, and efficiency is low, promptness is poor and waste of manpower resource.The present invention comprises the following steps：1）Deep learning model configures；2）Feature tag is complained to refine；3）Complain sample format；4）Model learning is trained；5）It is doubtful to complain identification；6）It is doubtful to complain classification.The technical program complains inventory to carry out data scrubbing sequence, complains a series of activities such as the refinement of tendency word, data modeling tuning, sample iterative learning are trained, machine learning is predicted by depth learning technology, realize doubtful complaint work order deep learning Intelligent Recognition and classification, lift intelligent work experience, management and control operating efficiency of improving service quality.
Description
Technical field
The present invention relates to the complaint recognition methods of power customer service, more particularly to a kind of electric power based on deep learning to doubt
Like complaint work order recognition methods.
Background technology
From 95598 incoming call analysis, though a large amount of clients do not complain directly, or seat judges non-complaint by accident, by seeking advice from,
The expression such as opinions and suggestions is discontented with to electric service, if dealing with improperly or not in time, may upgrade to customer complaint.How to subtract
The complaint amount of few user, improving the satisfaction of user turns into current power supply enterprise's focus of attention.To effective analyzer tube of complaint
Reason, can improve the satisfaction and loyalty of client, the persistently discovery and improvement to power service weak spot well, and lifting supplies
The service quality of electric enterprise, enterprise image have important and profound significance.
At present, service lifting be unable to do without client's demand analysis, but always enters by artificial inspection data, manual cleanup data
The modes such as row index prediction do not catch up with growth requirement seriously, and efficiency is low, promptness is poor and waste of manpower resource.The whole province's year is talked about
Business amount is up to 8,000,000, and center is using tradition sampling recording quality inspection pattern, by manually being repeated to listen to recording, work effect one by one
Rate is low and can not accurately and efficiently extract the complaint point of client, dissatisfied point etc..Only single doubtful complaint quality inspection normality task
Year 3456 people's hours of input, and this in following client's demand excacation proportion comprehensively far away from 1%.
In addition, a number of wrong job order of complaint in 95598 work orders be present, Guo Wang sales departments irregularly send nearly 30,000
Bar work order, it is desirable to quality inspection is completed in 6 days and is verified, time is pressing and the task is heavy, and center input 6-7 people works extra shifts or extra hours and could complete work.
Quality inspection workload input is big, efficiency is low and loss is higher, simultaneously because the deviation on understanding occurs in different quality inspection personnels
Cause Different Results, quality inspection work can not effectively be carried out.
The content of the invention
The technical problem to be solved in the present invention and the technical assignment proposed are prior art to be improved with being improved,
There is provided a kind of electric power based on deep learning doubtful complaint work order recognition methods, to reach the purpose for improving operating efficiency.Therefore,
The present invention takes following technical scheme.
A kind of electric power based on deep learning is doubtful to complain work order recognition methods, it is characterised in that comprises the following steps：
1）Deep learning model configures：Algorithm parameter for being related to learning template carries out unified configuration management, implementation model
Parameter subitem condition monitoring and dynamic configuration, being related to the information of deep learning model parameter includes pattern number, model name, work(
Can business description, regular parameter, random number, matrix line number, matrix columns, iterations, learning rate, model object class, god
Through the network number of plies；
2）Feature tag is complained to refine：Handled for complaining sample work order to accept content to history by Word2Vector classes, knot
Close Baidu's dictionary to carry out being formatted participle to work order content, sample machine learning is complained to history by system, carried automatically
Take and complain peculiar label word；Consider to need to recombinate between indivedual phrases simultaneously, by these label words in original sample work order content
Carry out system mark, individual distinguishing label word is finally subjected to restructuring refinement under manual intervention, is formed and complain characteristic vector label；Throw
Tell that characteristic vector label includes：Meter line wrong, civil compensation, over time limit, workmen service to have a power failure, is barbarous in violation of rules and regulations, without reason
Construct, behave badly, troubleshooting imperfection, very discontented, obsolete equipment are cleared up, mistakes family number, frequently has a power failure, frequently jumping
Lock, quality of voltage are abnormal for a long time, business apply to install it is over time limit, have a strong impact on, household electrical appliances damage, personnel's violation, personal injury, personnel
Service regulation, attitude be poor, not by power failure plan stop power transmission, frequency of supply for a long time exception, low-voltage, link process problem, stop
Power transmission information bulletin accuracy, device location, business hall service, the upgrading of rural power grids, repairing quality, it is over time limit in it is a variety of or whole；
3）Complain sample format：For filtering redundancy word, participle formatting processing to complaining sample to carry out, in dictionary is disabled
Increase needs the word filtered, including greeting, polite information, acquisition complaint structured language expression formula and write-back number in work order
According to storehouse；
4）Model learning is trained：Text value translation statement is complained, using vector space model, text is divided into some features
, calculate weight of each characteristic item in the text, and then by whole text to characteristic item weight for component to
Measure to represent, text is expressed as mathematical modeling with the mode of characteristic vector, then be iterated based on complaining sample vector to be grouped
Study；To it is doubtful complaint work order identification learning model realize artificial real-time oversight learn again or unartificial pattern under learn by oneself
Practise, and support that show backstage by learning training progress window exports to model deep learning process and study；
5）It is doubtful to complain identification：Judge that carrying out doubtful complain identifies by text similarity；Document participle uses space vector table
State, the semantic similarity between text is measured by the geometrical relationship between the two vectors in space；Learned based on above-mentioned
The Model Results of habit accept work order to all 95598 incoming calls and judged one by one；
6）It is doubtful to complain classification：The doubtful complaint work order judged is divided one by one based on the above-mentioned Model Results learnt
Class；Including first-level class, secondary classification, three-level classification, first-level class includes：Service tear open tell, do business lodge a complaint, stop power transmission lodge a complaint,
Power supply quality, power grid construction.
The technical program complains inventory to carry out data scrubbing sequence, complains tendency word to refine, number by deep learning technology
According to a series of activities such as modeling tuning, the training of sample iterative learning, machine learning predictions, doubtful complaint work order deep learning is realized
Intelligent Recognition and classification, lifting intelligent work experience, management and control operating efficiency of improving service quality.Using the depth of neutral net
Framework is practised, can build, shape and dispose neutral net, and interface and Hadoop and Spark effective integrations are provided, can be carried out big
Data cloud computing analyzes and processes, and can support the deep learning of the diversified forms such as data, text, image, voice.
The statement variation of client's demand content, wherein polite, redundancy word can shape to model learning and similarity judges band
Carry out many problems, the technical program sets deactivation dictionary, and automatic identification rejects these influence factors, and these words are added into stop words,
So as to the automatic rejection during structuring participle.
As further improving and supplementing to above-mentioned technical proposal, present invention additionally comprises following additional technical feature.
Further, in step 3）The step of middle complaint sample format, includes：
301）Sample is complained to import database；
302）Filter lodge a complaint content data and English；
303）Using segmenter combination dictionary and characteristic vector label is complained to carry out content participle；
304）Filter stop words；
305）Obtain and complain structured statement expression formula and write-back storehouse.
Further, in step 4）In, model learning train the step of comprise the following steps：
401）Learning model parameter, including regular parameter, iterations, learning rate, the neutral net number of plies are obtained according to model；
402）Learning model object class is positioned by above-mentioned model object class parameter；
403）Judge whether the model has the sample data that can learn；If so, then enter step 404）, otherwise skip to step 412）；
404）Create learning model；
405）Acquisition is lodged a complaint sample format data；
406）Sample vectorization processing, formatting is lodged a complaint into content with complaining classification to enter row vector according to defined learning model
Array instantiates；
407）By examples detailed above vector data order random alignment；
408）Whether judgement is less than study iterations；If less than study iterations, into step 409）, otherwise,
Skip to step 412)；
409）Database synchronization recording learning progress, including current study number, learning time interval；
410）Instantiation vector data is studied in groups；
411）The completion that studies in groups reinitializes vector index, is back to step 408)；
412）Learning tasks are completed by learning model write-back storehouse.
Random alignment again is carried out to the order of learning sample vector data in model learning training process；Reason is such as
Under：In model learning training process, complain sample to be converted into certain dry group vector data, according to vector space model mechanism, enter
Row is learnt by group, if undue concentration semanteme identical sample data set in some groups of vectors, can cause commenting for model
Divide to compare and lay particular stress on this group of data, so as to influence other group of aggregation of data scoring.Therefore sample is carried out when sample vector initializes
It is randomly ordered, so as to form a series of sample datas random, differentiation is semantic, be so advantageous to avoid model learning from training
During there is hypersaturated state, finally improve it is doubtful throwing work order identification with classification accuracy.
Further, in step 5）In doubtful identification of lodging a complaint comprise the following steps：
501）The complaint sample pattern learnt is obtained from database；
502）Based on the Model instantiation model object learnt；
503）Work order identifies whether to be less than work order quantity to be identified one by one；If so, then enter step 504）；Otherwise terminate；
504）Content sentence word segmentation processing is accepted to work order to be identified；
505）Sentence participle steering volume processing；
506）Similarity identification judgement is carried out based on learning model；
507）Doubtful complain judges, judges whether similarity exceedes setting value；If so, then enter step 508）；Otherwise it is back to step
Rapid 503）；
508）Doubtful complaint work order is obtained, and is back to step 503）.
Further, in step 6）During doubtful complaint classification, it includes step：
601）The complaint sample pattern learnt is obtained from database；
602）Based on the Model instantiation model object learnt；
603）Work order identifies whether to be less than work order quantity to be identified one by one；If so, then enter step 604）；Otherwise terminate；
604）Treat classification work order and accept content sentence word segmentation processing；
605）Sentence participle steering volume processing；
606）The classification output vector of lodging a complaint is obtained based on learning model；
607）Traversal classification output vector, which obtains, complains three-level classification corresponding to its maximum；And it is back to step 603）.
Further, servicing the secondary classification lodged a complaint includes service behavior, services channels；The secondary classification bag that business is lodged a complaint
Include Business Process System, expense is urged in electricity consumption change, meter reading, the electricity price electricity charge, electric energy metrical, business charge；Stop the secondary classification of power transmission complaint
Including：Stop power transmission information bulletin, power cut problem, repairing service, value-added service；The secondary classification of power supply quality includes：Voltage matter
Amount, frequency of supply, power supply reliability；Power grid construction includes：Power supply facilities, power construction.
Beneficial effect：
First, power industry takes the lead in attempting deep learning technology practical application in 95598 work order quality inspections at home, propose to be based on
The unartificial enforcement mechanisms electric power of deep learning is doubtful to complain work order identification technology, and it can be based on increment and complain sample to roll
Practise memory, and dynamic model adaptively reparation etc..
Second, using Deeplearning4j frameworks as technological break-through mouth, institute is functional to realize that assembly type encapsulates, autgmentability compared with
Good, strong adaptability, increasing change function newly can be with customer demand personalized customization, and all learning models can pass through interface configurations
Mode is completed, and reduces developer's pressure, lifts demand response promptness.
Third, it as the doubtful complaint work order identification technology of the electric power based on deep learning, 95598 peculiar business of fusion should
With, and demand personalization customization is carried out, there is function self-defined dynamic configuration, the study of unartificial enforcement mechanisms, learning process to exist
The features such as line monitoring, real-time learning memory, intelligent Accurate Prediction；Potential complaint risk can be efficiently identified, alleviates a line work
Make personnel service's pressure, lifting service risk surveillance management and control ability, it is pre- index analysis can efficiently to be obtained from big figureofmerit
It is alert, efficient, collaboration, the working mechanism of intelligence are realized, reaches the purpose increased efficiency by downsizing payrolls.By complaint handling from it is original " afterwards more
Mend " it is changed into " Prior Control " so that power supply enterprise is changed into actively, so as to be greatly reduced when handling complaint problem from passive
Customer complaint rate, lifting Electric Power Service in High are horizontal.
Brief description of the drawings
Fig. 1 is the complaint composition of sample implementation process figure of the present invention.
Fig. 2 is the model learning training implementation process figure of the present invention.
Fig. 3 is the doubtful complaint identification implementation process figure of the present invention.
Fig. 4, which is that the present invention is doubtful, complains classification implementation process figure.
Embodiment
Technical scheme is described in further detail below in conjunction with Figure of description.
The technical program is realized based on Deeplearning4j technologies, is configured by deep learning model, is complained feature mark
Label refinement, complaint sample format, model learning training, doubtful complaint identification, these key links of doubtful complaint classification are complete
Into a kind of complain work order identification technology towards power customer service and be based on the electric power that deep learning technology is realized is doubtful
System.It specifically includes following steps
1）Deep learning model configures：The process realizes that the algorithm parameter being related to learning template carries out unified configuration management, real
Existing model parameter subitem condition monitoring and dynamic configuration, being related to the details of deep learning model parameter has pattern number, mould
Type title, function service description, regular parameter (L2), random number (rngSeed), matrix line number (numRows), matrix columns
(numColumns), iterations (iterations), learning rate (learningRate), model object class (class), god
Through network number of plies (list) etc..
2）Feature tag is complained to refine：The process realizes that complaining sample work order to accept content to history passes through
Word2Vector classes（Deeplearning4j tool-class）Processing, with reference to Baidu's dictionary（Choose the word of more than 2 in dictionary）Enter
Row is formatted participle to work order content, complains sample machine learning to history by system, automatically extracts and complain peculiar mark
Sign word；Consider to need to recombinate between indivedual phrases simultaneously, these label words be subjected to system mark in original sample work order content,
Individual distinguishing label word is finally subjected to restructuring refinement under manual intervention.Partly complaint feature tag is：Meter line wrong, civil compensation
Repay, be over time limit, workmen service have a power failure in violation of rules and regulations, without reason, it is barbarous construct, behave badly, troubleshooting imperfection, very not
Full, obsolete equipment cleaning, mistake family number, frequently powers failure, Frequent trip, quality of voltage for a long time exception, business apply to install it is over time limit,
Have a strong impact on, household electrical appliances damage, personnel's violation, personal injury, personnel service's specification, attitude are poor, do not stop power transmission by power failure plan, supply
Exception, low-voltage, link process problem, power transmission information bulletin accuracy of stopping, device location, business hall take electric frequency for a long time
Business, the upgrading of rural power grids, repairing quality, it is over time limit in.
3）Complain sample format：The process is realized filters redundancy word, participle formatting processing to complaining sample to carry out, can
Increase needs the word that filters in dictionary is disabled, such as the information such as greeting in work order, polite.
Complain sample formatization as shown in Figure 1, it comprises the following steps：
301）Sample is complained to import database；
302）Filter lodge a complaint content data and English；
303）Using segmenter combination dictionary and characteristic vector label is complained to carry out content participle；
304）Filter stop words；
305）Obtain and complain structured statement expression formula and write-back storehouse.
4）Model learning is trained：Model learning training process core content is exactly to solve to complain text value translation statement,
Using vector space model, i.e., text is divided into some characteristic items, each characteristic item is calculated at this by specific means
Weight in text, and then whole text is represented to the weight of characteristic item for the vector of component, by text feature to
The mode of amount is expressed as mathematical modeling, then is iterated study based on complaining sample vector to be grouped.Process specific implementation is to doubting
Like complain the learning model that work order identifies realize artificial real-time oversight learn again or unartificial pattern under self study, and support logical
Cross learning training progress window and show backstage to model deep learning process and study output.
Model learning is trained as shown in Fig. 2 comprising the following steps：
401）Learning model parameter, including regular parameter, iterations, learning rate, the neutral net number of plies are obtained according to model；
402）Learning model object class is positioned by above-mentioned model object class parameter；
403）Judge whether the model has the sample data that can learn；If so, then enter step 404）, otherwise skip to step 412）；
404）Create learning model；
405）Acquisition is lodged a complaint sample format data；
406）Sample vectorization processing, formatting is lodged a complaint into content with complaining classification to enter row vector according to defined learning model
Array instantiates；
407）By examples detailed above vector data order random alignment；
408）Whether judgement is less than study iterations；If less than study iterations, into step 409）, otherwise,
Skip to step 412)；
409）Database synchronization recording learning progress, including current study number, learning time interval；
410）Instantiation vector data is studied in groups；
411）The completion that studies in groups reinitializes vector index, is back to step 408)；
412）Learning tasks are completed by learning model write-back storehouse.
5）It is doubtful to complain identification：It is doubtful to complain identification then to judge that carrying out doubtful complain identifies by text similarity.Once
Document participle is stated using space vector, and the semantic similarity can between text passes through between the two vectors in space
Geometrical relationship is measured.Similarity is assessed by model training and sets 70%, then complains recognition accuracy to reach 91.5% or so.Should
Process realizes that accept work order to all 95598 incoming calls based on the above-mentioned Model Results learnt is judged one by one.
Doubtful to lodge a complaint identification as shown in Fig. 3, it comprises the following steps：
501）The complaint sample pattern learnt is obtained from database；
502）Based on the Model instantiation model object learnt；
503）Work order identifies whether to be less than work order quantity to be identified one by one；If so, then enter step 504）；Otherwise terminate；
504）Content sentence word segmentation processing is accepted to work order to be identified；
505）Sentence participle steering volume processing；
506）Similarity identification judgement is carried out based on learning model；
507）Doubtful complain judges, judges whether similarity exceedes setting value；If so, then enter step 508）；Otherwise it is back to step
Rapid 503）；
508）Doubtful complaint work order is obtained, and is back to step 503）.
6）It is doubtful to complain classification：The process is realized based on the above-mentioned Model Results learnt to the doubtful complaint work that has judged
It is single to be classified one by one.Wherein criteria for classification is complained to be shown in Table shown in one：
The electric power of table one complains criterion with complaining classification detailed rules and regulations
It is doubtful to complain classification to realize as shown in figure 4, including step：
601）The complaint sample pattern learnt is obtained from database；
602）Based on the Model instantiation model object learnt；
603）Work order identifies whether to be less than work order quantity to be identified one by one；If so, then enter step 604）；Otherwise terminate；
604）Treat classification work order and accept content sentence word segmentation processing；
605）Sentence participle steering volume processing；
606）The classification output vector of lodging a complaint is obtained based on learning model；
607）Traversal classification output vector, which obtains, complains three-level classification corresponding to its maximum；And it is back to step 603）.
The technical program complains work order identification technology using electric power based on deep learning is doubtful, and it has, and following some is excellent
Point：First, power industry takes the lead in attempting deep learning technology practical application in 95598 work order quality inspections at home, propose based on deep
The doubtful complaint work order identification technology of unartificial enforcement mechanisms electric power of study is spent, it can be based on increment and complain sample to roll study
Memory, and dynamic model adaptively reparation etc..Second, using Deeplearning4j frameworks as technological break-through mouth, the functional realization of institute
Assembly type encapsulates, and autgmentability is preferable, strong adaptability, and increasing change function newly can be with customer demand personalized customization, all study
Model can be completed by interface configurations mode, reduce developer's pressure, lift demand response promptness.Third, it is as base
In the doubtful complaint work order identification technology of the electric power of deep learning, 95598 peculiar service applications are merged, and carry out demand personalization and determine
System, function have self-defined dynamic configuration, the study of unartificial enforcement mechanisms, learning process on-line monitoring, real-time learning memory, intelligence
The features such as energy Accurate Prediction；And this key technology can efficiently identify potential complaint risk in practical application, alleviate a line
Staff's service pressure, lifting service risk surveillance management and control ability, can efficiently obtain index analysis from big figureofmerit
Early warning, efficient, collaboration, the working mechanism of intelligence are realized, reaches the purpose increased efficiency by downsizing payrolls.By complaint handling from it is original " afterwards
Make up " it is changed into " Prior Control " so that power supply enterprise is changed into actively when handling complaint problem from passive, so as to significantly drop
Low customer complaint rate, lifting Electric Power Service in High are horizontal.
In the present embodiment, using the Deeplearning4j deep learning technologies increased income, it is the nerve based on Java
The deep learning framework of network, can build, shapes and dispose neutral net, and provide interface and Hadoop and Spark are effective
It is integrated, big data cloud computing analyzing and processing can be carried out.The depth of the diversified forms such as data, text, image, voice can be supported
Practise.
When deep learning nerual network technique is realized, a group model, excellent by model learning, last model again is defined first
Change type selecting.Complex Neural Network is broadly divided into input layer, hidden layer and output layer, and the right pair is it should be seen that coding is part in program
Code.Data are explained by machine sensory perceptual system, is marked or clusters to being originally inputted.Can identification pattern be included in
Numeric form in amount, therefore the data of all real worlds such as image, sound, text, time series must be converted into numerical value.
Help is clustered and classified.A cluster and classification on the data for storing and managing can be understood as establishing
Layer.For unlabelled data, neutral net can be according to the similarity of input sample by packet；If can be with marked
Data set sizing, neutral net can to data carry out genealogical classification.
There is ambiguity in the demand that client is fed back by 95598 hot lines, for example short message sending correlation work order belongs to complaint a bit
Some are not belonging to complain again, and this brings many challenges to machine learning and feature extraction.Need to comb comprehensively and complain learning sample
Data, precisely refine and complain feature tag, the design of optimization neural network hidden layer, complaint can be properly increased and judge similarity, from
And lift the accurate identification classification of doubtful complaint.
The statement variation of client's demand content, wherein polite, redundancy word can shape to model learning and similarity judges band
Carry out many problems, it is necessary to system automatic identification rejects these influence factors, these words are added into stop words, so as in structuring point
Automatic rejection during word.
In model learning training process, sample is complained to be converted into certain dry group vector data, according to vector space model machine
System, is learnt, if undue concentration semanteme identical sample data set in some groups of vectors, can cause model by group
Scoring compare and lay particular stress on this group of data, so as to influence the scoring of other group of aggregation of data.Therefore carried out when sample vector initializes
Sample is randomly ordered, so as to form a series of sample datas random, differentiation is semantic, is so advantageous to avoid model learning
Occur hypersaturated state in training process, finally improve the accuracy of doubtful throwing work order identification and classification.
The doubtful complaint work order recognition methods of a kind of electric power based on deep learning shown in figure 1 above -4 is the tool of the present invention
Body embodiment, substantive distinguishing features of the present invention and progress are embodied, can be according to the use needs of reality, in the enlightenment of the present invention
Under, carry out equivalent modifications to it, this programme protection domain row.
Claims (6)
1. a kind of electric power based on deep learning is doubtful to complain work order recognition methods, it is characterised in that comprises the following steps：
1）Deep learning model configures：Algorithm parameter for being related to learning template carries out unified configuration management, implementation model
Parameter subitem condition monitoring and dynamic configuration, being related to the information of deep learning model parameter includes pattern number, model name, work(
Can business description, regular parameter, random number, matrix line number, matrix columns, iterations, learning rate, model object class, god
Through the network number of plies；
2）Feature tag is complained to refine：Handled for complaining sample work order to accept content to history by Word2Vector classes, knot
Close Baidu's dictionary to carry out being formatted participle to work order content, sample machine learning is complained to history by system, carried automatically
Take and complain peculiar label word；Consider to need to recombinate between indivedual phrases simultaneously, by these label words in original sample work order content
Carry out system mark, individual distinguishing label word is finally subjected to restructuring refinement under manual intervention, is formed and complain characteristic vector label；Throw
Tell that characteristic vector label includes：Meter line wrong, civil compensation, over time limit, workmen service to have a power failure, is barbarous in violation of rules and regulations, without reason
Construct, behave badly, troubleshooting imperfection, very discontented, obsolete equipment are cleared up, mistakes family number, frequently has a power failure, frequently jumping
Lock, quality of voltage are abnormal for a long time, business apply to install it is over time limit, have a strong impact on, household electrical appliances damage, personnel's violation, personal injury, personnel
Service regulation, attitude be poor, not by power failure plan stop power transmission, frequency of supply for a long time exception, low-voltage, link process problem, stop
Power transmission information bulletin accuracy, device location, business hall service, the upgrading of rural power grids, repairing quality, it is over time limit in it is a variety of or complete
Portion；
3）Complain sample format：For filtering redundancy word, participle formatting processing to complaining sample to carry out, in dictionary is disabled
Increase needs the word filtered, including greeting, polite information, acquisition complaint structured language expression formula and write-back number in work order
According to storehouse；
4）Model learning is trained：Text value translation statement is complained, using vector space model, text is divided into some features
, calculate weight of each characteristic item in the text, and then by whole text to characteristic item weight for component to
Measure to represent, text is expressed as mathematical modeling with the mode of characteristic vector, then be iterated based on complaining sample vector to be grouped
Study；To it is doubtful complaint work order identification learning model realize artificial real-time oversight learn again or unartificial pattern under learn by oneself
Practise, and support that show backstage by learning training progress window exports to model deep learning process and study；
5）It is doubtful to complain identification：Judge that carrying out doubtful complain identifies by text similarity；Document participle uses space vector table
State, the semantic similarity between text is measured by the geometrical relationship between the two vectors in space；Learned based on above-mentioned
The Model Results of habit accept work order to all 95598 incoming calls and judged one by one；
6）It is doubtful to complain classification：The doubtful complaint work order judged is divided one by one based on the above-mentioned Model Results learnt
Class；Including first-level class, secondary classification, three-level classification, first-level class includes：Service tear open tell, do business lodge a complaint, stop power transmission lodge a complaint,
Power supply quality, power grid construction.
2. a kind of electric power based on deep learning according to claim 1 is doubtful to complain work order recognition methods, its feature exists
In：In step 3）The step of middle complaint sample format, includes：
301）Sample is complained to import database；
302）Filter lodge a complaint content data and English；
303）Using segmenter combination dictionary and characteristic vector label is complained to carry out content participle；
304）Filter stop words；
305）Obtain and complain structured statement expression formula and write-back storehouse.
3. a kind of electric power based on deep learning according to claim 2 is doubtful to complain work order recognition methods, its feature exists
In：In step 4）In, model learning train the step of comprise the following steps：
401）Learning model parameter, including regular parameter, iterations, learning rate, the neutral net number of plies are obtained according to model；
402）Learning model object class is positioned by above-mentioned model object class parameter；
403）Judge whether the model has the sample data that can learn；If so, then enter step 404）, otherwise skip to step
412）；
404）Create learning model；
405）Acquisition is lodged a complaint sample format data；
406）Sample vectorization processing, formatting is lodged a complaint into content with complaining classification to enter row vector according to defined learning model
Array instantiates；
407）By examples detailed above vector data order random alignment；
408）Whether judgement is less than study iterations；If less than study iterations, into step 409）, otherwise,
Skip to step 412)；
409）Database synchronization recording learning progress, including current study number, learning time interval；
410）Instantiation vector data is studied in groups；
411）The completion that studies in groups reinitializes vector index, is back to step 408)；
412）Learning tasks are completed by learning model write-back storehouse.
4. a kind of electric power based on deep learning according to claim 3 is doubtful to complain work order recognition methods, its feature exists
In：In step 5）In doubtful identification of lodging a complaint comprise the following steps：
501）The complaint sample pattern learnt is obtained from database；
502）Based on the Model instantiation model object learnt；
503）Work order identifies whether to be less than work order quantity to be identified one by one；If so, then enter step 504）；Otherwise terminate；
504）Content sentence word segmentation processing is accepted to work order to be identified；
505）Sentence participle steering volume processing；
506）Similarity identification judgement is carried out based on learning model；
507）Doubtful complain judges, judges whether similarity exceedes setting value；If so, then enter step 508）；Otherwise it is back to step
Rapid 503）；
508）Doubtful complaint work order is obtained, and is back to step 503）.
5. a kind of electric power based on deep learning according to claim 4 is doubtful to complain work order recognition methods, its feature exists
In：In step 6）During doubtful complaint classification, it includes step：
601）The complaint sample pattern learnt is obtained from database；
602）Based on the Model instantiation model object learnt；
603）Work order identifies whether to be less than work order quantity to be identified one by one；If so, then enter step 604）；Otherwise terminate；
604）Treat classification work order and accept content sentence word segmentation processing；
605）Sentence participle steering volume processing；
606）The classification output vector of lodging a complaint is obtained based on learning model；
607）Traversal classification output vector, which obtains, complains three-level classification corresponding to its maximum；And it is back to step 603）.
6. a kind of electric power based on deep learning according to claim 5 is doubtful to complain work order recognition methods, its feature exists
In：The secondary classification that service is lodged a complaint includes service behavior, services channels；The secondary classification lodged a complaint of doing business includes Business Process System, used
Expense, the electricity price electricity charge, electric energy metrical, business charge are urged in electricity change, meter reading；Stopping the secondary classification of power transmission complaint includes：Stop power transmission letter
Cease bulletin, power cut problem, repairing service, value-added service；The secondary classification of power supply quality includes：Quality of voltage, frequency of supply, confession
Electric reliability；Power grid construction includes：Power supply facilities, power construction.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710938687.7A CN107861942B (en) | 2017-10-11 | 2017-10-11 | Suspected power complaint work order identification method based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710938687.7A CN107861942B (en) | 2017-10-11 | 2017-10-11 | Suspected power complaint work order identification method based on deep learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107861942A true CN107861942A (en) | 2018-03-30 |
| CN107861942B CN107861942B (en) | 2021-10-26 |
Family
ID=61698361
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710938687.7A Active CN107861942B (en) | 2017-10-11 | 2017-10-11 | Suspected power complaint work order identification method based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107861942B (en) |
Cited By (36)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108710651A (en) * | 2018-05-08 | 2018-10-26 | 华南理工大学 | A kind of large scale customer complaint data automatic classification method |
| CN109190716A (en) * | 2018-10-23 | 2019-01-11 | 深圳增强现实技术有限公司 | Processing method, device and the electronic equipment of low-voltage collecting meter reading failure |
| CN109389418A (en) * | 2018-08-17 | 2019-02-26 | 国家电网有限公司客户服务中心 | Electric service client's demand recognition methods based on LDA model |
| CN109409553A (en) * | 2018-10-29 | 2019-03-01 | 国网河北省电力有限公司电力科学研究院 | A kind of power marketing business application system |
| CN109492660A (en) * | 2018-09-26 | 2019-03-19 | 平安科技（深圳）有限公司 | Calling information processing method, device, computer equipment and storage medium |
| CN109558486A (en) * | 2018-10-30 | 2019-04-02 | 国家电网有限公司客户服务中心 | Electric power customer service client's demand intelligent identification Method |
| CN109635292A (en) * | 2018-12-05 | 2019-04-16 | 杭州东方通信软件技术有限公司 | Work order quality detecting method and device based on machine learning algorithm |
| CN109710766A (en) * | 2018-12-29 | 2019-05-03 | 云南电网有限责任公司电力科学研究院 | A kind of the complaint trend analysis method for early warning and device of work order data |
| CN109726290A (en) * | 2018-12-29 | 2019-05-07 | 咪咕数字传媒有限公司 | Complaint classification model determination method and device and computer-readable storage medium |
| CN109726283A (en) * | 2018-12-03 | 2019-05-07 | 国家电网有限公司客户服务中心 | A kind of electric service client's demand recognition methods based on text similarity measurement |
| CN109740155A (en) * | 2018-12-27 | 2019-05-10 | 广州云趣信息科技有限公司 | A kind of customer service system artificial intelligence quality inspection rule self concludes the method and system of model |
| CN109783637A (en) * | 2018-12-12 | 2019-05-21 | 国网浙江省电力有限公司杭州供电公司 | Electric power overhaul text mining method based on deep neural network |
| CN109872162A (en) * | 2018-11-21 | 2019-06-11 | 阿里巴巴集团控股有限公司 | A kind of air control classifying identification method and system handling customer complaint information |
| CN110020777A (en) * | 2019-02-21 | 2019-07-16 | 国网山东省电力公司临沂供电公司 | A kind of power customer business worksheet system and method |
| CN110032643A (en) * | 2019-04-02 | 2019-07-19 | 上海建工四建集团有限公司 | A kind of building maintenance work order analysis method, device, storage medium and client |
| CN110163432A (en) * | 2019-05-17 | 2019-08-23 | 北京三快在线科技有限公司 | A kind of method and device of order identification |
| CN110245959A (en) * | 2019-04-17 | 2019-09-17 | 阿里巴巴集团控股有限公司 | The treating method and apparatus of specific aim request |
| CN110414819A (en) * | 2019-07-19 | 2019-11-05 | 中国电信集团工会上海市委员会 | A kind of work order methods of marking |
| CN110705245A (en) * | 2018-07-09 | 2020-01-17 | 中国移动通信集团有限公司 | Method and device for acquiring reference processing scheme and storage medium |
| CN110929036A (en) * | 2019-11-29 | 2020-03-27 | 南方电网数字电网研究院有限公司 | Electric power marketing inspection management method and device, computer equipment and storage medium |
| CN111144113A (en) * | 2019-12-31 | 2020-05-12 | 安徽智恒信科技股份有限公司 | Capability model and work order matching method and system based on machine learning |
| CN111161094A (en) * | 2019-12-12 | 2020-05-15 | 国网浙江省电力有限公司 | Electric power work order demand point identification method based on deep learning |
| CN111191529A (en) * | 2019-12-17 | 2020-05-22 | 中移（杭州）信息技术有限公司 | Method and system for processing abnormal work order |
| CN111340323A (en) * | 2018-12-19 | 2020-06-26 | 中国移动通信集团湖南有限公司 | Automatic complaint service request sending method and system |
| CN111708868A (en) * | 2020-01-15 | 2020-09-25 | 国网浙江省电力有限公司杭州供电公司 | Text classification method, device and equipment for electric power operation and inspection events |
| CN111931511A (en) * | 2019-04-26 | 2020-11-13 | 中国电力科学研究院有限公司 | Semantic analysis method and system based on wide-area distributed architecture |
| CN112633904A (en) * | 2020-12-30 | 2021-04-09 | 中国平安财产保险股份有限公司 | Complaint behavior analysis method, complaint behavior analysis device, complaint behavior analysis equipment and computer-readable storage medium |
| CN112685555A (en) * | 2019-10-17 | 2021-04-20 | 中国移动通信集团浙江有限公司 | Complaint work order quality detection method and device |
| CN112711947A (en) * | 2021-01-09 | 2021-04-27 | 国网湖北省电力有限公司电力科学研究院 | Text vectorization-based handling reference method in fault power failure repair work |
| CN113064962A (en) * | 2021-03-16 | 2021-07-02 | 北京工业大学 | Environment complaint reporting event similarity analysis method |
| CN114066483A (en) * | 2021-11-15 | 2022-02-18 | 国家电网有限公司客户服务中心 | Suspected information collecting client identification method, system, equipment and medium |
| CN115018472A (en) * | 2022-08-03 | 2022-09-06 | 中国电子科技集团公司第五十四研究所 | Interactive incremental information analysis system based on interpretable mechanism |
| CN117172508A (en) * | 2023-10-31 | 2023-12-05 | 无锡容智技术有限公司 | Automatic dispatch method and system based on city complaint worksheet recognition |
| CN117725214A (en) * | 2023-12-05 | 2024-03-19 | 广州市白云区政务服务数据管理局 | Method, device, equipment and storage medium for identifying problematic work orders |
| CN117743956A (en) * | 2024-01-05 | 2024-03-22 | 北京数字政通科技股份有限公司 | Method and system for carrying out hotline early warning by intelligent tag algorithm |
| CN111931511B (en) * | 2019-04-26 | 2024-11-08 | 中国电力科学研究院有限公司 | Semantic analysis method and system based on wide area distributed architecture |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6212532B1 (en) * | 1998-10-22 | 2001-04-03 | International Business Machines Corporation | Text categorization toolkit |
| CN103617157A (en) * | 2013-12-10 | 2014-03-05 | 东北师范大学 | Text similarity calculation method based on semantics |
| CN105335496A (en) * | 2015-10-22 | 2016-02-17 | 国网山东省电力公司电力科学研究院 | Customer service repeated call treatment method based on cosine similarity text mining algorithm |
| CN105760493A (en) * | 2016-02-18 | 2016-07-13 | 国网江苏省电力公司电力科学研究院 | Automatic work order classification method for electricity marketing service hot spot 95598 |
| CN106503101A (en) * | 2016-10-14 | 2017-03-15 | 五邑大学 | Electric business customer service automatically request-answering system sentence keyword extracting method |
-
2017
- 2017-10-11 CN CN201710938687.7A patent/CN107861942B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6212532B1 (en) * | 1998-10-22 | 2001-04-03 | International Business Machines Corporation | Text categorization toolkit |
| CN103617157A (en) * | 2013-12-10 | 2014-03-05 | 东北师范大学 | Text similarity calculation method based on semantics |
| CN105335496A (en) * | 2015-10-22 | 2016-02-17 | 国网山东省电力公司电力科学研究院 | Customer service repeated call treatment method based on cosine similarity text mining algorithm |
| CN105760493A (en) * | 2016-02-18 | 2016-07-13 | 国网江苏省电力公司电力科学研究院 | Automatic work order classification method for electricity marketing service hot spot 95598 |
| CN106503101A (en) * | 2016-10-14 | 2017-03-15 | 五邑大学 | Electric business customer service automatically request-answering system sentence keyword extracting method |
Non-Patent Citations (3)
| Title |
|---|
| 周文杰 等: "基于深度学习的用户投诉预测模型研究", 《计算机应用研究》 * |
| 李济汉 等: "面向电信客户投诉和建议的智能分析模型", 《现代电信科技》 * |
| 王琴明 等: "供电企业建立投诉管理体系的探索与实践", 《电力需求侧管理》 * |
Cited By (53)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108710651A (en) * | 2018-05-08 | 2018-10-26 | 华南理工大学 | A kind of large scale customer complaint data automatic classification method |
| CN110705245B (en) * | 2018-07-09 | 2023-04-28 | 中移(成都)信息通信科技有限公司 | Method and device for acquiring reference processing scheme and storage medium |
| CN110705245A (en) * | 2018-07-09 | 2020-01-17 | 中国移动通信集团有限公司 | Method and device for acquiring reference processing scheme and storage medium |
| CN109389418A (en) * | 2018-08-17 | 2019-02-26 | 国家电网有限公司客户服务中心 | Electric service client's demand recognition methods based on LDA model |
| CN109492660A (en) * | 2018-09-26 | 2019-03-19 | 平安科技（深圳）有限公司 | Calling information processing method, device, computer equipment and storage medium |
| CN109492660B (en) * | 2018-09-26 | 2024-07-02 | 平安科技（深圳）有限公司 | Complaint information processing method, apparatus, computer device and storage medium |
| CN109190716A (en) * | 2018-10-23 | 2019-01-11 | 深圳增强现实技术有限公司 | Processing method, device and the electronic equipment of low-voltage collecting meter reading failure |
| CN109409553A (en) * | 2018-10-29 | 2019-03-01 | 国网河北省电力有限公司电力科学研究院 | A kind of power marketing business application system |
| CN109558486A (en) * | 2018-10-30 | 2019-04-02 | 国家电网有限公司客户服务中心 | Electric power customer service client's demand intelligent identification Method |
| CN109872162A (en) * | 2018-11-21 | 2019-06-11 | 阿里巴巴集团控股有限公司 | A kind of air control classifying identification method and system handling customer complaint information |
| CN109726283A (en) * | 2018-12-03 | 2019-05-07 | 国家电网有限公司客户服务中心 | A kind of electric service client's demand recognition methods based on text similarity measurement |
| CN109635292B (en) * | 2018-12-05 | 2023-07-28 | 杭州东方通信软件技术有限公司 | Work order quality inspection method and device based on machine learning algorithm |
| CN109635292A (en) * | 2018-12-05 | 2019-04-16 | 杭州东方通信软件技术有限公司 | Work order quality detecting method and device based on machine learning algorithm |
| CN109783637A (en) * | 2018-12-12 | 2019-05-21 | 国网浙江省电力有限公司杭州供电公司 | Electric power overhaul text mining method based on deep neural network |
| CN111340323A (en) * | 2018-12-19 | 2020-06-26 | 中国移动通信集团湖南有限公司 | Automatic complaint service request sending method and system |
| CN111340323B (en) * | 2018-12-19 | 2023-09-05 | 中国移动通信集团湖南有限公司 | Automatic dispatch method and system for complaint service request |
| CN109740155A (en) * | 2018-12-27 | 2019-05-10 | 广州云趣信息科技有限公司 | A kind of customer service system artificial intelligence quality inspection rule self concludes the method and system of model |
| CN109710766B (en) * | 2018-12-29 | 2023-01-20 | 云南电网有限责任公司电力科学研究院 | Complaint tendency analysis early warning method and device for work order data |
| CN109726290B (en) * | 2018-12-29 | 2020-12-22 | 咪咕数字传媒有限公司 | Complaint classification model determination method and device and computer-readable storage medium |
| CN109726290A (en) * | 2018-12-29 | 2019-05-07 | 咪咕数字传媒有限公司 | Complaint classification model determination method and device and computer-readable storage medium |
| CN109710766A (en) * | 2018-12-29 | 2019-05-03 | 云南电网有限责任公司电力科学研究院 | A kind of the complaint trend analysis method for early warning and device of work order data |
| CN110020777A (en) * | 2019-02-21 | 2019-07-16 | 国网山东省电力公司临沂供电公司 | A kind of power customer business worksheet system and method |
| CN110032643B (en) * | 2019-04-02 | 2021-04-27 | 上海建工四建集团有限公司 | Building maintenance work order analysis method and device, storage medium and client |
| CN110032643A (en) * | 2019-04-02 | 2019-07-19 | 上海建工四建集团有限公司 | A kind of building maintenance work order analysis method, device, storage medium and client |
| CN110245959A (en) * | 2019-04-17 | 2019-09-17 | 阿里巴巴集团控股有限公司 | The treating method and apparatus of specific aim request |
| CN111931511A (en) * | 2019-04-26 | 2020-11-13 | 中国电力科学研究院有限公司 | Semantic analysis method and system based on wide-area distributed architecture |
| CN111931511B (en) * | 2019-04-26 | 2024-11-08 | 中国电力科学研究院有限公司 | Semantic analysis method and system based on wide area distributed architecture |
| CN110163432A (en) * | 2019-05-17 | 2019-08-23 | 北京三快在线科技有限公司 | A kind of method and device of order identification |
| CN110414819A (en) * | 2019-07-19 | 2019-11-05 | 中国电信集团工会上海市委员会 | A kind of work order methods of marking |
| CN110414819B (en) * | 2019-07-19 | 2023-05-26 | 中国电信集团工会上海市委员会 | Work order scoring method |
| CN112685555A (en) * | 2019-10-17 | 2021-04-20 | 中国移动通信集团浙江有限公司 | Complaint work order quality detection method and device |
| CN112685555B (en) * | 2019-10-17 | 2023-04-18 | 中国移动通信集团浙江有限公司 | Complaint work order quality detection method and device |
| CN110929036A (en) * | 2019-11-29 | 2020-03-27 | 南方电网数字电网研究院有限公司 | Electric power marketing inspection management method and device, computer equipment and storage medium |
| CN110929036B (en) * | 2019-11-29 | 2023-05-05 | 南方电网数字电网研究院有限公司 | Electric power marketing inspection management method, electric power marketing inspection management device, computer equipment and storage medium |
| CN111161094A (en) * | 2019-12-12 | 2020-05-15 | 国网浙江省电力有限公司 | Electric power work order demand point identification method based on deep learning |
| CN111191529B (en) * | 2019-12-17 | 2023-04-28 | 中移（杭州）信息技术有限公司 | Method and system for processing abnormal worksheets |
| CN111191529A (en) * | 2019-12-17 | 2020-05-22 | 中移（杭州）信息技术有限公司 | Method and system for processing abnormal work order |
| CN111144113B (en) * | 2019-12-31 | 2024-02-06 | 安徽智恒信科技股份有限公司 | Method and system for matching capability model with work order based on machine learning |
| CN111144113A (en) * | 2019-12-31 | 2020-05-12 | 安徽智恒信科技股份有限公司 | Capability model and work order matching method and system based on machine learning |
| CN111708868A (en) * | 2020-01-15 | 2020-09-25 | 国网浙江省电力有限公司杭州供电公司 | Text classification method, device and equipment for electric power operation and inspection events |
| CN112633904A (en) * | 2020-12-30 | 2021-04-09 | 中国平安财产保险股份有限公司 | Complaint behavior analysis method, complaint behavior analysis device, complaint behavior analysis equipment and computer-readable storage medium |
| CN112633904B (en) * | 2020-12-30 | 2024-04-30 | 中国平安财产保险股份有限公司 | Complaint behavior analysis method, apparatus, device and computer readable storage medium |
| CN112711947A (en) * | 2021-01-09 | 2021-04-27 | 国网湖北省电力有限公司电力科学研究院 | Text vectorization-based handling reference method in fault power failure repair work |
| CN112711947B (en) * | 2021-01-09 | 2023-08-22 | 国网湖北省电力有限公司电力科学研究院 | Text vectorization-based fault power failure emergency repair handling reference method |
| CN113064962B (en) * | 2021-03-16 | 2024-03-15 | 北京工业大学 | Environment complaint reporting event similarity analysis method |
| CN113064962A (en) * | 2021-03-16 | 2021-07-02 | 北京工业大学 | Environment complaint reporting event similarity analysis method |
| CN114066483A (en) * | 2021-11-15 | 2022-02-18 | 国家电网有限公司客户服务中心 | Suspected information collecting client identification method, system, equipment and medium |
| CN115018472B (en) * | 2022-08-03 | 2022-11-11 | 中国电子科技集团公司第五十四研究所 | Interactive incremental information analysis system based on interpretable mechanism |
| CN115018472A (en) * | 2022-08-03 | 2022-09-06 | 中国电子科技集团公司第五十四研究所 | Interactive incremental information analysis system based on interpretable mechanism |
| CN117172508B (en) * | 2023-10-31 | 2024-02-27 | 无锡容智技术有限公司 | Automatic dispatch method and system based on city complaint worksheet recognition |
| CN117172508A (en) * | 2023-10-31 | 2023-12-05 | 无锡容智技术有限公司 | Automatic dispatch method and system based on city complaint worksheet recognition |
| CN117725214A (en) * | 2023-12-05 | 2024-03-19 | 广州市白云区政务服务数据管理局 | Method, device, equipment and storage medium for identifying problematic work orders |
| CN117743956A (en) * | 2024-01-05 | 2024-03-22 | 北京数字政通科技股份有限公司 | Method and system for carrying out hotline early warning by intelligent tag algorithm |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107861942B (en) | 2021-10-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107861942A (en) | A kind of electric power based on deep learning is doubtful to complain work order recognition methods | |
| Singh et al. | Investigation and modeling of lean six sigma barriers in small and medium-sized industries using hybrid ISM-SEM approach | |
| CN105930347B (en) | Text analysis based power outage cause recognition system | |
| CN106529804A (en) | Client complaint early-warning monitoring analyzing method based on text mining technology | |
| CN110175324B (en) | Power grid operation instruction verification method and system based on data mining | |
| CN112541600A (en) | Knowledge graph-based auxiliary maintenance decision method | |
| EP3038025A1 (en) | Retention risk determiner | |
| CN104331502B (en) | The recognition methods of courier's data in being marketed for courier periphery crowd | |
| CN106874134A (en) | The processing method of work order type, apparatus and system | |
| CN106599160A (en) | Content rule base management system and encoding method thereof | |
| CN108985467A (en) | Secondary device lean management-control method based on artificial intelligence | |
| CN109740838A (en) | Provider service evaluation method and relevant device based on big data | |
| CN107194617A (en) | A kind of app software engineers soft skill categorizing system and method | |
| CN114218402B (en) | Method for recommending computer hardware fault replacement parts | |
| CN107730158A (en) | Power network project intelligent evaluation method and system | |
| CN107239798A (en) | A kind of feature selection approach of software-oriented defect number prediction | |
| CN109344227A (en) | Worksheet method, system and electronic equipment | |
| CN116861924A (en) | Project risk early warning method and system based on artificial intelligence | |
| CN113868422A (en) | Multi-label inspection work order problem traceability identification method and device | |
| CN115358481A (en) | Early warning and identification method, system and device for enterprise ex-situ migration | |
| CN108154276A (en) | Attendance analysis method based on image identification and machine learning | |
| CN115660464A (en) | Intelligent equipment maintenance method and terminal based on big data and physical ID | |
| CN115619117A (en) | Power grid intelligent scheduling method based on duty system | |
| CN107229732A (en) | A kind of fault data information processing method and device | |
| CN113378024A (en) | Deep learning-based public inspection field-oriented related event identification method |
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 | ||
| CB02 | Change of applicant information | ||
| CB02 | Change of applicant information |
Address after: The eight district of Hangzhou city in Zhejiang province 310014 Huadian Zhaohui under No. 1 Lane Applicant after: ELECTRIC POWER SCIENTIFIC RESEARCH INSTITUTE OF STATE GRID ZHEJIANG ELECTRIC POWER Co.,Ltd. Applicant after: ZHEJIANG HUAYUN INFORMATION TECHNOLOGY Co.,Ltd. Address before: The eight district of Hangzhou city in Zhejiang province 310014 Huadian Zhaohui under No. 1 Lane Applicant before: ELECTRIC POWER RESEARCH INSTITUTE OF STATE GRID ZHEJIANG ELECTRIC POWER Co. Applicant before: ZHEJIANG HUAYUN INFORMATION TECHNOLOGY Co.,Ltd. |
|
| TA01 | Transfer of patent application right | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20210319 Address after: 311100 Building 5, 138 Yunlian Road, Yuhang District, Hangzhou City, Zhejiang Province Applicant after: Marketing service center of State Grid Zhejiang Electric Power Co., Ltd Applicant after: ZHEJIANG HUAYUN INFORMATION TECHNOLOGY Co.,Ltd. Address before: The eight district of Hangzhou city in Zhejiang province 310014 Huadian Zhaohui under No. 1 Lane Applicant before: ELECTRIC POWER SCIENTIFIC RESEARCH INSTITUTE OF STATE GRID ZHEJIANG ELECTRIC POWER Co.,Ltd. Applicant before: ZHEJIANG HUAYUN INFORMATION TECHNOLOGY Co.,Ltd. |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant |