CN109376237A - Prediction technique, device, computer equipment and the storage medium of client's stability - Google Patents
Prediction technique, device, computer equipment and the storage medium of client's stability Download PDFInfo
- Publication number
- CN109376237A CN109376237A CN201811025974.XA CN201811025974A CN109376237A CN 109376237 A CN109376237 A CN 109376237A CN 201811025974 A CN201811025974 A CN 201811025974A CN 109376237 A CN109376237 A CN 109376237A
- Authority
- CN
- China
- Prior art keywords
- target
- information
- customer
- client
- 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 34
- 238000012544 monitoring process Methods 0.000 claims abstract description 47
- 238000004458 analytical method Methods 0.000 claims description 55
- 238000004590 computer program Methods 0.000 claims description 16
- 238000012549 training Methods 0.000 claims description 15
- 210000002569 neuron Anatomy 0.000 claims description 10
- 238000012545 processing Methods 0.000 claims description 8
- 239000011159 matrix material Substances 0.000 claims description 7
- 238000012546 transfer Methods 0.000 claims description 6
- 238000003062 neural network model Methods 0.000 claims description 5
- 230000011218 segmentation Effects 0.000 claims description 5
- 238000013507 mapping Methods 0.000 claims description 3
- 238000013473 artificial intelligence Methods 0.000 abstract 1
- 230000002996 emotional effect Effects 0.000 description 9
- 238000004422 calculation algorithm Methods 0.000 description 8
- 239000000284 extract Substances 0.000 description 8
- 230000008451 emotion Effects 0.000 description 7
- 230000004927 fusion Effects 0.000 description 7
- 230000006399 behavior Effects 0.000 description 6
- 238000010586 diagram Methods 0.000 description 6
- 238000010801 machine learning Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000013145 classification model Methods 0.000 description 3
- 238000004891 communication Methods 0.000 description 3
- 238000005194 fractionation Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 238000007477 logistic regression Methods 0.000 description 3
- 239000000203 mixture Substances 0.000 description 3
- 238000012216 screening Methods 0.000 description 3
- 241001269238 Data Species 0.000 description 2
- 230000003542 behavioural effect Effects 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000012512 characterization method Methods 0.000 description 2
- 238000003066 decision tree Methods 0.000 description 2
- 238000007637 random forest analysis Methods 0.000 description 2
- 230000002441 reversible effect Effects 0.000 description 2
- 230000002123 temporal effect Effects 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 239000004744 fabric Substances 0.000 description 1
- 238000000855 fermentation Methods 0.000 description 1
- 230000004151 fermentation Effects 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 230000003862 health status Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 230000003211 malignant effect Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003014 reinforcing effect Effects 0.000 description 1
- 230000006641 stabilisation Effects 0.000 description 1
- 238000011105 stabilization Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 230000001360 synchronised 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
- G06F40/30—Semantic analysis
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
Prediction technique, device, computer equipment and the storage medium for client's stability based on artificial intelligence that this application involves a kind of.The described method includes: multiple target informations of monitoring network platform publication；Obtain the corresponding customer data of target customer；The customer data includes product identification；The corresponding public opinion index of the product identification is calculated based on the target information；Count access data of the target customer in the monitoring period to target information；The affective characteristics of the target customer are determined according to the customer data；Accordingly and the affective characteristics input preset informational influence prediction model by the public opinion index, the access number, the stability parameter of the target customer is exported.Client's stability can be predicted using this method in time and improve predictablity rate.
Description
Technical field
This application involves field of computer technology, prediction technique, device, calculating more particularly to a kind of client's stability
Machine equipment and storage medium.
Background technique
As the influence power of the Internet media becomes larger, also got over by the network platform that internet releases news to user
Come more.Although the information of network platform publication itself may be free of any emotion word, accessing this category information can allow people to produce
Certain raw Sentiment orientation, and some Sentiment orientations then directly affect the stability of corporate client.Although current most enterprise can needle
When being responded to the information of network platform publication, information etc. for example, publication is refuted a rumour, but responding relevant information, often exist
Message event is issued and is passed through after public opinion fermentation, corresponds to for enterprise at this time, has had already appeared the bad phenomenons such as customer churn.
As it can be seen that the prior art lacks the early warning scheme for providing to enterprise and being directed to client's stability.
Summary of the invention
Based on this, it is necessary to which in view of the above technical problems, client's stability can be predicted in time and improve pre- by providing one kind
Survey prediction technique, device, computer equipment and the storage medium of client's stability of accuracy rate.
A kind of prediction technique of client's stability, which comprises multiple target informations of monitoring network platform publication；
Obtain the corresponding customer data of target customer；The customer data includes product identification；Based on described in target information calculating
The corresponding public opinion index of product identification；Count access data of the target customer in the monitoring period to target information；According to institute
State the affective characteristics that customer data determines the target customer；Accordingly and the emotion by the public opinion index, the access number
Feature inputs preset informational influence prediction model, exports the stability parameter of the target customer.
Multiple target informations of the monitoring network platform publication in one of the embodiments, comprising: monitoring network is flat
The raw information of platform publication；Word segmentation processing is carried out to the raw information, obtains the corresponding information mark of each raw information
Label；Multiple target keywords are obtained, identify whether the information labels include the target keyword；If so, will be corresponding original
Information flag is target information.
It is described in one of the embodiments, to be referred to based on the corresponding public sentiment of the target information calculating product identification
Number, comprising: the target information is split, multiple short texts are obtained；Product identification is extracted in the short text, by institute
Product identification is stated to be associated with corresponding short text；The corresponding emotion of each short text is calculated using preset the analysis of public opinion model
Index；Determine the corresponding influence power weight of multiple short texts；According to the affection index and shadow of associated short text
Power weight is rung, corresponding product is calculated and identifies corresponding public opinion index.
The statistics target customer is in the monitoring period to the access number of target information in one of the embodiments,
According to, comprising: the identification field is sent to the network platform by the identification field for obtaining target customer；Receive the network
The associated access data that platform is returned according to the identification field；In the associated access data extract message reference field,
Information collection field and information forward field；Based on the message reference field, information collection field and information forwarding field system
The target customer is counted in information access volume, information collection amount and the information transfer amount in monitoring period.
The affective characteristics that the target customer is determined according to the customer data in one of the embodiments, packet
It includes: obtaining multiple submodels, determine the corresponding weight of multiple submodels；According to multiple submodels and divide
Not corresponding weight generates the first model；Obtain client's sample data and corresponding affective tag；By client's sample
Notebook data inputs first model, obtains intermediate sentiment analysis result；Calculate the intermediate sentiment analysis result and the feelings
The difference for feeling label, is adjusted first model according to the difference, obtains the second model；The customer data is defeated
Enter second model, exports the affective characteristics of the target customer.
