CN114417045A - Insurance case spot inspection method, system, equipment and storage medium based on neural network - Google Patents
Insurance case spot inspection method, system, equipment and storage medium based on neural network Download PDFInfo
- Publication number
- CN114417045A CN114417045A CN202210091571.5A CN202210091571A CN114417045A CN 114417045 A CN114417045 A CN 114417045A CN 202210091571 A CN202210091571 A CN 202210091571A CN 114417045 A CN114417045 A CN 114417045A
- Authority
- CN
- China
- Prior art keywords
- case
- insurance
- neural network
- spot
- preset
- 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
- 238000000034 method Methods 0.000 title claims abstract description 57
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 50
- 238000007689 inspection Methods 0.000 title claims abstract description 42
- 230000004927 fusion Effects 0.000 claims abstract description 55
- 238000003062 neural network model Methods 0.000 claims description 52
- 238000005070 sampling Methods 0.000 claims description 24
- 238000004590 computer program Methods 0.000 claims description 17
- 238000012549 training Methods 0.000 claims description 15
- 238000004364 calculation method Methods 0.000 claims description 10
- 238000000605 extraction Methods 0.000 claims description 6
- 238000001914 filtration Methods 0.000 claims description 6
- 238000013473 artificial intelligence Methods 0.000 abstract description 10
- 230000004069 differentiation Effects 0.000 abstract description 4
- 230000008569 process Effects 0.000 description 12
- 239000013598 vector Substances 0.000 description 12
- 238000005516 engineering process Methods 0.000 description 8
- 230000006870 function Effects 0.000 description 8
- 210000002569 neuron Anatomy 0.000 description 7
- 238000004422 calculation algorithm Methods 0.000 description 6
- 238000010801 machine learning Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 230000006872 improvement Effects 0.000 description 4
- 238000005457 optimization Methods 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 230000000295 complement effect Effects 0.000 description 3
- 239000000284 extract Substances 0.000 description 3
- 238000012216 screening Methods 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 2
- 239000003814 drug Substances 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 238000007639 printing Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000004083 survival effect Effects 0.000 description 1
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
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
- G06F16/5846—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content using extracted text
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/5866—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using information manually generated, e.g. tags, keywords, comments, manually generated location and time information
-
- 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
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- 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
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/08—Insurance
-
- 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
- G06Q50/10—Services
- G06Q50/18—Legal services
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Business, Economics & Management (AREA)
- General Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Computational Linguistics (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Library & Information Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Business, Economics & Management (AREA)
- Databases & Information Systems (AREA)
- Marketing (AREA)
- Software Systems (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Tourism & Hospitality (AREA)
- Technology Law (AREA)
- Mathematical Physics (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Development Economics (AREA)
- Human Resources & Organizations (AREA)
- Primary Health Care (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
The invention relates to the field of artificial intelligence, and provides a neural network-based insurance case spot-check method, which comprises the following steps: acquiring preset keyword information of the insurance case to be sampled and inspected; extracting the word frequency characteristic, the word position characteristic and the context semantic characteristic in the preset keyword information respectively, and fusing the extracted characteristics to obtain fusion characteristics; acquiring a plurality of reference insurance history cases with the highest similarity according to the similarity between the fusion features and the fusion features corresponding to each insurance history case in a preset insurance case library; and acquiring the selective examination result of the case to be subjected to the selective examination according to the claim settlement condition of each reference insurance history case. The embodiment of the invention can effectively reduce the labor cost, and the intelligent spot inspection has certain objectivity, so that the quality result differentiation of the spot inspection caused by the professional skill and medical background difference of personnel can be avoided.
Description
Technical Field
The invention relates to the technical field of artificial intelligence, in particular to a neural network-based insurance case spot-checking method, system, equipment and storage medium.
Background
Insurance companies in the market all have the requirements of carrying out spot check and reexamination on the claim settling cases of the insurance companies, and the insurance companies are obligated and responsible for carrying out spot check and quality check on the claim settling cases no matter whether the insurance cases are automatically audited to settle the case or whether operators operate the audited to settle the case.
However, in the current insurance industry, most of the solutions adopted by insurance companies are to invest personnel in a certain proportion of the case quantity, manually perform a spot check on the case and check the auditing result data, if the case is found to have a problem after the manual spot check, the case is reprocessed after the complaint is initiated, and the solutions adopted by most of the insurance companies have the following defects:
because the manual selective examination needs to put in human resources in a certain proportion according to the amount of the cases, the human input amount is large, the manual selective examination is easy to bring personal subjectivity, the mode of the manual selective examination of the cases is also easy to be different from person to person, and the cases which are possibly problematic are easily omitted due to the fact that the manual selective examination is carried out according to personal experience.
Disclosure of Invention
The invention provides a neural network-based insurance case spot check method, system, equipment and storage medium, and mainly aims to realize automatic spot check of insurance cases and effectively improve the efficiency of the insurance case spot check.
In a first aspect, an embodiment of the present invention provides a neural network-based insurance case spot-check method, including:
acquiring preset keyword information of the insurance case to be sampled and inspected;
extracting word frequency characteristics, word position characteristics and context semantic characteristics in the preset keyword information respectively based on a fusion unit in the case spot check neural network model, and fusing the extracted characteristics to obtain fusion characteristics;
based on a similarity calculation unit in the case spot check neural network model, acquiring a plurality of reference insurance history cases with highest similarity according to the similarity between the fusion features and the fusion features corresponding to each insurance history case in a preset insurance case library;
and acquiring the selective examination result of the case to be subjected to selective examination according to the claim settlement condition of each reference insurance history case based on a prediction unit in the case selective examination neural network model, wherein the case selective examination neural network model is obtained by training samples and labels.
Preferably, the obtaining a plurality of reference insurance history cases with the highest similarity according to the similarity between the fusion feature and the fusion feature corresponding to each insurance history case in the preset insurance case library includes:
calculating the cosine similarity between the fusion features and the fusion features corresponding to each insurance history case in the preset insurance case library;
and taking a plurality of insurance history cases with the maximum cosine similarity as the reference insurance history cases.
Preferably, the extracting the word frequency feature, the word position feature and the context semantic feature in the preset keyword information respectively includes:
extracting word frequency-inverse document characteristics from the preset keyword information to obtain the word frequency characteristics;
and performing word2vec feature extraction on the preset keyword information to obtain word position features and the context semantic features in the text.
Preferably, the acquiring preset keyword information of the insurance case to be spot-checked includes:
acquiring a picture corresponding to the insurance case recording document to be spot-checked;
extracting representation information corresponding to preset keywords in the picture;
and acquiring the preset keyword information according to the representation information corresponding to the preset keyword.
Preferably, the extracting of the representation information corresponding to the preset keyword in the picture further includes, before:
and filtering the picture to remove invalid information in the picture.
Preferably, the method further comprises the following steps:
and taking the preset keyword information of the insurance case to be subjected to the spot check as a sample, and taking the spot check result of the insurance case to be subjected to the spot check as a label so as to train the case spot check neural network model again.
Preferably, the method further comprises the following steps:
and if the selective examination result is negative, sending a reminding short message to a preset mobile phone number associated with the selective examination person, and/or sending a reminding mail to a preset mailbox account associated with the selective examination person.
In a second aspect, an embodiment of the present invention provides a neural network-based insurance case spot-check system, including:
the keyword module is used for acquiring preset keyword information of the insurance case to be subjected to spot check;
the fusion module is used for extracting the word frequency characteristics, the word position characteristics and the context semantic characteristics in the preset keyword information respectively based on a fusion unit in the case spot check neural network model, and fusing the extracted characteristics to obtain fusion characteristics;
the similarity module is used for acquiring a plurality of reference insurance history cases with the highest similarity according to the similarity between the fusion characteristics and the fusion characteristics corresponding to each insurance history case in a preset insurance case library based on a similarity calculation unit in the case spot check neural network model;
and the sampling inspection module is used for acquiring sampling inspection results of the cases to be sampled and inspected according to the claim settlement condition of each reference insurance history case based on a prediction unit in the case sampling inspection neural network model, wherein the case sampling inspection neural network model is obtained by training samples and labels.
In a third aspect, an embodiment of the present invention provides a computer device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, where the processor implements the steps of the neural network based insurance case spot check method when executing the computer program.
In a fourth aspect, an embodiment of the present invention provides a computer storage medium, where a computer program is stored, and the computer program, when executed by a processor, implements the steps of the method for sampling and checking a safety case based on a neural network.
According to the method, the system, the equipment and the storage medium for the selective inspection of the insurance cases based on the neural network, artificial intelligence is applied to the field of the selective inspection of the insurance cases, the preset keyword information of the cases to be subjected to the selective inspection is firstly extracted, the cases to be subjected to the selective inspection are quantized through the preset keyword information, and the description of the cases to be subjected to the selective inspection can be simplified; then inputting preset keyword information of a case to be subjected to spot check into a case spot check neural network model, extracting word frequency characteristics, word position characteristics in a document and context characteristics from the preset keyword information, and then fusing the characteristics, wherein a plurality of simple characteristics are fused instead of simply connected in series, and if the plurality of simple characteristics are connected in series, although certain performance improvement can be obtained, the dimensionality of the characteristics can be greatly increased, the problem of dimensionality disaster is caused, and the complexity of putting in the insurance case to be subjected to spot check is increased; the case selective examination neural network model is obtained after a sample is trained, and the selective examination result of the case to be subjected to selective examination can be directly predicted, so that the manual selective examination work is avoided. The embodiment of the invention can effectively reduce the labor cost, and the case quantity needing to be checked is not increased due to the increase of the case quantity, so that a certain proportion of labor needs to be synchronously increased, and the cost can be effectively reduced; and the intelligent spot check has certain objectivity, and the quality result differentiation of the spot check cannot be caused by the professional skill of personnel and the medical background difference.
Drawings
Fig. 1 is a schematic view of an application scenario of a neural network based insurance case spot inspection method according to an embodiment of the present invention;
FIG. 2 is a flowchart of a neural network-based insurance case spot check method according to an embodiment of the present invention;
fig. 3 is a schematic structural diagram of a neural network based insurance case spot inspection system according to an embodiment of the present invention;
fig. 4 is a schematic structural diagram of a computer device according to an embodiment of the present invention.
The implementation, functional features and advantages of the objects of the present invention will be further explained with reference to the accompanying drawings.
Detailed Description
It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Fig. 1 is a schematic view of an application scenario of a neural network based insurance case spot check method according to an embodiment of the present invention, as shown in fig. 1, a user inputs a record document of an insurance case to be spot checked at a client, the client sends the insurance case to be spot checked to a server after obtaining the record document of the insurance case to be spot checked, and the server executes the neural network based insurance case spot check method after receiving the insurance case to be spot checked, so as to obtain a spot check result of the insurance case, and returns the spot check result to the client for the user to refer to.
It should be noted that the server may be implemented by an independent server or a server cluster composed of a plurality of servers. The client may be, but is not limited to, a smart phone, a tablet computer, a notebook computer, a desktop computer, and the like. The client and the server may be connected through bluetooth, USB (Universal Serial Bus), or other communication connection manners, which is not limited in this embodiment of the present invention.
The embodiment of the invention can acquire and process related data based on an artificial intelligence technology. Artificial Intelligence (AI) is a theory, method, technique and application system that simulates, extends and expands human Intelligence using a digital computer or a machine controlled by a digital computer, senses the environment, acquires knowledge and uses the knowledge to obtain the best result.
The artificial intelligence infrastructure generally includes technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technologies, operation/interaction systems, mechatronics, and the like. The artificial intelligence software technology mainly comprises a computer vision technology, a robot technology, a biological recognition technology, a voice processing technology, a natural language processing technology, machine learning, deep learning and the like.
Fig. 2 is a flowchart of a neural network based insurance case spot check method according to an embodiment of the present invention, and as shown in fig. 2, the method includes:
the embodiment of the invention is applied to the field of insurance, is used for carrying out selective examination on the insurance company insurance condition, quantifies and represents the selective examination insurance case by taking the existing selective examination insurance case of the existing insurance company as a training sample and taking the selective examination result as a label through a large amount of machine learning so as to facilitate the machine learning, and realizes the intelligent selective examination of the insurance case through deep learning of a large amount of claim settlement data.
S210, acquiring preset keyword information of the insurance case to be subjected to spot inspection;
the method comprises the steps of firstly obtaining preset keyword information of the insurance case to be subjected to spot inspection, wherein the preset keyword information can also be regarded as representing characteristics of the insurance case to be subjected to spot inspection and used for representing the insurance case to be subjected to spot inspection, representing and describing the insurance case to be subjected to spot inspection by utilizing some preset keyword information, describing the insurance case through documents in general insurance companies, extracting keywords from original documents due to the fact that the content of the original documents is complex and has a bit of redundancy, describing the insurance case to be subjected to spot inspection through the preset keywords, simplifying the description of the insurance case to be subjected to spot inspection, and unifying the description information and description format of each insurance case. The preset keywords may be the same or different in different companies or different applications, so that the representation features in different companies or different applications may also be the same or different, which may be determined specifically according to the actual situation, and the specific content of the representation features in the embodiment of the present invention is not specifically limited.
For example, the representation characteristics in the embodiment of the invention comprise relevant information of the insured person, relevant information of medical treatment, relevant information of claim paying and auxiliary information, wherein the relevant information of the insured person comprises sex, age, type of medical insurance, location of medical insurance, time of occurrence, location of occurrence, hospital of medical treatment, survival status, date of occurrence, date of identification of disabilities and date of accurate diagnosis of serious diseases. The information related to the visit includes billing information, hospital information, expense information, amount information, medical insurance information and treatment information, the billing information includes the person, the time and date of the visit, the hospital information includes the name of the hospital, the type of the hospital (a special hospital or a comprehensive hospital), the hospital grade and the nature of the hospital (a private hospital or a public hospital), the expense information includes expense items (such as treatment fee, examination fee, western medicine fee and the like), expense price, measurement amount and unit, and medicine detail, the amount information includes the total amount of the bill, self-payment amount, medical insurance payment amount, cash payment amount and personal account payment amount, the treatment information includes the type of treatment (such as general outpatient service, in-patient operation, specialist outpatient service and special-need outpatient service), and the auxiliary information includes medical record number, social security card number and inpatient number. The claim related information includes product category of the claim (medical insurance, severe insurance, life insurance, etc.), product validity period (long term, one year period, very short term), insurance payment mode (wholesale, term payment), first insurance or renewal, claim risk type terms, responsibility, claim amount, claim mode (normal claim, agreement claim, partial claim, general claim, rejection), exemption claim for use of the claim, limit, guarantee amount and claim description. And the information forms preset keyword information of the case to be sampled.
S220, extracting word frequency characteristics, word position characteristics and context semantic characteristics in the preset keyword information respectively based on a fusion unit in the case spot check neural network model, and fusing the extracted characteristics to obtain fusion characteristics;
the extracted preset keyword information is input into a case spot check neural network model, in the embodiment of the invention, the case spot check neural network model is formed by sequentially connecting a fusion unit, a similarity calculation unit and a prediction unit end to end, the preset keyword information is specifically input into the fusion unit of the case spot check neural network model, the word frequency characteristic, the word position characteristic and the context semantic characteristic in the preset keyword information are respectively extracted, and the extracted characteristics are fused to obtain the fusion characteristics.
The word frequency characteristics, the word position characteristics and the context characteristics in the document are extracted from the preset keyword information, and then the characteristics are fused, wherein a plurality of simple characteristics are fused instead of simply connected in series, and if the plurality of simple characteristics are connected in series, although certain performance improvement can be achieved, the dimensionality of the characteristics can be greatly increased, the problem of dimensionality disaster is caused, and therefore the complexity of putting the insurance case is increased. In the embodiment of the invention, a plurality of simple features are effectively fused, so that the dimensionality of the fused features is reduced, and the complementary characteristics of different modal features of the insurance case are fully utilized.
S230, based on a similarity calculation unit in the case spot check neural network model, acquiring a plurality of reference insurance history cases with highest similarity according to the similarity between the fusion features and the fusion features corresponding to each insurance history case in a preset insurance case library;
and then calculating the similarity between the fusion characteristics and the fusion characteristics corresponding to each insurance history case in a preset insurance case library, and screening out one or more insurance history cases with the highest similarity.
And then, calculating the similarity between the insurance cases to be sampled and inspected and the fusion characteristics of each insurance history case, and extracting a plurality of insurance history cases with the highest similarity to serve as reference insurance history cases.
S240, based on the prediction unit in the case selective examination neural network model, obtaining the selective examination result of the case to be subjected to selective examination according to the claim settlement condition of each reference insurance history case, wherein the case selective examination neural network model is obtained by training samples and labels.
And inputting the extracted reference insurance history cases with higher similarity into a prediction unit of the case selective examination neural network model, and obtaining the selective examination result of the case to be selectively examined by combining the claim settlement condition of each reference insurance history case.
In the embodiment of the invention, the case selective examination neural network model is a machine learning model, and before the case selective examination neural network model is used, a sample and a label are required to be trained.
Specifically, the sampling result in different companies or different applications may be the same or different, so that the sampling result in different companies or different applications may also be the same or different, which may be determined specifically according to the actual situation, and the specific content of the sampling result in the embodiment of the present invention is not specifically limited.
For example, the spot check result in the embodiment of the present invention includes decision information and claim settlement adjustment information, where the decision information is "pass" or "deny" by indicating that the original claim settlement scheme is agreed, and "deny" the original claim settlement scheme is indicated, and the claim settlement adjustment information includes risk terms, responsibilities, a claim amount, a claim payment manner (normal claim payment, agreement claim payment, partial claim payment, converged claim payment, claim denial) of the adjusted claim payment, a claim exemption amount used for the claim payment, a limit, a guarantee amount, and a claim payment description.
And obtaining the trained case selective examination neural network model, and inputting the representation characteristics of the insurance case to be subjected to selective examination into the case selective examination neural network model to obtain the selective examination result of the insurance case to be subjected to selective examination.
The invention provides a neural network-based insurance case spot check method, which applies artificial intelligence in the field of insurance case spot check, firstly extracts preset keyword information of a case to be spot checked, quantizes the case to be spot checked through the preset keyword information, and can simplify the description of the case to be spot checked; then inputting preset keyword information of a case to be subjected to spot check into a case spot check neural network model, extracting word frequency characteristics, word position characteristics in a document and context characteristics from the preset keyword information, and then fusing the characteristics, wherein a plurality of simple characteristics are fused instead of simply connected in series, and if the plurality of simple characteristics are connected in series, although certain performance improvement can be obtained, the dimensionality of the characteristics can be greatly increased, the problem of dimensionality disaster is caused, and the complexity of putting in the insurance case to be subjected to spot check is increased; the case selective examination neural network model is obtained after a sample is trained, and the selective examination result of the case to be subjected to selective examination can be directly predicted, so that the manual selective examination work is avoided. The embodiment of the invention can effectively reduce the labor cost, and the case quantity needing to be checked is not increased due to the increase of the case quantity, so that a certain proportion of labor needs to be synchronously increased, and the cost can be effectively reduced; and the intelligent spot check has certain objectivity, and the quality result differentiation of the spot check cannot be caused by the professional skill of personnel and the medical background difference.
On the basis of the foregoing embodiment, preferably, the obtaining, according to the similarity between the fusion feature and the fusion feature corresponding to each reference insurance history case in the preset insurance case library, a plurality of insurance history cases with the highest similarity includes:
calculating the cosine similarity between the fusion features and the fusion features corresponding to each insurance history case in the preset insurance case library;
and taking a plurality of insurance history cases with the maximum cosine similarity as the reference insurance history cases.
Specifically, the similarity in the embodiment of the invention refers to cosine similarity, and the reference insurance history case is screened out according to the cosine similarity by calculating the cosine similarity between the fusion feature and the fusion feature corresponding to each insurance history case in the preset insurance case library.
On the basis of the foregoing embodiment, preferably, the extracting the word frequency feature, the word position feature, and the context semantic feature in the preset keyword information respectively includes:
extracting word frequency-inverse document characteristics from the preset keyword information to obtain the word frequency characteristics;
and performing word2vec feature extraction on the preset keyword information to obtain word position features and the context semantic features in the text.
In the embodiment of the invention, word Frequency characteristics are extracted from preset keyword information through a Term Frequency-Inverse Document (TF-IDF for short), word position characteristics and context semantic characteristics in a text are extracted through word2vec, specifically, word2vec is a tool for word vector calculation which is sourced from Google 2013, the tool is a model for learning semantic knowledge in an unsupervised learning mode from a large amount of semantics, and word2vec is learned through a large amount of corpus and represents semantic information of words, namely word vectors, in a vector mode. According to the method, semantically similar words are close to each other in the space through an embedding space, word2vec is an Estimator, a series of words representing documents are adopted to train a word2vec model, each word is mapped into a vector with a fixed size through the model, the word2vec skip-gram model is used, vector representations of all words in pre-processed committee document data are obtained firstly, then all word vectors belonging to one committee document are averaged, and the average value is used as a word2vec feature vector of the referee document.
TF-IDF is a commonly used weighting technique for information retrieval and data mining. TF is Term Frequency (Term Frequency) and IDF is Inverse text Frequency index (Inverse Document Frequency). TF-IDF is a feature weight algorithm widely used in text mining and based on word frequency, and can effectively evaluate the importance degree of feature words to a text set or a corpus. In the specific implementation process, a sklern library in Python is utilized to directly extract TF-IDF feature vectors of the referee document, and in the sklern library, the countVectorizer only considers the frequency of each vocabulary in the training text and simultaneously calculates the reciprocal of the number of the texts containing the vocabulary.
And then, fusing the word2vec word vector and the TF-IDF vector, thereby not only reducing the dimensionality of fused features, but also fully utilizing the complementary characteristics of different modal features of preset keyword information.
On the basis of the above embodiment, preferably, the acquiring preset keyword information of the insurance case to be spot-checked includes:
acquiring a picture corresponding to the insurance case recording document to be spot-checked;
extracting representation information corresponding to preset keywords in the picture;
and acquiring the preset keyword information according to the representation information corresponding to the preset keyword.
The insurance case recording document to be spot-checked is description information about the insurance case to be spot-checked, information such as insurance person information, policy information, insurance information, treatment information, claim settlement information and the like is recorded in the insurance case recording document to be spot-checked, the general insurance case recording document to be spot-checked is in a PDF format, a general system cannot directly process the PDF format, and the content on the insurance case recording document to be spot-checked needs to be recognized and extracted firstly. In the embodiment of the invention, the specific extraction mode is to identify the committee document by using the universal character identification software, the universal character identification software can provide identification of a universal printing form and identification of a universal handwriting form, and finally, the content in the insurance case recording document to be subjected to selective examination is extracted.
On the basis of the foregoing embodiment, preferably, the extracting the representation information corresponding to the preset keyword in the picture further includes:
and filtering the picture to remove invalid information in the picture.
In the specific implementation process, before the insurance case recording documents to be subjected to the selective examination are identified and extracted, preprocessing is required to be performed firstly, the structured data of the insurance case recording documents to be subjected to the selective examination are extracted, the whole insurance case recording documents to be subjected to the selective examination are divided into a title, a case number, a case type, a case order and the like according to elements, and the preprocessing mode adopted in the embodiment of the invention is to combine the insurance case recording documents to be subjected to the selective examination into a complete text which can be input by a model and add a category label to the text. Secondly, in order to extract the class characteristics of the insurance case recording documents to be sampled, word segmentation processing and stop word processing are required to be carried out on the documents. Firstly, a jieba word segmentation tool is adopted to segment words of the recording documents of the insurance cases to be sampled, and then a certain open source word list is used to remove characters which have no practical significance, such as symbols, numbers, English and the like in original data.
The general character recognition software is OCR software, when the insurance case recording document to be subjected to spot inspection is recognized through the OCR software, PDF (portable document format) can not be recognized under certain conditions, in order to solve the problems, the PDF format of the insurance case recording document to be subjected to spot inspection is converted into a picture, then the picture is recognized through the OCR software, and finally the obtained recognition effect is good. In addition, the conventional steps are to extract the converted pictures according to the number of the PDF pages, convert each page of PDF into each picture, and when taking out the pictures, the pictures are more and more disordered, which may cause the loss of the content. Therefore, the operation steps adopted in the embodiment of the invention are to convert the whole PDF document into one picture, rather than converting each page of PDF document into one picture, so that the messy condition can be avoided, the content loss can not be generated, and the content is very regular.
In addition, when PDF is converted into a picture to interface with OCR software standard, because the transmitted PDF is sometimes too small, which makes recognition unclear, it is necessary to first adjust the PDF picture to an appropriate size and then recognize the PDF picture by OCR software.
On the basis of the above embodiment, it is preferable to further include:
and taking the preset keyword information of the insurance case to be subjected to the spot check as a sample, and taking the spot check result of the insurance case to be subjected to the spot check as a label so as to train the case spot check neural network model again.
Specifically, after the selective examination of the insurance case to be subjected to the selective examination is finished, the insurance case to be subjected to the selective examination can be used as a sample again, and then the neural network model for the selective examination of the case can be updated when the neural network model for the selective examination of the case is trained again next time. The intelligent sampling inspection method also has deep machine learning capability, and the case sampling inspection neural network model can continuously adapt to new products and new rules through learning, so that the intelligent sampling inspection capability is continuously improved, the probability of missing sampling inspection is reduced, the sampling inspection quality is improved, and the like.
The case selective examination neural network model in the embodiment of the invention belongs to one of neural networks, and needs to be trained or updated before the case selective examination neural network model is used, and the case selective examination neural network model is trained through the obtained samples and labels. The training process of the case sampling inspection neural network model can be divided into three steps: defining the structure of a case sampling inspection neural network model and an output result of forward propagation; defining a loss function and a back propagation optimization algorithm; finally, a session is generated and a back propagation optimization algorithm is run repeatedly on the training data.
The neuron is the minimum unit forming the neural network, one neuron can have a plurality of inputs and one output, and the input of each neuron can be the output of other neurons or the input of the whole neural network. The output of the neural network is the weighted sum of the inputs of all the neurons, the weight of different inputs is the neuron parameter, and the optimization process of the neural network is the process of optimizing the value of the neuron parameter.
The effect and optimization goal of the neural network are defined by a loss function, the loss function gives a calculation formula of the difference between the output result of the neural network and the real label, and supervised learning is a way of training the neural network, and the idea is that on a labeled data set of known answers, the result given by the neural network is as close as possible to the real answer (namely, the label). The training data is fitted by adjusting parameters in the neural network so that the neural network provides predictive power to unknown samples.
The back propagation algorithm realizes an iterative process, when each iteration starts, a part of training data is taken first, and the prediction result of the neural network is obtained through the forward propagation algorithm. Because the training data all have correct answers, the difference between the predicted result and the correct answer can be calculated. Based on the difference, the back propagation algorithm can correspondingly update the value of the neural network parameter, so that the neural network parameter is closer to the real answer.
After the training process is completed by the method, the trained target ranking model can be used for application.
On the basis of the above embodiment, it is preferable to further include:
and if the selective examination result is negative, sending a reminding short message to a preset mobile phone number associated with the selective examination person, and/or sending a reminding mail to a preset mailbox account associated with the selective examination person.
According to the selective examination result and the claim adjustment information provided by the case selective examination neural network model, the selective examination personnel is reminded in a mode suitable for business operation, for example, mail short messages are sent to the selective examination personnel, or a special page searching and viewing function is provided in a corresponding management system, so that the selective examination personnel can follow up and adjust the selective examination personnel in time, if the system can automatically apply for the automatic selective examination result and automatically perform the calculation and the examination, the AI capability can be accessed, and the case adjustment can be completed without manual entering.
To sum up, the insurance case spot check method based on the neural network provided by the embodiment of the invention applies artificial intelligence in the field of insurance case spot check, firstly extracts the preset keyword information of the case to be spot checked, quantizes the case to be spot checked through the preset keyword information, and can simplify the description of the case to be spot checked; then inputting preset keyword information of a case to be subjected to spot check into a case spot check neural network model, extracting word frequency characteristics, word position characteristics in a document and context characteristics from the preset keyword information, and then fusing the characteristics, wherein a plurality of simple characteristics are fused instead of simply connected in series, and if the plurality of simple characteristics are connected in series, although certain performance improvement can be obtained, the dimensionality of the characteristics can be greatly increased, the problem of dimensionality disaster is caused, and the complexity of putting in the insurance case to be subjected to spot check is increased; the case selective examination neural network model is obtained after a sample is trained, and the selective examination result of the case to be subjected to selective examination can be directly predicted, so that the manual selective examination work is avoided. The embodiment of the invention can effectively reduce the labor cost, and the case quantity needing to be checked is not increased due to the increase of the case quantity, so that a certain proportion of labor needs to be synchronously increased, and the cost can be effectively reduced; and the intelligent spot check has certain objectivity, and the quality result differentiation of the spot check cannot be caused by the professional skill of personnel and the medical background difference.
Moreover, word2vec word vectors and TF-IDF vectors are fused, so that the dimensionality of fused features is reduced, and complementary characteristics of different modal features of preset keyword information are fully utilized.
Finally, the intelligent sampling inspection method also has deep machine learning capability, and the case sampling inspection neural network model can continuously adapt to new products and new rules through learning, so that the intelligent sampling inspection capability is continuously improved, the probability of missing sampling inspection is reduced, the sampling inspection quality is improved, and the like.
Fig. 3 is a schematic structural diagram of a neural network-based insurance case spot-checking system according to an embodiment of the present invention, and as shown in fig. 3, the system includes a keyword module 310, a fusion module 320, a similarity module 330, and a spot-checking module 340, where:
the keyword module 310 is configured to obtain preset keyword information of the insurance case to be spot-checked;
the fusion module 320 is configured to extract the word frequency features, the word position features and the context semantic features in the preset keyword information respectively based on a fusion unit in the case spot check neural network model, and fuse the extracted features to obtain fusion features;
the similarity module 330 is configured to, based on a similarity calculation unit in the case spot-check neural network model, obtain a plurality of reference insurance history cases with the highest similarity according to the similarity between the fusion feature and the fusion feature corresponding to each insurance history case in a preset insurance case library;
the selective examination module 340 is configured to obtain a selective examination result of the case to be subjected to selective examination according to the claim settlement condition of each reference insurance history case based on a prediction unit in the case selective examination neural network model, where the case selective examination neural network model is obtained by training samples and labels.
The present embodiment is a system embodiment corresponding to the above method embodiment, the specific implementation process is the same as the above method embodiment, please refer to the above method embodiment for details, and the system embodiment is not described herein again.
On the basis of the foregoing embodiment, preferably, the similarity module includes a cosine unit and a screening unit, where:
the cosine unit is used for calculating cosine similarity between the fusion features and fusion features corresponding to each insurance history case in the preset insurance case library;
the screening unit is used for taking a plurality of insurance history cases with the largest cosine similarity as the reference insurance history cases.
On the basis of the foregoing embodiment, preferably, the fusion module includes a word frequency unit and a semantic unit, where:
the word frequency unit is used for carrying out word frequency-inverse document feature extraction on the preset keyword information to obtain the word frequency features;
and the semantic unit is used for carrying out word2vec feature extraction on the preset keyword information to obtain word position features and the context semantic features in the text.
On the basis of the foregoing embodiment, preferably, the keyword module includes a picture unit, a representation unit, and a key unit, where:
the picture unit is used for acquiring a picture corresponding to the insurance case recording document to be subjected to spot inspection;
the representation unit is used for extracting representation information corresponding to preset keywords in the picture;
the key unit is used for acquiring the preset keyword information according to the representation information corresponding to the preset keyword.
On the basis of the foregoing embodiment, preferably, the keyword module further includes a filtering unit, wherein:
the filtering unit is used for filtering the picture and removing invalid information in the picture.
On the basis of the foregoing embodiment, it is preferable that the mobile terminal further includes an update module, where:
the updating module is used for taking the preset keyword information of the insurance case to be subjected to the selective examination as a sample and taking the selective examination result of the insurance case to be subjected to the selective examination as a label so as to train the case selective examination neural network model again.
On the basis of the above embodiment, preferably, the system further includes an alarm module, wherein:
and the alarm module is used for sending a reminding short message to a preset mobile phone number associated with the random inspector and/or sending a reminding mail to a preset mailbox account associated with the random inspector if the random inspection result is negative.
The modules in the neural network based insurance case spot-checking system can be wholly or partially realized by software, hardware and a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.
Fig. 4 is a schematic structural diagram of a computer device according to an embodiment of the present invention, where the computer device may be a server, and an internal structural diagram of the computer device may be as shown in fig. 4. The computer device includes a processor, a memory, a network interface, and a database connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a computer storage medium and an internal memory. The computer storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of an operating system and computer programs in the computer storage media. The database of the computer device is used for storing data generated or acquired in the process of executing the neural network-based insurance case spot check method, such as preset keyword information and fusion characteristics. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to implement a neural network-based insurance case spot check method.
In one embodiment, a computer device is provided, which includes a memory, a processor, and a computer program stored on the memory and executable on the processor, and when the processor executes the computer program, the steps of the neural network based insurance case spot check method in the above embodiments are implemented. Alternatively, the processor, when executing the computer program, implements the functions of the modules/units in the embodiment of the neural network based insurance case spot-checking system.
In an embodiment, a computer storage medium is provided, on which a computer program is stored, which when executed by a processor implements the steps of the neural network based insurance case spot check method in the above embodiments. Alternatively, the computer program is executed by a processor to implement the functions of the modules/units in the embodiment of the neural network based insurance case spot-checking system.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in the embodiments provided herein may include non-volatile and/or volatile memory, among others. Non-volatile memory can include read-only memory (ROM), Programmable ROM (PROM), Electrically Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus Direct RAM (RDRAM), direct bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
It will be apparent to those skilled in the art that, for convenience and brevity of description, only the above-mentioned division of the functional units and modules is illustrated, and in practical applications, the above-mentioned function distribution may be performed by different functional units and modules according to needs, that is, the internal structure of the apparatus is divided into different functional units or modules to perform all or part of the above-mentioned functions.
The above-mentioned embodiments are only used for illustrating the technical solutions of the present invention, and not for limiting the same; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; such modifications and substitutions do not substantially depart from the spirit and scope of the embodiments of the present invention, and are intended to be included within the scope of the present invention.
Claims (10)
1. A neural network-based insurance case spot-check method is characterized by comprising the following steps:
acquiring preset keyword information of the insurance case to be sampled and inspected;
extracting word frequency characteristics, word position characteristics and context semantic characteristics in the preset keyword information respectively based on a fusion unit in the case spot check neural network model, and fusing the extracted characteristics to obtain fusion characteristics;
based on a similarity calculation unit in the case spot check neural network model, acquiring a plurality of reference insurance history cases with highest similarity according to the similarity between the fusion features and the fusion features corresponding to each insurance history case in a preset insurance case library;
and acquiring the selective examination result of the case to be subjected to selective examination according to the claim settlement condition of each reference insurance history case based on a prediction unit in the case selective examination neural network model, wherein the case selective examination neural network model is obtained by training samples and labels.
2. The neural network based insurance case spot-check method according to claim 1, wherein the obtaining of the plurality of insurance history cases with the highest similarity according to the similarity between the fusion features and the fusion features corresponding to each reference insurance history case in a preset insurance case library comprises:
calculating the cosine similarity between the fusion features and the fusion features corresponding to each insurance history case in the preset insurance case library;
and taking a plurality of insurance history cases with the maximum cosine similarity as the reference insurance history cases.
3. The insurance case spot-check method based on neural network as claimed in claim 1, wherein said extracting the word frequency feature, the word position feature and the context semantic feature in the preset keyword information respectively comprises:
extracting word frequency-inverse document characteristics from the preset keyword information to obtain the word frequency characteristics;
and performing word2vec feature extraction on the preset keyword information to obtain word position features and the context semantic features in the text.
4. The neural network-based insurance case spot-checking method according to claim 1, wherein the acquiring of the preset keyword information of the insurance case to be spot-checked comprises:
acquiring a picture corresponding to the insurance case recording document to be spot-checked;
extracting representation information corresponding to preset keywords in the picture;
and acquiring the preset keyword information according to the representation information corresponding to the preset keyword.
5. The neural network based insurance case spot-checking method according to claim 4, wherein the extracting of the representation information corresponding to the preset keywords in the picture further comprises:
and filtering the picture to remove invalid information in the picture.
6. The neural network based insurance case spot-check method of any one of claims 1 to 4, further comprising:
and taking the preset keyword information of the insurance case to be subjected to the spot check as a sample, and taking the spot check result of the insurance case to be subjected to the spot check as a label so as to train the case spot check neural network model again.
7. The neural network based insurance case spot-check method of any one of claims 1 to 4, further comprising:
and if the selective examination result is negative, sending a reminding short message to a preset mobile phone number associated with the selective examination person, and/or sending a reminding mail to a preset mailbox account associated with the selective examination person.
8. A neural network based insurance case spot-check system, comprising:
the keyword module is used for acquiring preset keyword information of the insurance case to be subjected to spot check;
the fusion module is used for extracting the word frequency characteristics, the word position characteristics and the context semantic characteristics in the preset keyword information respectively based on a fusion unit in the case spot check neural network model, and fusing the extracted characteristics to obtain fusion characteristics;
the similarity module is used for acquiring a plurality of reference insurance history cases with the highest similarity according to the similarity between the fusion characteristics and the fusion characteristics corresponding to each insurance history case in a preset insurance case library based on a similarity calculation unit in the case spot check neural network model;
and the sampling inspection module is used for acquiring sampling inspection results of the cases to be sampled and inspected according to the claim settlement condition of each reference insurance history case based on a prediction unit in the case sampling inspection neural network model, wherein the case sampling inspection neural network model is obtained by training samples and labels.
9. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor when executing the computer program implements the steps of the neural network based insurance case spot check method of any one of claims 1 to 7.
10. A computer storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the steps of the neural network based insurance case spot check method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210091571.5A CN114417045A (en) | 2022-01-26 | 2022-01-26 | Insurance case spot inspection method, system, equipment and storage medium based on neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210091571.5A CN114417045A (en) | 2022-01-26 | 2022-01-26 | Insurance case spot inspection method, system, equipment and storage medium based on neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114417045A true CN114417045A (en) | 2022-04-29 |
Family
ID=81277854
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210091571.5A Pending CN114417045A (en) | 2022-01-26 | 2022-01-26 | Insurance case spot inspection method, system, equipment and storage medium based on neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114417045A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116307607A (en) * | 2023-03-24 | 2023-06-23 | 探保网络科技(广州)有限公司 | Insurance core system monitoring system and method |
-
2022
- 2022-01-26 CN CN202210091571.5A patent/CN114417045A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116307607A (en) * | 2023-03-24 | 2023-06-23 | 探保网络科技(广州)有限公司 | Insurance core system monitoring system and method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20230222366A1 (en) | Systems and methods for semantic analysis based on knowledge graph | |
| Capuano et al. | Sentiment analysis for customer relationship management: an incremental learning approach | |
| WO2020134008A1 (en) | Method and apparatus for matching semantic text data with tags, and computer readable storage medium storing instruction | |
| US10733675B2 (en) | Accuracy and speed of automatically processing records in an automated environment | |
| CN111859916B (en) | Method, device, equipment and medium for extracting key words of ancient poems and generating poems | |
| US11847245B2 (en) | Privacy preserving data labeling | |
| CN112256863B (en) | Method and device for determining corpus intention and electronic equipment | |
| CN113112282A (en) | Method, device, equipment and medium for processing consult problem based on client portrait | |
| CN111462752A (en) | Client intention identification method based on attention mechanism, feature embedding and BI-L STM | |
| Rafail et al. | Natural language processing | |
| CN116628163A (en) | Customer service processing method, customer service processing device, customer service processing equipment and storage medium | |
| CN115017879A (en) | Text comparison method, computer device and computer storage medium | |
| CN114757178A (en) | Core product word extraction method, device, equipment and medium | |
| CN114417045A (en) | Insurance case spot inspection method, system, equipment and storage medium based on neural network | |
| CN113240531A (en) | Data processing method and device | |
| CN113011156A (en) | Quality inspection method, device and medium for audit text and electronic equipment | |
| CN112464670A (en) | Recognition method, recognition model training method, device, equipment and storage medium | |
| CN112528887B (en) | Auditing method and device | |
| CN113901817A (en) | Document classification method and device, computer equipment and storage medium | |
| CN114358647A (en) | Case recommendation method, system, device and storage medium based on neural network | |
| CN113742452B (en) | Text classification-based public opinion monitoring method, device, equipment and medium | |
| US20230042449A1 (en) | System and method for decreasing turnaround for pre-authorizations using a smart request for information model | |
| CN112133308B (en) | Method and device for classifying multiple tags of speech recognition text | |
| US20240236191A1 (en) | Segmented hosted content data streams | |
| KR20230126390A (en) | Method and system for providing user-customized interview feedback for educational purposes based on deep learning |
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 |