The informational influence prediction model is obtained using deep neural network model training in one of the embodiments,；
It is described that the public opinion index, access data and affective characteristics are inputted into preset informational influence prediction model, export the target
The stability parameter of client, comprising: the public opinion index, access data and affective characteristics are pre-processed, client characteristics square is obtained
Battle array；Input layer sequence is obtained according to the client characteristics matrix；The input layer sequence is projected, obtains
The corresponding hidden node sequence of one hidden layer, using first hidden layer as currently processed hidden layer；Obtain the currently processed hidden layer
The weight and deviation of corresponding each neuron node；According to the corresponding hidden node sequence of the currently processed hidden layer and each
The weight and deviation of neuron node obtain the hidden node sequence of next hidden layer using Nonlinear Mapping；Next hidden layer is made
It is iterated for currently processed hidden layer, until output layer；Obtain the corresponding stabilization of the target customer of the output layer output
Property parameter.
A kind of prediction meanss of client's stability, described device includes: information analysis module, is sent out for monitoring network platform
Multiple target informations of cloth；Obtain the corresponding customer data of target customer；The customer data includes product identification；Based on described
Target information calculates the corresponding public opinion index of the product identification；Customer analysis module, for counting target customer in the prison
The period is controlled to the access data of target information；The affective characteristics of the target customer are determined according to the customer data；It influences pre-
Survey module, for by the public opinion index, the access number accordingly and the affective characteristics input the prediction of preset informational influence
Model exports the stability parameter of the target customer.
A kind of computer equipment, including memory and processor, the memory are stored with computer program, the processing
Device realizes the prediction technique of the client's stability provided in any one embodiment of the application when executing the computer program
Step.
A kind of computer readable storage medium, is stored thereon with computer program, and the computer program is held by processor
The step of prediction technique of the client's stability provided in any one embodiment of the application is provided when row.
Prediction technique, device, computer equipment and the storage medium of above-mentioned client's stability, to the more of network platform publication
A target information real-time monitoring；The corresponding customer data of the target customer analyzed as needed, available one or more productions
Product mark；According to the target information monitored, different product can be calculated and identify corresponding public opinion index；Pass through statistics target visitor
Family monitors the period to the access data of target information described, and the emotion of the target customer is determined according to the customer data
Feature can obtain the stability parameter of target customer based on informational influence prediction model.Due to not only predicting target information pair
The influence of product obtains the corresponding public opinion index of product identification, also using target customer to the degree of understanding of target information, into one
Step considers the affective characteristics whether client influences vulnerable to target information, and comprehensively considering Multiple factors can be improved informational influence prediction
Accuracy rate；The many factors data being calculated, which are directly inputted preset informational influence prediction model, can be obtained prediction knot
Fruit can predict client's stability in time and improve predictablity rate.
Detailed description of the invention
Fig. 1 is the application scenario diagram of the prediction technique of client's stability in one embodiment；
Fig. 2 is the flow diagram of the prediction technique of client's stability in one embodiment；
Fig. 3 is the flow diagram that product public opinion index step is calculated in one embodiment；
Fig. 4 is the flow diagram that client's affective characteristics step is determined in one embodiment；
Fig. 5 is the structural block diagram of the prediction meanss of client's stability in one embodiment；
Fig. 6 is the internal structure chart of computer equipment in one embodiment.
Specific embodiment
It is with reference to the accompanying drawings and embodiments, right in order to which the objects, technical solutions and advantages of the application are more clearly understood
The application is further elaborated.It should be appreciated that specific embodiment described herein is only used to explain the application, not
For limiting the application.
The prediction technique of client's stability provided by the present application, can be applied in application environment as shown in Figure 1.Its
In, terminal 102 is communicated with server 104 by network.Wherein, terminal 102 can be, but not limited to be various individual calculus
Machine, laptop, smart phone, tablet computer and portable wearable device, server 104 can use independent server
The either server cluster of multiple servers composition is realized.To 104 monitoring network platform of server in the publication of monitoring period
Raw information, relevant to business event target information is screened from raw information.Server 104 is sent according to terminal 102
Informational influence predictions request or whether the stability of target customer can be made according to preset time frequency predication target information
At influence.Specifically, server 104 obtains the corresponding customer data of target customer.Customer data includes product identification.Server
104 calculate the corresponding public opinion index of product identification based on target information, i.e. prediction target information is to the corresponding product of product identification
Price etc. influence.Server 104 counts access data of the target customer in the monitoring period to target information, such as believes target
The behavioral data of browsing, forwarding or the collection of breath etc..Server 104 determines the affective characteristics of target customer according to customer data,
Influence to change the degree determined vulnerable to extraneous.Server 104 has been pre-stored informational influence prediction model.Informational influence predicts mould
Type can be to be obtained using deep neural network model training.Server 104 by public opinion index, access number accordingly and emotion is special
Sign input informational influence prediction model, exports the stability parameter of target customer.Above- mentioned information influence prediction process, not only predict
Influence of the target information to product obtains the corresponding public opinion index of product identification, also using target customer to target information
Solution degree further considers the affective characteristics whether client influences vulnerable to target information, and comprehensively considering Multiple factors can be improved
Informational influence predictablity rate.
In one embodiment, it as shown in Fig. 2, providing a kind of prediction technique of client's stability, applies in this way
It is illustrated for server in Fig. 1, comprising the following steps:
Step 202, multiple target informations of monitoring network platform publication.
The network platform can be preassigned much information reader, as UC browser, QQ browser, today's tops,
Flash report etc. everyday.The network platform can also be preassigned a variety of social platforms, as wechat, microblogging, QQ, discussion bar, forum or
Know.Target information has corresponding influence object.Target information will affect the emotional attitude of people, and then to influence object
Benign or malignant influence is caused, such as traveller is lost, resource is devalued.Influence object type can be communication equipment, house property is built,
Virtual resource, other products etc..Wherein, virtual resource can be product etc..The monitoring period refers to collection of server target information
Temporal frequency, can freely set according to demand, such as 1 week, with no restriction to this.Server is crawled from the different network platforms
The temporal frequency of target information can be different.
In one embodiment, multiple target informations of monitoring network platform publication, comprising: the publication of monitoring network platform
Raw information；Word segmentation processing is carried out to raw information, obtains the corresponding information labels of each raw information；Multiple targets are obtained to close
Whether keyword, identification information label include target keyword；If so, corresponding raw information is labeled as target information.
The network platform may send out a plurality of raw information in monitoring period not timing.But and not all raw information will affect
Client's stability, screening server raw information relevant to business event.Specifically, server is original to collected every
Information is segmented, goes the processing such as stop words and name entity replacement, and multiple information keys are obtained.Server calculates each letter
Cease TF-IDF value (termfrequency-inverse document frequency, the reverse file frequency of word frequency-of keyword
Rate).Specifically, the number that server occurs in all information keys by counting each information key, calculates corresponding
The TF value (term frequency, word frequency) of information key；Pass through the total quantity and packet of informative statement in statistics raw information
The ratio of the quantity of informative statement containing some information key calculates the IDF value (inverse of corresponding information keyword
Document frequency, reverse document-frequency).Server calculates phase according to the TF value and IDF value of each information key
Answer the TF-IDF value of information key.TF-IDF value can reflect the class of service separating capacity of information key.Server root
Information key is screened according to TF-IDF value, screening obtains the high information key of preset quantity TF-IDF value.
The information key that server by utilizing is screened generates the corresponding information labels of raw information.Server is pre-stored
Multiple target keywords relevant to business.Server matches information labels with target keyword, identification information mark
Whether label include target keyword.If information labels include target keyword, indicate that raw information is related to business event, service
The raw information is labeled as target information by device.The present embodiment screens collected information, allow server only
The analysis of public opinion is carried out to target information relevant to business event, it is possible to reduce server data volume to be treated, and then subtract
Few server resource occupies, and improves informational influence forecasting efficiency.
Step 204, the corresponding customer data of target customer is obtained；Customer data includes product identification.
Target customer can be existing client, be also possible to potential customers.For potential customers, server can pass through net
The mode of network crawler obtains corresponding customer data.Customer data includes the build-in attributes such as gender, age, occupation, also includes net
The dynamic behaviours data such as page browsing, information forwarding, dynamic release, can be text, voice, video or picture etc..For existing
Client, server directly can obtain corresponding customer data in operation system.The customer data of existing client includes having bought
Or prepare the mark (hereinafter referred to as " product identification ") of the product of purchase.For example, client bought and it is the currently active, bought
But failed and prepare purchase insurance products product identification.
Step 206, the corresponding public opinion index of product identification is calculated based on target information.
Server splits target information, obtains multiple short texts.Server extracts in target information being capable of table
The keyword for levying the object (hereinafter referred to as " influencing object ") that it may be influenced determines the target pair according to the keyword extracted
As corresponding influence object type.A variety of influence object types have been stored in advance in server and every kind of influence object type is corresponding
The public sentiment factor and the analysis of public opinion model.According to target information, server obtain it is corresponding influence the corresponding public sentiment of object type because
Son extracts target keyword in multiple short texts according to the public sentiment factor respectively, the target keyword extracted is inputted the shadow
The corresponding the analysis of public opinion model of object type is rung, the corresponding affection index of target information is calculated.Server can be to virtual
A variety of target informations for influencing object type such as resource, communication equipment are analyzed.When influencing object type is virtual resource,
Server is also used to calculate the corresponding public opinion index of virtual resource according to affection index.Public opinion index can characterize target information pair
The influence degree of price, audient crowd, the popularity of different product etc..
Step 208, statistics target customer is in the monitoring period to the access data of target information.
Server collects access of the target customer in the monitoring period to target information based on customer ID in the network platform
Data.Access data include to the browsing of one or more target information record, such as the browsing time, whether forward, whether collect,
Browse duration, comment information etc..Server is according to access data statistics target customer in information access volume, the information for monitoring the period
Amount of collection and information transfer amount.
In one embodiment, statistics target customer is in the monitoring period to the access data of target information, comprising: obtains mesh
The identification field for marking client, is sent to the network platform for identification field；Receive the association that the network platform is returned according to identification field
Access data；Message reference field, information collection field and information are extracted in associated access data forwards field；Based on information
Field, information collection field and information forwarding statistics target customer is accessed to receive in the information access volume in monitoring period, information
Reserve and information transfer amount.
Target customer has corresponding identification field.Server extracts in the identity information that enterprise retains from target customer
Basic identification field.Identification field can be relatives or the friend's (hereinafter referred to as " association pair of target customer and target customer
As ") identification field.Identification field include name, identification card number, cell-phone number, Email Accounts, network account, often
With facility information etc..Commonly used equipment information can be IMEI (International Mobile Equipment Identity,
International mobile equipment identification number), IP address, device-fingerprint, operating system version number, sequence number etc..
Heterogeneous networks platform has been run on different Internet Servers.Target customer is using in various kinds of equipment access mechanism
When the outer network platform, access record will be left in corresponding Internet Server.Access record can be with log or file etc.
Form storage.Server generates data retrieval request, data retrieval request is sent out according to the basic identification field of target customer
It send to Internet Server.Internet Server is searched the access comprising identification field and is recorded, and the access found record is returned
It is back to server.Access record refers to that target customer is based on the hair such as mobile terminal, automobile, intelligent robot, intelligent wearable device
The behavioral data of raw message reference behavior (such as browsing behavior, comment behavior, forwarding behavior, collection behavior) (hereinafter referred to as " closes
Connection access data ").Server extracts message reference field, information collection field and information forwarding word in associated access data
Section etc., it is for statistical analysis to the field information extracted, it obtains target customer and is visited in information of the monitoring period to target information
The amount of asking, information collection amount and information transfer amount etc..
Step 210, the affective characteristics of target customer are determined according to customer data.
The customer data training in server by utilizing history monitoring period obtains the analysis model in different monitoring period.Each prison
One analysis model of period generation is controlled, there is each disaggregated model corresponding model identification (can be the date, such as 201708)
With weight factor W.Server chooses the analysis model of preset quantity based on preset sliding window function, the analysis obtained according to screening
Model and corresponding weight factor, construct initial machine learning model.Customer data of the server based on the current monitor period
Preset machine learning model is trained, affective characteristics model is obtained.Server is when reaching in the monitoring period, by target visitor
The customer data of the multiple dimensions in family inputs affective characteristics model, obtains the affective characteristics of target customer.Affective characteristics can be use
It is influenced and the degree (i.e. degree of influence of the public opinion to the client) for fluctuation of producing a feeling in characterization client vulnerable to public opinion
Quantitatively or qualitatively parameter value.For example, the affective characteristics of -100~100 characterization clients, the bigger table of parameter absolute value can be used
Show that public opinion is bigger to the degree of influence of client, parameter value is bigger to indicate public opinion to the positive influences dynamics of client more
Greatly, the smaller expression public opinion of parameter value is bigger to the negative effect dynamics of client.
Step 212, accordingly and affective characteristics input preset informational influence prediction model by public opinion index, access number, defeated
The stability parameter of target customer out.
Server has been pre-stored informational influence prediction model.Informational influence prediction model can be using deep neural network
What model training obtained.Specifically, informational influence prediction model includes input layer and output layer.Between input layer and output layer also
Including with multiple hidden layers.Full connection between layers.Every layer includes multiple neurons, and the neuron of same layer inputs parameter phase
Together.
In one embodiment, informational influence prediction model is obtained using deep neural network model training；Public sentiment is referred to
Number, access data and affective characteristics input preset informational influence prediction model, export the stability parameter of target customer, wrap
It includes: public opinion index, access data and affective characteristics being pre-processed, client characteristics matrix is obtained；It is obtained according to client characteristics matrix
Input layer sequence；Input layer sequence is projected, the corresponding hidden node sequence of the first hidden layer is obtained, by first
Hidden layer is as currently processed hidden layer；Obtain weight and deviation that currently processed hidden layer corresponds to each neuron node；According to current
The weight and deviation for handling the corresponding hidden node sequence of hidden layer and each neuron node, are obtained down using Nonlinear Mapping
The hidden node sequence of one hidden layer；It is iterated next hidden layer as currently processed hidden layer, until output layer；Obtain output layer
The corresponding stability parameter of the target customer of output.
In the present embodiment, to the network platform in the multiple target information real-time monitorings for monitoring period publication；Divide as needed
The corresponding customer data of the target customer of analysis, available one or more product identification；It, can according to the target information monitored
Corresponding public opinion index is identified to calculate different product；By statistics target customer in the monitoring period to the access number of target information
According to, and determine according to customer data the affective characteristics of target customer, target customer can be obtained based on informational influence prediction model
Stability parameter.Due to not only predicting influence of the target information to product, the corresponding public opinion index of product identification is obtained, is also adopted
With target customer to the degree of understanding of target information, the affective characteristics whether client influences vulnerable to target information are further considered,
Comprehensively considering Multiple factors can be improved informational influence predictablity rate；The many factors data being calculated are directly inputted pre-
Prediction result can be obtained in the informational influence prediction model set, and can predict client's stability in time and improve predictablity rate.
In one embodiment, it as shown in figure 3, calculating the corresponding public opinion index of product identification based on target information, that is, counts
The step of calculating product public opinion index, comprising:
Step 302, target information is split, obtains multiple short texts.
Step 304, product identification is extracted in short text, and product identification is associated with corresponding short text.
Step 306, the corresponding affection index of each short text is calculated using preset the analysis of public opinion model.
Step 308, the corresponding influence power weight of multiple short texts is determined.
Step 310, it according to the affection index of associated short text and influence power weight, calculates corresponding product mark and corresponds to
Public opinion index.
Target information can be text, voice, video or picture etc..If target information is voice, video or picture, incite somebody to action
It is first converted to text.Target information after conversion is the long text for including multiple fractionation identifiers.Server is by each fractionation
Identifier position is determined as splitting position, is split in each fractionation position of long text, obtains multiple short texts.It tears open
Divide identifier can be with statement terminator, such as fullstop, exclamation mark.
Server extracts interim key word in each short text.Specifically, server carries out word segmentation processing to short text, if
Participle includes stop words or punctuation mark, is filtered to multiple participles, stop words and punctuation mark is deleted, to save server
Memory space.Server carries out synonym replacement and name entity replacement to filtered multiple participles.Server is deposited in advance
Synonym table and name entity are stored up.Synonym replacement can carry out unification to a variety of expression ways of same concept, so that
The key concept of short essay does not highlight, the difficulty that server carries out the analysis of public opinion according to interim key word is reduced, so as to mention
High the analysis of public opinion efficiency and accuracy rate.The replacement of name entity can reduce the granularity of the analysis of public opinion, can be further improved carriage
The efficiency of mutual affection analysis.The corresponding public sentiment factor of a variety of influence object types according to the pre-stored data, after server will be replaced
One or more participles be determined as interim key word.The public sentiment factor, which refers to, may influence user feeling state in such target information
The factor of degree.
The analysis of public opinion model has been stored in advance in server.The analysis of public opinion model can obtain machine learning classification model training
It arrives.Specifically, server, which is based on word2vec model, is separately converted to corresponding term vector for multiple interim key words, and right
Each term vector adds corresponding tag along sort.Term vector and corresponding tag along sort composing training collection, based on training set to machine
Device learning classification model is trained, and obtains the analysis of public opinion model.Machine learning classification model can be GBDT model or
XGBOOST model etc..Server, which inputs the interim key word extracted, accordingly influences the corresponding the analysis of public opinion mould of object type
The corresponding affection index of target information is calculated in type.
Each target information has corresponding profile information, such as issuing time, publication medium, publication author.Server
Profile information based on target information calculates the influence power weight of each target information.For example, influence power weight can be the time
Weight, media weight and author's weight etc. cumulative and.It is readily appreciated that, multiple short texts pair that same target information is split
The influence power weight answered is identical.
The affection index for the target information that server is calculated refers to including the accordingly corresponding emotion of multiple short texts
Number.Server extracts product identification by dictionary tree (trie) algorithm in short text.Product identification can be name of product or
Product number etc..In other words, the interim key word that server extracts in certain short texts includes product identification.Server exists
Identical or different product identification can be extracted in different short texts.Server closes product identification and corresponding short text
Connection.It is readily appreciated that, identical product mark may be associated with multiple short texts from multiple target informations.Server is according to product
It identifies the affection index of corresponding short text and corresponds to influence power weight, calculate the corresponding target public opinion index of corresponding product.
For example, the corresponding public opinion index of each product identification can be and the affection index of the associated whole short texts of the product identification
Weighted sum, such as product A public opinion index=short text 11* influence power weight 11+ short text 12* influence power weight 12+...+ short essay
This 21* influence power weight 21.
In the present embodiment, the influence power weight calculation different target information of combining target information influences different product, i.e.,
The analysis of public opinion accuracy can be improved in public opinion index.
In one embodiment, as shown in figure 4, determining the affective characteristics of target customer according to customer data, that is, visitor is determined
The step of family affective characteristics, comprising:
Step 402, multiple submodels are obtained, determine the corresponding weight of multiple submodels.
Step 404, according to multiple submodels and corresponding weight, the first model is generated.
Step 406, client's sample data and corresponding affective tag are obtained.
Step 408, client's sample data is inputted into the first model, obtains intermediate sentiment analysis result.
Step 410, the difference for calculating intermediate sentiment analysis result and affective tag, adjusts the first model according to difference
It is whole, obtain the second model.
Step 412, customer data is inputted into the second model, exports the affective characteristics of target customer.
Server monitors the emotional feature analysis model in period building corresponding monitoring period every one.Monitor the period when
Between length can freely set according to demand, such as 1 year.The current monitor period, corresponding emotional feature analysis model can be utilization
The emotional feature analysis model construction in multiple history monitoring periods forms.For convenience, by the emotion in history monitoring period
Characteristic Analysis Model is referred to as " submodel ".Initial submodel can be a large amount of client's sample datas of server by utilizing to introductory die
Type training obtains.
Server obtains client's sample data of multiple historical periods.Historical period is opposite with the above-mentioned history monitoring period
It answers.Server adds corresponding quality label for client's sample data of each client.In order to reduce manually mark it is cumbersome,
Server establishes customer portrait according to client's sample data, and the quality label of respective client is automatically generated based on customer portrait.Tool
Body, server carries out client's sample data the processing such as to clean, and the corresponding multiple attribute tags of client is obtained, such as the year of user
Age, gender, occupation, marital status, schooling, occupation, property guarantee, health status etc..Multiple categories that server will acquire
Property set of tags become a text vector, using the text vector of composition as the customer portrait of the client.Customer portrait is as real
The virtual representations of border client are often built according to product and market, have been reacted the feature of actual customer and have been needed
It asks.Server has been pre-stored a variety of attribute tags combinations and the corresponding quality label of every kind of combination.Server is based on pre-
The corresponding relationship of the attribute tags combination and quality label of storage, converts customer portrait, obtains respective client mark pair
The quality label answered.Quality label can be the quantitative targets such as score value, be also possible to the qualitative indexes such as excellent, good, poor.
Server is based on a large amount of client's sample datas and corresponding quality label to initial model training, obtains corresponding submodule
Type.Initial model can be tagsort model and Fusion Features models fitting obtains.Initial model includes multiple client's indexs,
Every kind of client's index has corresponding a variety of client properties, as the corresponding client properties of client's index " gender " can be " male " or
" female ".Server calculates the corresponding entropy gain of each client's index.The formula for calculating entropy gain may is that
Wherein, GA indicates the entropy gain of the client's index A calculated；M indicates that client's affective characteristics index reaches threshold value
Probability；Ai indicates that the quantity of the client properties i of corresponding client's index A accounts for the ratio of the total quantity of client properties in client's sample data
Example, ai indicate that client properties i reaches the general of threshold value by client's quality affective characteristics index of radix of the quantity of client's index A
Rate, n indicate the number of the client properties of corresponding client's index A.Server is by the entropy gain weighted sum of multiple client's indexs
Obtain the corresponding entropy gain of respective client indicator combination.Server increases according to the corresponding entropy of each client's indicator combination
Benefit and quality label, are trained the first preset model by tagsort algorithm, obtain tagsort model.Tagsort
Algorithm can be GBDT (Gradient Boost Decision Tree, gradient promote tree algorithm) and (Logistic
Regression, logistic regression algorithm) combination.
Server is based on the training of client's sample data and obtains Fusion Features model.Specifically, if customer data is open network
What platform crawled, heterogeneous networks platform may be different to the naming method of same client's index, in order to reduce name difference
Influence to model training, server carry out synonymous extension process to each client's index, obtain each client's indicator combination point
Not corresponding extended counter combination.Server obtains the corresponding synonym of each participle in client's index respectively, will participle with it is right
The synonym answered forms extension set of words.There are corresponding extension set of words, such as client's indicator combination A to be for each participle
{ a, b, c }, then each client's index in client's indicator combination is there are corresponding extension set of words, and such as a pairs of client's index
The extension set of words answered is { a, a1, a2 }.Server according to the sequence occurred with client's index each in client's indicator combination,
A word is arbitrarily selected from the corresponding extension set of words of each client's index, forms an extended counter collection in order
It closes.When selecting different words from extension set of words, then different extended counter set, different extended counters are formed
Collection is combined into extended counter combination.Server is combined according to each extended counter and corresponding sentiment analysis is as a result, pass through
Feature Fusion Algorithm is trained the second preset model, obtains Fusion Features model.Feature Fusion Algorithm can be random gloomy
Woods algorithm etc..
It is initially formed the corresponding extension set of words of each client's index, then each client is formed by extension set of words and is referred to
Mark combines corresponding extended counter combination, substantially increases the divergence of client's index, each client's index expression after extension
The meaning same or similar with original client's index, improves the effective coverage range of client's index, thus subsequent defeated
After entering the Fusion Features model trained, emotional feature analysis accuracy can be improved.
Tagsort model and Fusion Features models fitting are obtained corresponding submodel by server.It is specific real at one
Apply in example, server to Logic Regression Models, GBDT (Gradient Boost Decision Tree, nonlinear model),
(Logistic Regression, Logic Regression Models), Random Forest model carry out linear fit, obtain submodel.For example,
Submodel=Logic Regression Models * W1+GBDT*W2+LR*W3+ Random Forest model * W4.Wherein, Wi is weight factor.It is different
There are ROC (receiver operating characteristic curve, Receiver Operating Characteristics) differences for Type model
Property, here different type models fitting, it can be improved client's emotional feature analysis accuracy.
Each submodel has corresponding time tag.Time tag can be to be generated according to the building period of submodel, such as
2017,20170317 etc..Server is according to time attenuation functionIt determines the contribution rate of each submodel, that is, determines multiple
The corresponding weight of submodel.Wherein, Δ t is the time difference of time tag and current time；T is that Best Times are long
Degree.For example, the time difference Δ t=1 in time tag " 2017 " year and current time " 2018 "；It is corresponding that T can be sliding window function
Time span, that is, the quantity of the submodel screened.It is readily appreciated that, the submodel remoter from present period is gone through using relatively early
The training of history customer data, the reference significance (i.e. contribution rate) for analyzing present period client's affective characteristics are smaller.In other words, when
Between difference it is bigger, corresponding submodel is smaller to the contribution rate of attributional analysis, so as to determine multiple sons based on time attenuation function
The weight of model.
Server is based on multiple submodels and corresponding weight carries out linear regression operation, obtains the first mould
Type.In order to improve the accuracy of the first model, server is trained reinforcing to the first model.Specifically, server acquisition is worked as
Client's sample data of multiple clients in the preceding monitoring period.Client's sample data has corresponding tag along sort.Client's sample number
According to the information including the multiple dimensions of client, such as age, occupation, kinsfolk.Server is by client's sample in current monitor period
Notebook data inputs the first model, obtains middle classification result.Server calculates the difference of middle classification result and tag along sort, root
The first model is adjusted according to difference, obtains the second model.
In the present embodiment, due to constructing emotional feature analysis model in advance using client's sample data, it is based on analysis model
Only the affective characteristics of respective client can need to be quickly obtained using customer data as ginseng is entered, the multiple dimensions of client can also be comprehensively considered
The information of degree also improves client's emotional feature analysis accuracy rate to not only improve client's emotional feature analysis efficiency.
It should be understood that although each step in the flow chart of Fig. 2~Fig. 4 is successively shown according to the instruction of arrow,
But these steps are not that the inevitable sequence according to arrow instruction successively executes.Unless expressly state otherwise herein, these
There is no stringent sequences to limit for the execution of step, these steps can execute in other order.Moreover, in Fig. 2~Fig. 4
At least part step may include that perhaps these sub-steps of multiple stages or stage are not necessarily same to multiple sub-steps
One moment executed completion, but can execute at different times, and the execution in these sub-steps or stage sequence is also not necessarily
Be successively carry out, but can at least part of the sub-step or stage of other steps or other steps in turn or
Alternately execute.
In one embodiment, as shown in figure 5, providing a kind of prediction meanss of client's stability, comprising: information analysis
Module 502, customer analysis module 504 and influence prediction module 506, in which:
Information analysis module 502, multiple target informations for the publication of monitoring network platform；It is corresponding to obtain target customer
Customer data；Customer data includes product identification；The corresponding public opinion index of product identification is calculated based on target information.
Customer analysis module 504, for counting access data of the target customer in the monitoring period to target information；According to visitor
User data determines the affective characteristics of target customer.
Influence prediction module 506, for by public opinion index, access number accordingly and affective characteristics input preset informational influence
Prediction model exports the stability parameter of target customer.
In one embodiment, information analysis module 502 is also used to the raw information of monitoring network platform publication；To original
Information carries out word segmentation processing, obtains the corresponding information labels of each raw information；Obtain multiple target keywords, identification information mark
Whether label include target keyword；If so, corresponding raw information is labeled as target information.
In one embodiment, information analysis module 502 is also used to split target information, obtains multiple short essays
This；Product identification is extracted in short text, and product identification is associated with corresponding short text；Utilize preset the analysis of public opinion model meter
Calculate the corresponding affection index of each short text；Determine the corresponding influence power weight of multiple short texts；According to associated short
The affection index and influence power weight of text calculate corresponding product and identify corresponding public opinion index.
In one embodiment, customer analysis module 504 is also used to obtain the identification field of target customer, by identification field
It is sent to the network platform；Receive the associated access data that the network platform is returned according to identification field；It is mentioned in associated access data
Take message reference field, information collection field and information forwarding field；Based on message reference field, information collection field and information
Forward statistics target customer in information access volume, information collection amount and the information transfer amount in monitoring period.
In one embodiment, customer analysis module 504 is also used to obtain multiple submodels, determines multiple submodel difference
Corresponding weight；According to multiple submodels and corresponding weight, the first model is generated；Obtain client's sample number
According to and corresponding affective tag；Client's sample data is inputted into the first model, obtains intermediate sentiment analysis result；It calculates intermediate
The difference of sentiment analysis result and affective tag is adjusted the first model according to difference, obtains the second model；By client's number
According to the second model is inputted, the affective characteristics of target customer are exported.
In one embodiment, informational influence prediction model is obtained using deep neural network model training；Influence prediction
Module 506 is also used to obtain client characteristics matrix to public opinion index, access data and affective characteristics pretreatment；According to client spy
Sign matrix obtains input layer sequence；Input layer sequence is projected, the corresponding hidden node of the first hidden layer is obtained
Sequence, using the first hidden layer as currently processed hidden layer；Obtain currently processed hidden layer correspond to each neuron node weight and partially
Difference；It is non-thread according to the corresponding hidden node sequence of currently processed hidden layer and the weight and deviation of each neuron node, use
Property maps to obtain the hidden node sequence of next hidden layer；It is iterated next hidden layer as currently processed hidden layer, until output
Layer；Obtain the corresponding stability parameter of target customer of output layer output.
The specific of prediction meanss about client's stability limits the prediction that may refer to above for client's stability
The restriction of method, details are not described herein.Modules in the prediction meanss of above-mentioned client's stability can be fully or partially through
Software, hardware and combinations thereof are realized.Above-mentioned each module can be embedded in the form of hardware or independently of the place in computer equipment
It manages in device, can also be stored in a software form in the memory in computer equipment, in order to which processor calls execution or more
The corresponding operation of modules.
In one embodiment, a kind of computer equipment is provided, which can be server, internal junction
Composition can be as shown in Figure 6.The computer equipment include by system bus connect processor, memory, network interface and
Database.Wherein, the processor of the computer equipment is for providing calculating and control ability.The memory packet of the computer equipment
Include non-volatile memory medium, built-in storage.The non-volatile memory medium is stored with operating system, computer program and data
Library.The built-in storage provides environment for the operation of operating system and computer program in non-volatile memory medium.The calculating
The database of machine equipment is for storing customer data.The network interface of the computer equipment is used to pass through network with external terminal
Connection communication.A kind of prediction technique of client's stability is realized when the computer program is executed by processor.
It will be understood by those skilled in the art that structure shown in Fig. 6, only part relevant to application scheme is tied
The block diagram of structure does not constitute the restriction for the computer equipment being applied thereon to application scheme, specific computer equipment
It may include perhaps combining certain components or with different component layouts than more or fewer components as shown in the figure.
A kind of computer readable storage medium is stored thereon with computer program, when computer program is executed by processor
The step of prediction technique of the client's stability provided in any one embodiment of the application is provided.
Those of ordinary skill in the art will appreciate that realizing all or part of the process in above-described embodiment method, being can be with
Instruct relevant hardware to complete by computer program, computer program to can be stored in a non-volatile computer readable
It takes in storage medium, the computer program is when being executed, it may include such as the process of the embodiment of above-mentioned each method.Wherein, this Shen
Please provided by any reference used in each embodiment to memory, storage, database or other media, may each comprise
Non-volatile and/or volatile memory.Nonvolatile memory may include read-only memory (ROM), programming ROM
(PROM), electrically programmable ROM (EPROM), electrically erasable ROM (EEPROM) or flash memory.Volatile memory may include
Random access memory (RAM) or external cache.By way of illustration and not limitation, RAM is available in many forms,
Such as static state RAM (SRAM), dynamic ram (DRAM), synchronous dram (SDRAM), double data rate sdram (DDRSDRAM), enhancing
Type SDRAM (ESDRAM), synchronization link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM
(RDRAM), direct memory bus dynamic ram (DRDRAM) and memory bus dynamic ram (RDRAM) etc..
Each technical characteristic of above embodiments can be combined arbitrarily, for simplicity of description, not to above-described embodiment
In each technical characteristic it is all possible combination be all described, as long as however, the combination of these technical characteristics be not present lance
Shield all should be considered as described in this specification.
Above embodiments only express the several embodiments of the application, and the description thereof is more specific and detailed, but can not
Therefore it is construed as limiting the scope of the patent.It should be pointed out that for those of ordinary skill in the art,
Under the premise of not departing from the application design, various modifications and improvements can be made, these belong to the protection scope of the application.
Therefore, the scope of protection shall be subject to the appended claims for the application patent.
Claims (10)
1. a kind of prediction technique of client's stability, which comprises
Multiple target informations of monitoring network platform publication；
Obtain the corresponding customer data of target customer；The customer data includes product identification；
The corresponding public opinion index of the product identification is calculated based on the target information；
Count access data of the target customer in the monitoring period to target information；
The affective characteristics of the target customer are determined according to the customer data；
Accordingly and the affective characteristics input preset informational influence prediction model by the public opinion index, the access number, defeated
The stability parameter of the target customer out.
2. the method according to claim 1, wherein the monitoring network platform publication multiple target informations,
Include:
The raw information of monitoring network platform publication；
Word segmentation processing is carried out to the raw information, obtains the corresponding information labels of each raw information；
Multiple target keywords are obtained, identify whether the information labels include the target keyword；
If so, corresponding raw information is labeled as target information.
3. the method according to claim 1, wherein described calculate the product identification based on the target information
Corresponding public opinion index, comprising:
The target information is split, multiple short texts are obtained；
Product identification is extracted in the short text, the product identification is associated with corresponding short text；
The corresponding affection index of each short text is calculated using preset the analysis of public opinion model；
Determine the corresponding influence power weight of multiple short texts；
According to the affection index of associated short text and influence power weight, calculates corresponding product and identify corresponding public opinion index.
4. the method according to claim 1, wherein the statistics target customer is in the monitoring period to target
The access data of information, comprising:
The identification field for obtaining target customer, is sent to the network platform for the identification field；
Receive the associated access data that the network platform is returned according to the identification field；
Message reference field, information collection field and information are extracted in the associated access data forwards field；
Based on target customer described in the message reference field, information collection field and information forwarding statistics in the monitoring period
Information access volume, information collection amount and information transfer amount.
5. the method according to claim 1, wherein described determine the target customer according to the customer data
Affective characteristics, comprising:
Multiple submodels are obtained, determine the corresponding weight of multiple submodels；
According to multiple submodels and corresponding weight, the first model is generated；
Obtain client's sample data and corresponding affective tag；
Client's sample data is inputted into first model, obtains intermediate sentiment analysis result；
The difference for calculating intermediate the sentiment analysis result and the affective tag, according to the difference to first model into
Row adjustment, obtains the second model；
The customer data is inputted into second model, exports the affective characteristics of the target customer.
6. the method according to claim 1, wherein the informational influence prediction model uses deep neural network
Model training obtains；It is described that the public opinion index, access data and affective characteristics are inputted into preset informational influence prediction model,
Export the stability parameter of the target customer, comprising:
To the public opinion index, access data and affective characteristics pretreatment, client characteristics matrix is obtained；
Input layer sequence is obtained according to the client characteristics matrix；
The input layer sequence is projected, the corresponding hidden node sequence of the first hidden layer is obtained, it is hidden by described first
Layer is used as currently processed hidden layer；
Obtain weight and deviation that the currently processed hidden layer corresponds to each neuron node；According to the currently processed hidden layer pair
The weight and deviation of the hidden node sequence and each neuron node answered obtain the hidden of next hidden layer using Nonlinear Mapping
Node layer sequence；
It is iterated next hidden layer as currently processed hidden layer, until output layer；Obtain the mesh of the output layer output
Mark the corresponding stability parameter of client.
7. a kind of prediction meanss of client's stability, which is characterized in that described device includes:
Information analysis module, multiple target informations for the publication of monitoring network platform；Obtain the corresponding client's number of target customer
According to；The customer data includes product identification；The corresponding public opinion index of the product identification is calculated based on the target information；
Customer analysis module, for counting access data of the target customer in the monitoring period to target information；According to described
Customer data determines the affective characteristics of the target customer；
Influence prediction module, for by the public opinion index, the access number accordingly and the affective characteristics input preset letter
Retire into private life and ring prediction model, exports the stability parameter of the target customer.
8. device according to claim 7, which is characterized in that the information analysis module is also used to the target information
It is split, obtains multiple short texts；Product identification is extracted in the short text, by the product identification and corresponding short text
Association；The corresponding affection index of each short text is calculated using preset the analysis of public opinion model；Determine multiple short essays
This corresponding influence power weight；According to the affection index of associated short text and influence power weight, corresponding product is calculated
Identify corresponding public opinion index.
9. a kind of computer equipment, including memory and processor, the memory are stored with computer program, feature exists
In the step of processor realizes any one of claims 1 to 6 the method when executing the computer program.
10. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the computer program
The step of method described in any one of claims 1 to 6 is realized when being executed by processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811025974.XA CN109376237B (en) | 2018-09-04 | 2018-09-04 | Client stability prediction method, device, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811025974.XA CN109376237B (en) | 2018-09-04 | 2018-09-04 | Client stability prediction method, device, computer equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109376237A true CN109376237A (en) | 2019-02-22 |
| CN109376237B CN109376237B (en) | 2024-05-28 |
Family
ID=65405116
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811025974.XA Active CN109376237B (en) | 2018-09-04 | 2018-09-04 | Client stability prediction method, device, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109376237B (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110083759A (en) * | 2019-03-15 | 2019-08-02 | 深圳壹账通智能科技有限公司 | Public opinion information crawler method, apparatus, computer equipment and storage medium |
| CN110348914A (en) * | 2019-07-19 | 2019-10-18 | 中国银行股份有限公司 | Customer churn data analysing method and device |
| CN110688553A (en) * | 2019-08-13 | 2020-01-14 | 平安科技（深圳）有限公司 | Information pushing method and device based on data analysis, computer equipment and storage medium |
| CN110956303A (en) * | 2019-10-12 | 2020-04-03 | 未鲲(上海)科技服务有限公司 | Information prediction method, device, terminal and readable storage medium |
| CN111158732A (en) * | 2019-12-23 | 2020-05-15 | 中国平安人寿保险股份有限公司 | Access data processing method and device, computer equipment and storage medium |
| CN111506727A (en) * | 2020-04-16 | 2020-08-07 | 腾讯科技（深圳）有限公司 | Text content category acquisition method and device, computer equipment and storage medium |
| US20200327582A1 (en) * | 2019-04-15 | 2020-10-15 | Yandex Europe Ag | Method and system for determining result for task executed in crowd-sourced environment |
| CN111950623A (en) * | 2020-08-10 | 2020-11-17 | 中国平安人寿保险股份有限公司 | Data stability monitoring method and device, computer equipment and medium |
| CN112836749A (en) * | 2021-02-03 | 2021-05-25 | 中国工商银行股份有限公司 | System resource adjusting method, device and equipment |
| CN113643060A (en) * | 2021-08-12 | 2021-11-12 | 工银科技有限公司 | Product price prediction method and device |
| CN113657635A (en) * | 2020-05-12 | 2021-11-16 | 中国移动通信集团湖南有限公司 | Method for predicting communication user loss and electronic equipment |
| US11727329B2 (en) | 2020-02-14 | 2023-08-15 | Yandex Europe Ag | Method and system for receiving label for digital task executed within crowd-sourced environment |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130103667A1 (en) * | 2011-10-17 | 2013-04-25 | Metavana, Inc. | Sentiment and Influence Analysis of Twitter Tweets |
| CN103544255A (en) * | 2013-10-15 | 2014-01-29 | 常州大学 | Text semantic relativity based network public opinion information analysis method |
| CN104462096A (en) * | 2013-09-13 | 2015-03-25 | 北大方正集团有限公司 | Public opinion monitoring and analysis method and device |
| US20160162582A1 (en) * | 2014-12-09 | 2016-06-09 | Moodwire, Inc. | Method and system for conducting an opinion search engine and a display thereof |
| US20160267377A1 (en) * | 2015-03-12 | 2016-09-15 | Staples, Inc. | Review Sentiment Analysis |
| CN106097114A (en) * | 2016-06-27 | 2016-11-09 | 江苏易乐网络科技有限公司 | A kind of game public sentiment monitoring and player's behavior analysis cloud service platform |
| CN106126558A (en) * | 2016-06-16 | 2016-11-16 | 东软集团股份有限公司 | A kind of public sentiment monitoring method and device |
| CN106649270A (en) * | 2016-12-19 | 2017-05-10 | 四川长虹电器股份有限公司 | Public opinion monitoring and analyzing method |
| CN107038178A (en) * | 2016-08-03 | 2017-08-11 | 平安科技（深圳）有限公司 | The analysis of public opinion method and apparatus |
| CN107341270A (en) * | 2017-07-28 | 2017-11-10 | 东北大学 | Towards the user feeling influence power analysis method of social platform |
| CN107391493A (en) * | 2017-08-04 | 2017-11-24 | 青木数字技术股份有限公司 | A kind of public feelings information extracting method, device, terminal device and storage medium |
| CN107908619A (en) * | 2017-11-15 | 2018-04-13 | 中国平安人寿保险股份有限公司 | Processing method, device, terminal and computer-readable storage medium based on public sentiment monitoring |
-
2018
- 2018-09-04 CN CN201811025974.XA patent/CN109376237B/en active Active
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130103667A1 (en) * | 2011-10-17 | 2013-04-25 | Metavana, Inc. | Sentiment and Influence Analysis of Twitter Tweets |
| CN104462096A (en) * | 2013-09-13 | 2015-03-25 | 北大方正集团有限公司 | Public opinion monitoring and analysis method and device |
| CN103544255A (en) * | 2013-10-15 | 2014-01-29 | 常州大学 | Text semantic relativity based network public opinion information analysis method |
| US20160162582A1 (en) * | 2014-12-09 | 2016-06-09 | Moodwire, Inc. | Method and system for conducting an opinion search engine and a display thereof |
| US20160267377A1 (en) * | 2015-03-12 | 2016-09-15 | Staples, Inc. | Review Sentiment Analysis |
| CN106126558A (en) * | 2016-06-16 | 2016-11-16 | 东软集团股份有限公司 | A kind of public sentiment monitoring method and device |
| CN106097114A (en) * | 2016-06-27 | 2016-11-09 | 江苏易乐网络科技有限公司 | A kind of game public sentiment monitoring and player's behavior analysis cloud service platform |
| CN107038178A (en) * | 2016-08-03 | 2017-08-11 | 平安科技（深圳）有限公司 | The analysis of public opinion method and apparatus |
| CN106649270A (en) * | 2016-12-19 | 2017-05-10 | 四川长虹电器股份有限公司 | Public opinion monitoring and analyzing method |
| CN107341270A (en) * | 2017-07-28 | 2017-11-10 | 东北大学 | Towards the user feeling influence power analysis method of social platform |
| CN107391493A (en) * | 2017-08-04 | 2017-11-24 | 青木数字技术股份有限公司 | A kind of public feelings information extracting method, device, terminal device and storage medium |
| CN107908619A (en) * | 2017-11-15 | 2018-04-13 | 中国平安人寿保险股份有限公司 | Processing method, device, terminal and computer-readable storage medium based on public sentiment monitoring |
Non-Patent Citations (1)
| Title |
|---|
| 陈振春;刘学军;李斌: "基于内容和信任度的舆情扩散研究", 《计算机应用与软件》, vol. 34, no. 10, pages 59 - 65 * |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110083759A (en) * | 2019-03-15 | 2019-08-02 | 深圳壹账通智能科技有限公司 | Public opinion information crawler method, apparatus, computer equipment and storage medium |
| US11727336B2 (en) * | 2019-04-15 | 2023-08-15 | Yandex Europe Ag | Method and system for determining result for task executed in crowd-sourced environment |
| US20200327582A1 (en) * | 2019-04-15 | 2020-10-15 | Yandex Europe Ag | Method and system for determining result for task executed in crowd-sourced environment |
| CN110348914A (en) * | 2019-07-19 | 2019-10-18 | 中国银行股份有限公司 | Customer churn data analysing method and device |
| CN110688553A (en) * | 2019-08-13 | 2020-01-14 | 平安科技（深圳）有限公司 | Information pushing method and device based on data analysis, computer equipment and storage medium |
| CN110956303A (en) * | 2019-10-12 | 2020-04-03 | 未鲲(上海)科技服务有限公司 | Information prediction method, device, terminal and readable storage medium |
| CN111158732A (en) * | 2019-12-23 | 2020-05-15 | 中国平安人寿保险股份有限公司 | Access data processing method and device, computer equipment and storage medium |
| CN111158732B (en) * | 2019-12-23 | 2024-04-02 | 中国平安人寿保险股份有限公司 | Access data processing method, device, computer equipment and storage medium |
| US11727329B2 (en) | 2020-02-14 | 2023-08-15 | Yandex Europe Ag | Method and system for receiving label for digital task executed within crowd-sourced environment |
| CN111506727B (en) * | 2020-04-16 | 2023-10-03 | 腾讯科技（深圳）有限公司 | Text content category acquisition method, apparatus, computer device and storage medium |
| CN111506727A (en) * | 2020-04-16 | 2020-08-07 | 腾讯科技（深圳）有限公司 | Text content category acquisition method and device, computer equipment and storage medium |
| CN113657635A (en) * | 2020-05-12 | 2021-11-16 | 中国移动通信集团湖南有限公司 | Method for predicting communication user loss and electronic equipment |
| CN113657635B (en) * | 2020-05-12 | 2023-10-27 | 中国移动通信集团湖南有限公司 | Method for predicting loss of communication user and electronic equipment |
| CN111950623A (en) * | 2020-08-10 | 2020-11-17 | 中国平安人寿保险股份有限公司 | Data stability monitoring method and device, computer equipment and medium |
| CN111950623B (en) * | 2020-08-10 | 2023-11-14 | 中国平安人寿保险股份有限公司 | Data stability monitoring method, device, computer equipment and medium |
| CN112836749A (en) * | 2021-02-03 | 2021-05-25 | 中国工商银行股份有限公司 | System resource adjusting method, device and equipment |
| CN113643060A (en) * | 2021-08-12 | 2021-11-12 | 工银科技有限公司 | Product price prediction method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109376237B (en) | 2024-05-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109376237A (en) | Prediction technique, device, computer equipment and the storage medium of client's stability | |
| CN109241427B (en) | Information pushing method, device, computer equipment and storage medium | |
| CN108874992B (en) | Public opinion analysis method, system, computer equipment and storage medium | |
| CN111382361B (en) | Information pushing method, device, storage medium and computer equipment | |
| CN111444428A (en) | Information recommendation method and device based on artificial intelligence, electronic equipment and storage medium | |
| CN107590224B (en) | Big data based user preference analysis method and device | |
| CN109272396A (en) | Customer risk method for early warning, device, computer equipment and medium | |
| CN111859160B (en) | Session sequence recommendation method and system based on graph neural network | |
| US11574326B2 (en) | Identifying topic variances from digital survey responses | |
| CN110110233B (en) | Information processing method, device, medium and computing equipment | |
| CN109543925A (en) | Risk Forecast Method, device, computer equipment and storage medium based on machine learning | |
| CN108563734A (en) | Institutional information querying method, device, computer equipment and storage medium | |
| CN113157898B (en) | Method and device for recommending candidate questions, computer equipment and storage medium | |
| CN112784168B (en) | Information push model training method and device, information push method and device | |
| CN112995414B (en) | Behavior quality inspection method, device, equipment and storage medium based on voice call | |
| CN113051468B (en) | Movie recommendation method and system based on knowledge graph and reinforcement learning | |
| CN109947902A (en) | A kind of data query method, apparatus and readable medium | |
| CN112633690B (en) | Service personnel information distribution method, device, computer equipment and storage medium | |
| CN114371946B (en) | Information push method and information push server based on cloud computing and big data | |
| CN108809718A (en) | Network Access Method, system, computer equipment based on virtual resource and medium | |
| CN112685639A (en) | Activity recommendation method and device, computer equipment and storage medium | |
| CN109325781B (en) | Customer quality analysis method, apparatus, computer device and storage medium | |
| CN113065880A (en) | Group dissatisfaction user identification method, device, equipment and storage medium | |
| CN114693409A (en) | Product matching method, device, computer equipment, storage medium and program product | |
| CN110162689B (en) | Information pushing method, device, computer equipment and storage medium |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |