CN105824806B - A kind of quality evaluating method and device of public's account - Google Patents
A kind of quality evaluating method and device of public's account Download PDFInfo
- Publication number
- CN105824806B CN105824806B CN201610420186.5A CN201610420186A CN105824806B CN 105824806 B CN105824806 B CN 105824806B CN 201610420186 A CN201610420186 A CN 201610420186A CN 105824806 B CN105824806 B CN 105824806B
- Authority
- CN
- China
- Prior art keywords
- public
- data
- account
- regression model
- sample data
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 38
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 48
- 238000012549 training Methods 0.000 claims description 84
- 238000012360 testing method Methods 0.000 claims description 59
- 235000010627 Phaseolus vulgaris Nutrition 0.000 claims description 42
- 244000046052 Phaseolus vulgaris Species 0.000 claims description 42
- 238000011156 evaluation Methods 0.000 claims description 18
- 238000004458 analytical method Methods 0.000 claims description 16
- 238000005457 optimization Methods 0.000 claims description 16
- 230000000694 effects Effects 0.000 claims description 15
- 238000012545 processing Methods 0.000 claims description 13
- 238000012216 screening Methods 0.000 claims description 13
- 238000013441 quality evaluation Methods 0.000 claims description 11
- 230000003542 behavioural effect Effects 0.000 claims description 7
- 230000002452 interceptive effect Effects 0.000 claims description 7
- 238000012106 screening analysis Methods 0.000 claims description 5
- 238000002372 labelling Methods 0.000 claims description 4
- 238000001303 quality assessment method Methods 0.000 claims description 4
- 230000008859 change Effects 0.000 claims description 3
- 230000003993 interaction Effects 0.000 claims description 2
- 238000004364 calculation method Methods 0.000 claims 1
- 238000010586 diagram Methods 0.000 description 9
- 230000008569 process Effects 0.000 description 7
- 239000013598 vector Substances 0.000 description 6
- 230000006399 behavior Effects 0.000 description 4
- 238000010801 machine learning Methods 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 238000004891 communication Methods 0.000 description 3
- 238000007477 logistic regression Methods 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 210000004209 hair Anatomy 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000009412 basement excavation Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 210000004709 eyebrow Anatomy 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000002085 persistent effect Effects 0.000 description 1
- 239000000843 powder Substances 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 238000012958 reprocessing Methods 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000011144 upstream manufacturing Methods 0.000 description 1
- 230000003442 weekly effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/205—Parsing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/205—Parsing
- G06F40/216—Parsing using statistical methods
-
- 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
- G06Q50/01—Social networking
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Databases & Information Systems (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Business, Economics & Management (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Economics (AREA)
- Computing Systems (AREA)
- Probability & Statistics with Applications (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of quality evaluating methods and device of public's account, make for realizing the quality to public's account and efficiently and accurately evaluating.The embodiment of the present invention provides a kind of quality evaluating method of public's account, including：Obtain by regression algorithm from the acquistion of sample data middle school to regression model, the sample data includes：Multiple public's accounts in public platform and the corresponding indicator-specific statistics data of the multiple public's account；Public's account to be evaluated is input in the regression model, fractional value prediction is carried out to public's account to be evaluated by the regression model；Obtain mass fraction of the fractional value exported after the forecast of regression model as public's account to be evaluated.
Description
Technical field
The present invention relates to field of computer technology more particularly to a kind of quality evaluating methods and device of public's account.
Background technology
Public platform is eventually positioned in service to the public, and user gives public platform very high expectation.The public is flat
Platform possesses a large amount of user, and public platform wants to constantly excavate the value of oneself user, is that this platform increase is more high-quality
Content, create preferably viscosity, formed a different ecological circulation, this is public platform in the more important of developing stage
Direction.Various tissues, enterprise and individual can register into public platform, and different registration users is different
Public's account, can be carried out on public platform from media activity using public's account, be exactly in simple terms carry out it is one-to-many
Media behavioral activity, such as businessman by apply public platform in public's account, can show the official website of businessman, various work
It moves, so as to form the marketing mode of on-line off-line interaction.
Public's account is in consequence in public platform, with the release and accumulation of public platform, the public at present
The quantity of account is more and more, even up to tens billion of ranks, but the quality of various public's accounts is irregular, for example exists
Many corpse public account, multiple level marketing public's account etc..With the quick development of public platform, public's account of high quality exists
There is extensive demand under many scenes, but the quality good or not of public's account can not have been made in current public platform
The evaluation of effect.
Invention content
An embodiment of the present invention provides a kind of quality evaluating methods and device of public's account, for realizing to public's account
Quality make effectively evaluating.
In order to solve the above technical problems, the embodiment of the present invention provides following technical scheme：
In a first aspect, the embodiment of the present invention provides a kind of quality evaluating method of public's account, including：
Obtain by regression algorithm from the acquistion of sample data middle school to regression model, the sample data includes：The public
Multiple public's accounts in platform and the corresponding indicator-specific statistics data of the multiple public's account；
Public's account to be evaluated is input in the regression model, by the regression model to described to be evaluated
Public's account carries out fractional value prediction；
Obtain mass fraction of the fractional value exported after the forecast of regression model as public's account to be evaluated.
Second aspect, the embodiment of the present invention also provide a kind of quality evaluation device of public's account, including：
Model acquisition module, for obtains pass through regression algorithm from the acquistion of sample data middle school to regression model, it is described
Sample data includes：Multiple public's accounts in public platform and the corresponding indicator-specific statistics data of the multiple public's account；
Model prediction module passes through the recurrence for public's account to be evaluated to be input in the regression model
Model carries out fractional value prediction to public's account to be evaluated；
Quality assessment module, for obtaining the fractional value exported after the forecast of regression model as the public affairs to be evaluated
The mass fraction of many accounts.
As can be seen from the above technical solutions, the embodiment of the present invention has the following advantages：
In embodiments of the present invention, first obtain by regression algorithm from the acquistion of sample data middle school to regression model,
Sample data includes：Multiple public's accounts in public platform and the corresponding indicator-specific statistics data of multiple public's accounts, then
Public's account to be evaluated is input in regression model, it is pre- to carry out fractional value to public's account to be evaluated by regression model
It surveys, finally obtains mass fraction of the fractional value exported after forecast of regression model as public's account to be evaluated.The present invention is real
Sample data can be extracted to train to obtain regression model from public platform by applying in example, can be to be evaluated by regression model
Public's account of valence carries out fractional value prediction, to obtain the mass fraction of public's account to be evaluated.Since regression model is
It trains to obtain by the sample data extracted from public platform, sample data is in public platform, public platform
Public's account can carry out quality evaluation by the regression model, so as to realize that the quality to public's account is made effectively
Evaluation.
Description of the drawings
To describe the technical solutions in the embodiments of the present invention more clearly, make required in being described below to embodiment
Attached drawing is briefly described, it should be apparent that, drawings in the following description are only some embodiments of the invention, for
For those skilled in the art, other drawings may also be obtained based on these drawings.
Fig. 1 is a kind of process blocks schematic diagram of the quality evaluating method of public's account provided in an embodiment of the present invention；
Fig. 2 is a kind of application scenarios schematic diagram of the quality evaluating method of public's account provided in an embodiment of the present invention；
Fig. 3-a are a kind of composed structure schematic diagram of the quality evaluation device of public's account provided in an embodiment of the present invention；
Fig. 3-b are the composed structure signal of the quality evaluation device of another public's account provided in an embodiment of the present invention
Figure；
Fig. 3-c are a kind of composed structure schematic diagram of model training module provided in an embodiment of the present invention；
Fig. 3-d are the composed structure schematic diagram of another model training module provided in an embodiment of the present invention；
Fig. 3-e are a kind of composed structure schematic diagram of first model processing modules provided in an embodiment of the present invention；
Fig. 4 is that the quality evaluating method of public's account provided in an embodiment of the present invention shows applied to the composed structure of server
It is intended to.
Specific implementation mode
An embodiment of the present invention provides a kind of quality evaluating methods and device of public's account, for realizing to public's account
Quality make effectively evaluating.
In order to make the invention's purpose, features and advantages of the invention more obvious and easy to understand, below in conjunction with the present invention
Attached drawing in embodiment, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that disclosed below
Embodiment be only a part of the embodiment of the present invention, and not all embodiments.Based on the embodiments of the present invention, this field
The every other embodiment that technical staff is obtained, shall fall within the protection scope of the present invention.
Term " comprising " and " having " in description and claims of this specification and above-mentioned attached drawing and they
Any deformation, it is intended that it includes so as to a series of process comprising units, method, system, product or to set to cover non-exclusive
It is standby to be not necessarily limited to those units, but may include not listing clearly or solid for these processes, method, product or equipment
The other units having.
It is described in detail separately below.
One embodiment of the quality evaluating method of public's account of the present invention, specifically can be applied to in public platform
Public's account carries out quality evaluation, which can be the account in social network sites, can also be the account in scene of game
Number, it can also be the account paid in scene, do not limit herein.Refering to Figure 1, provided by one embodiment of the present invention
The quality evaluating method of public's account, may include steps of：
101, obtain by regression algorithm from the acquistion of sample data middle school to regression model, sample data includes：The public
Multiple public's accounts in platform and the corresponding indicator-specific statistics data of multiple public's accounts.
In embodiments of the present invention, the regression model that can be used for predicting public's account quality score, this hair are got first
Bright embodiment predicts public's account quality score using regression model, the realization method of the regression model can there are many, example
Such as, logistic regression (full name in English:Logist Regression, English abbreviation:LR) model can also be that random forest returns
(full name in English：Random Frorest Regressor, English abbreviation：RFReg) model, stochastic gradient descent (full name in English：
Stochastic Gradient Descent Regressor, English abbreviation：SGD Reg), (English is complete for support vector regression
Claim：Support Vector Regression, English abbreviation：SVR) model etc., specific implementation does not limit.The present invention
Regression model includes from the acquistion of sample data middle school to, wherein sample data by regression algorithm in embodiment：In public platform
Multiple public's accounts and the corresponding indicator-specific statistics data of multiple public's accounts.Record has multiple public's accounts in public platform
Number, and each indicator-specific statistics data of public's account generation, sample data is extracted from public platform, passes through the sample number
It can be used for carrying out quality evaluation to public's account in public platform according to the regression model come is trained.
In some embodiments of the invention, step 101 obtain by regression algorithm from the acquistion of sample data middle school to
Before regression model, the quality evaluating method of public's account provided in an embodiment of the present invention can also include the following steps：
A1, multiple public's accounts and the corresponding indicator-specific statistics data of multiple public's accounts are got from public platform；
A2, using multiple public's accounts and the corresponding indicator-specific statistics data of multiple public's accounts as sample data, pass through
Regression algorithm is trained study, the regression model that output study obtains to regression model.
Wherein, in public platform record have a large amount of public's account, can from these public's accounts the selected part public
Account is used for sample data, further includes the corresponding index of multiple public's accounts in addition to including multiple public's accounts in sample data
Statistical data, the indicator-specific statistics data of public's account refer to be collected into from public platform respectively for multiple indexs with the public affairs
The related data of many accounts.The specific targets of public's account setting can be obtained according in public platform in practical applications
Statistical data.After getting sample data, regression algorithm can be used to be trained study to regression model, output learns
The regression model arrived.Wherein, workable regression algorithm can there are many, be illustrated below：Logistic regression, branch may be used
It holds the regression algorithms such as vector regression and study is trained to regression model, when the characteristic of regression model meets preset condition
It is required that when can export the obtained regression model of study, corresponding condition requirement can be arranged for different regression models, this
Place does not limit.
In some embodiments of the invention, step A2 is with multiple public's accounts and the corresponding index of multiple public's accounts
Statistical data is trained study, the recurrence mould that output study obtains as sample data, by regression algorithm to regression model
Type can specifically include following steps：
A20, using multiple public's accounts and the corresponding indicator-specific statistics data of multiple public's accounts as sample data, pass through
Regression algorithm is trained study, multiple regression models that output study obtains to multiple regression models respectively.
That is, multiple recurrence moulds can be trained using sample data in model training stage in the embodiment of the present invention
Type can then export multiple regression models that study obtains.Under this realization scene, step 101 acquisition passes through regression algorithm
From the acquistion of sample data middle school to regression model, can specifically include following steps：
1010, evaluation is carried out respectively to multiple regression models, assessment effect is selected from multiple regression models
The best regression model of fruit is as the regression model got.
Under the realization scene for executing step A20, the embodiment of the present invention can execute step 1010, to multiple regression models
Evaluation is carried out respectively, and Evaluated effect best regression model is selected from multiple regression models as getting
Regression model.Wherein it is possible to which using effect appraisal procedure assesses multiple regression models respectively, can be used for example
Effect evaluation method is least mean-square error (full name in English：Mean squared error, English abbreviation：MSE), can also adopt
With squared correlation coefficient (English name：Squared Correlation Coefficient), it does not limit herein specifically.From
The regression model selected in multiple regression models can be as the model used in prediction scoring in subsequent step 102.By more
The selection of a regression model, the regression model that Evaluated effect can be used best, to improve the quality score to public's account
The accuracy of prediction.
In some embodiments of the invention, the corresponding indicator-specific statistics data of public's account may include：Operation indicator is united
It counts, bean vermicelli indicator-specific statistics data, article indicator-specific statistics data and interactive indicator-specific statistics data.Wherein, operation indicator counts
Data refer to being directed to the data being collected into the operation indicator set by platform operation from public platform, which can refer to
Be the original degree of article, plagiarism ratio of public's account etc., operation indicator can be configured according to the specific implementation of public platform.
Bean vermicelli indicator-specific statistics data refer to the data being collected into from the bean vermicelli index of public's account in public platform, and bean vermicelli index can be with
Refer to the bean vermicelli data that public's account is possessed, such as the active degree etc. of the distribution of grades of bean vermicelli quantity, bean vermicelli, bean vermicelli.
Article indicator-specific statistics data also refer to the data being collected into from the article index that public platform is arranged for public's account, should
The article that article index also refers to public's account delivers the data etc. that data, the data that article is read, article are forwarded.
Interactive indicator-specific statistics data also refer to the data that public's account and the interactive index of bean vermicelli are collected into from public platform, example
The message data of message data, bean vermicelli message that such as public's account is sent.Refer to for operation indicator, the bean vermicelli of foregoing description
The data that mark, article index, interactive index are collected into respectively may be constructed the corresponding indicator-specific statistics data of public's account.
In other embodiments of the present invention, the corresponding indicator-specific statistics data of public's account may include：Bean vermicelli is to public affairs
Uplink behavioral indicator statistical data, public's account payment indicator-specific statistics data, the public's account of many accounts disappear to the downlink of bean vermicelli
Cease indicator-specific statistics data.Wherein, uplink behavioral indicator statistical data refers to the behavior that bean vermicelli actively executes from public platform,
Such as bean vermicelli sends messages to public's account, bean vermicelli reads the article that public's account is delivered, and bean vermicelli forwarding public's account is delivered
Article etc..Public's account payment indicator-specific statistics data refer to the payment data that bean vermicelli is transferred accounts to public's account in public platform,
Downstream message indicator-specific statistics data refer to the message data that public's account is replied for bean vermicelli in public platform.For foregoing description
Uplink behavioral indicator, the data that are collected into respectively of payment index, downstream message index may be constructed the corresponding finger of public's account
Mark statistical data.
It should be noted that in practical applications, the corresponding indicator-specific statistics data of public's account recorded in public platform
It can be specifically dependent upon concrete configuration of the public platform to public's account there are many realization method.It is illustrated below, in social activity
In the public platform of application, public's account and its bean vermicelli data and bean vermicelli can be collected from public platform to public's account
Article is read, menu is clicked, sends the indicator-specific statistics data such as message, public's account transmission and receive message data, the public
Data that the article that account is delivered is read or forwarded, corresponding bean vermicelli pass through the data etc. that public's account carries out payment behavior
Etc. the statistical data such as indexs of correlation.
In some embodiments of the invention, step A2 is with multiple public's accounts and the corresponding index of multiple public's accounts
Statistical data is trained study, the recurrence mould that output study obtains as sample data, by regression algorithm to regression model
Type can specifically include following steps：
A21, sample data is divided into two classes, obtains training sample data and test sample data, training sample data packet
It includes：Training public's account and the corresponding indicator-specific statistics data of training public's account, test sample data include：Test public's account
Number and the corresponding indicator-specific statistics data of test public's account, wherein the multiple public's accounts got from public platform are divided into
Two classes：Training public's account and test public's account；
A22, using training sample data, carry out Feature Engineering to regression model by regression algorithm and analyze to obtain to return mould
The fisrt feature data of type have been trained according to the fisrt feature data of regression model to being exported after regression model progress prediction optimization
At regression model；
A23, using test sample data, Feature Engineering analysis is carried out to the regression model that training is completed by regression algorithm
The second feature data of regression model are obtained, after carrying out evaluation and test optimization to regression model according to the second feature data of regression model
The regression model that output study obtains.
Wherein, in step A21, sample data is divided into two classes first, obtains training sample data and test sample number
According to training sample data are for training regression model, and test sample data are the regression model objects to be identified, in reality
In, sample data can be divided according to ratio data, sample data is divided into training sample data and test sample number
According to, for example, can be using the sample data of the 70%-75% of total number of samples as training sample data, remaining sample data is made
For test sample data.
After classifying to sample data, step A22 is executed to training sample data first, using training sample data,
Feature Engineering is carried out by regression algorithm to regression model to analyze to obtain the fisrt feature data of regression model, according to regression model
Fisrt feature data to regression model carry out prediction optimization after export training complete regression model.Wherein, Feature Engineering is
Initial data conversion is characterized, the more preferable practical problem for indicating prediction model processing promotes the accuracy for unknown data.
It is to generate, extract, delete or combine variation with the method for public's account knowledge where target problem either automation
Obtain feature.In the embodiment of the present invention, after carrying out Feature Engineering analysis to training sample data, regression model can be obtained
Fisrt feature data, the fisrt feature data are the feature vectors after being analyzed training sample data by regression algorithm,
Then the regression model trained and completed is exported after carrying out prediction optimization to regression model according to the fisrt feature data of regression model,
Specific prediction optimization mode can be completed in conjunction with the regression algorithm and regression model of use.
Further, in some embodiments of the invention, step A22 uses training sample data, passes through regression algorithm
Feature Engineering is carried out to regression model to analyze to obtain the fisrt feature data of regression model, can specifically include following steps：
A221, the corresponding indicator-specific statistics data of training public's account in training sample data are carried out according to feature importance
Qualitative character list is written in the characteristic screened by screening analysis；
A222, judge whether the characteristic in qualitative character list changes in historical time section, by quality spy
Fisrt feature data of the characteristic exported from high to low according to stability in sign list as regression model.
It wherein, can be to the corresponding indicator-specific statistics data of training public's account according to feature weight in training regression model
The property wanted carries out screening analysis, the characteristic screened can be written in qualitative character list according to significance level, in matter
It may include multiple characteristics in measure feature list, analysis of stability then carried out to the characteristic in qualitative character list again
Analysis, such as a historical time section can be set, whether the characteristic from public platform in qualitative character list is sent out
Variation, using the characteristic exported from high to low according to stability in qualitative character list as the fisrt feature number of regression model
According to.Feature is carried out according to processes such as importance analysis, stability analyses by the characteristic to collection in the embodiment of the present invention
Screening can retain in the qualitative character list finally obtained and predict helpful characteristic to final score value.
After exporting the regression model completed by the training of training sample data, step A23 can be executed, it is complete with training
At regression model test sample data are tested.Specifically, test sample data can be used, pass through regression algorithm pair
The regression model that training is completed carries out Feature Engineering and analyzes to obtain the second feature data of regression model, according to the of regression model
Output learns obtained regression model after two characteristics carry out evaluation and test optimization to regression model.The second feature data are to pass through
Regression algorithm test sample data are analyzed after feature vector, then according to the second feature data of regression model to return
The regression model that output study obtains after returning model to carry out evaluation and test optimization, specific prediction optimization mode can be in conjunction with use
Regression algorithm and regression model are completed.
In some embodiments of the invention, step A22 uses training sample data, by regression algorithm to regression model
Before progress Feature Engineering is analyzed to obtain the fisrt feature data of regression model, the matter of public's account provided in an embodiment of the present invention
Amount evaluation method can also include the following steps：
B1, data mark is carried out respectively to training sample data and test sample data, the training sample after being marked
Test sample data after data and mark；
B2, data screening is carried out respectively to the training sample data after mark and the test sample data after mark.
It wherein, can be according to after being classified to obtain training sample data and test sample data to sample data
Training sample data and test sample data carry out data mark respectively, to distinguish from training sample data and test sample data
Data are marked out, then training sample data is directed to and test sample data carries out data screening respectively.It is illustrated below, it is right
It can be screened according to same ratio according to 0~100 point of each fraction range in the public's account marked, so as to have
The guarantee sample of effect is balanced, such as filters out training sample data about 20,000 altogether, filters out test sample data about 10,000
It is a plurality of.
In some embodiments of the invention, public's account may include：Subscription type public account and service type public's account
Number.It for example, subject of operation is tissue (such as enterprise, media, commonweal organizations), can apply for service type public's account, run
Main body, which is organizations and individuals, can apply for subscription type public's account.It can then be pressed respectively during data mark and screen
Data mark and screening are carried out according to subscription type public account and service type public's account, to realize the equilibrium of sample data.
102, public's account to be evaluated is input in regression model, by regression model to public's account to be evaluated
Carry out fractional value prediction.
In embodiments of the present invention, get by regression algorithm from the acquistion of sample data middle school to regression model it
Afterwards, which can be used for the prediction of quality of public's account, by the description of abovementioned steps it is found that regression model passes through sample
Notebook data is trained using regression algorithm, and the data source of the sample data is multiple public's accounts in public platform, complete
The quality height of specific public's account can be identified at the regression model after training.Specifically, by public's account to be evaluated
It number is input in regression model, then score can be carried out to public's account of the evaluation by the characteristic in the regression model
Value prediction.Wherein, public's account to be evaluated can be public's account, can also refer to multiple public's accounts, at this time
Fractional value prediction can be carried out to multiple public's accounts to be evaluated by regression model.
103, mass fraction of the fractional value exported after forecast of regression model as public's account to be evaluated is obtained.
In embodiments of the present invention, fractional value prediction is carried out to public's account to be evaluated by step 102 regression model
It afterwards, can be using the fractional value exported after forecast of regression model as the mass fraction of public's account to be evaluated.For example, returning mould
The interval of the fractional value exported after type prediction is from 0 to 100, and each public's account to be evaluated passes through forecast of regression model
For the fractional value exported afterwards according to being ranked up from high to low, it is higher that the higher public's account of fractional value represents its quality.
Regression algorithm is passed through from sample data it is found that obtaining first to the description of the embodiment of the present invention by above example
The regression model that middle school's acquistion is arrived, sample data include：Multiple public's accounts in public platform and multiple public's accounts pair
Then public's account to be evaluated is input in regression model, by regression model to be evaluated by the indicator-specific statistics data answered
Public's account carry out fractional value prediction, the fractional value that exports is as public's account to be evaluated after finally obtaining forecast of regression model
Number mass fraction.Sample data can be extracted in the embodiment of the present invention from public platform to train to obtain regression model,
Fractional value prediction can be carried out to public's account to be evaluated, by regression model to obtain the matter of public's account to be evaluated
Measure score.Since regression model is to train to obtain by the sample data extracted from public platform, sample data derives from
Public platform, public's account in public platform can carry out quality evaluation by the regression model, so as to realize to public affairs
The quality of many accounts makes effectively evaluating.
For ease of being better understood from and implementing the said program of the embodiment of the present invention, corresponding application scenarios of illustrating below come
It is specifically described.
With the development of the commercialization process of public's account platform, there is high-quality article quality, height to enliven bean vermicelli etc. spy
The high quality public account of sign is in many business such as advertisement dispensing, trade marketing, reference business etc. have demand and make
It uses, and public's account of high quality can more achieve the effect that get half the result with twice the effort, therefore public's account excavation of high quality is compeled in eyebrow
It is anxious, the quality of public's account can be got through regression model to realize, and by public platform in the embodiment of the present invention
Some indicator-specific statistics data (such as weekly publication article number, article by reading ratio, enliven bean vermicelli etc.) carry out at reprocessing
Reason is aggregated into the characteristic of each latitude, and carries out quality score using machine learning model.Wherein, indicator-specific statistics data can
To include：Collected from public platform public's account and its bean vermicelli data and bean vermicelli to the reading article of public's account, click
Menu, transmission message, public's account send and receive message data, the article that public's account is delivered is read or forwarded
Data, corresponding bean vermicelli passes through public's account and carries out the statistical data such as the related operation indicator of data of payment behavior etc..
It please refers to shown in Fig. 2, is a kind of applied field of the quality evaluating method of public's account provided in an embodiment of the present invention
Scape schematic diagram.In the embodiment of the present invention can by the quality score problem of public's account by the regression model in machine learning come
It realizes, in conjunction with the data that public platform marks out, is trained first using regression algorithm such as logistic regression, support vector regression etc.
Go out regression model, final realize carries out Score on Prediction marking, fractional value Fan Weikeyiwei [ to each public's account;0,100],
In, fractional value the higher the better represent public's account quality it is more better more high-quality, wherein 0 point is worst, 100 points are best.It is whole
A regression model frame is as shown in Fig. 2, can be divided into:Early period, public's account data prepared analysis, training sample data and test
Several steps such as the mark of sample data and screening, Feature Engineering analysis, model prediction and evaluation and test optimization, prediction of result.Its
In, it needs that the feature of collection is carried out carrying out Feature Selection according to processes such as importance analysis, stability analyses in machine learning,
To obtain to the helpful feature of final result.
As shown in Fig. 2, in the quality of public's account provided in an embodiment of the present invention excavates frame, training sample data are first
Manual features project analysis first is carried out to regression model, to complete the training to regression model, test sample data are also to returning
Return model to carry out manual features project analysis, to complete the evaluation and test to regression model, next the regression model is carried out again
Iteration optimization, so that it is determined that going out the characteristic of regression model.The regression model can be used for public's account to be evaluated into
Row fractional value is predicted, prediction result is then exported.
In embodiments of the present invention, sample data can be selected from public platform, and the characteristic of regression model is logical
It crosses after being screened to indicator-specific statistics data and obtains.For example, bottom of the foundation characteristic data from public's account used in Fig. 2
Some statistical data mainly have the uplink behavioral data of bean vermicelli and public's account, public's account payment data, public's account to powder
The three categories data such as the downstream message data of silk.Wherein, public's account payment data refers to that bean vermicelli is paid to public's account
Data.Specifically, indicator-specific statistics data may include public's account related data and relevant attribute information, mainly there are public affairs
Many account bean vermicelli situations, public's account and bean vermicelli upstream message data, public's account payment data, public's account and bean vermicelli downlink
Message, public's Account interface data etc., wherein public's Account interface data refer to some opened inside public's account
Outside links data.To increase the confidence level and robustness of model, when model training, while the past period is considered
The ASSOCIATE STATISTICS numerical value of interior each category feature.
In embodiments of the present invention, data can also be carried out respectively according to subscription type public account and service type public account
Mark and model training, for example, can be according to 0~100 point of each fraction range according to phase for the public's account marked
Brush choosing is carried out in proportion, is ensured sample equalization problem in this way, is filtered out training sample data about 20,000, test sample number altogether
It is a plurality of according to about 10,000.
It should be noted that in embodiments of the present invention, there are many regression models that can be used, such as LR models,
Can be RFReg models, SGD Reg models, SVR models etc. do not limit, the regression model used in the embodiment of the present invention
It can also be other available models, such as decision tree and its variant, neural network, deep learning etc. model.The present invention is real
The replacement for mentioning characteristic in example is applied, the other data targets that can relate to of text, article etc. public's account can also be used instead
Etc..Machine learning model can be used in the embodiment of the present invention, can also be substituted for some artificial experiences and some correlations are referred to
Mark, for example calculate separately ASSOCIATE STATISTICS from operation indicator, bean vermicelli index, article index and interactive this four great s of index and refer to
Then mark is weighted the final fractional value of combination calculating to indicate the scheme of the quality point of this public's account.In addition, this hair
In bright embodiment labeled data in addition to it is shown in Fig. 2 it is artificial be labeled other than, Active Learning (full name in English can also be utilized：
Active learning) method gradually marked.For example, manually marking a small amount of sample every time, then regression model utilizes people
The sample of work mark, which expand, learns more samples, manually confirms that whole process is continuous to the sample of model mark again
Continue, until mark sample reaches sufficient amount.
By the aforementioned illustration to the present invention it is found that using regression algorithm to public's account in the embodiment of the present invention
Index carries out importance screening, and combines the situation of change of these index values in the past period, is aggregated into public's account
Qualitative character list, be then trained and predict using regression model, thus have more robustness and stability.The present invention
Public's account of the high quality of embodiment output has a wide range of applications scene, such as advertisement dispensing, trade marketing activity, sign
Communication service etc., it is generally the case that 20% high quality public account can cover 80% user group, therefore can be at this
It tends to achieve the effect that get half the result with twice the effort in a little application scenarios.
It should be noted that for each method embodiment above-mentioned, for simple description, therefore it is all expressed as a series of
Combination of actions, but those skilled in the art should understand that, the present invention is not limited by the described action sequence because
According to the present invention, certain steps can be performed in other orders or simultaneously.Secondly, those skilled in the art should also know
It knows, embodiment described in this description belongs to preferred embodiment, and involved action and module are not necessarily of the invention
It is necessary.
For ease of preferably implementing the said program of the embodiment of the present invention, the phase for implementing said program is also provided below
Close device.
It please refers to shown in Fig. 3-a, a kind of quality evaluation device 300 of public's account provided in an embodiment of the present invention, it can be with
Including：Model acquisition module 301, model prediction module 302 and quality assessment module 303, wherein
Model acquisition module 301, for obtains pass through regression algorithm from the acquistion of sample data middle school to regression model, institute
Stating sample data includes：Multiple public's accounts in public platform and the corresponding indicator-specific statistics number of the multiple public's account
According to；
Model prediction module 302 passes through described time for public's account to be evaluated to be input in the regression model
Model is returned to carry out fractional value prediction to public's account to be evaluated；
Quality assessment module 303, for obtaining the fractional value exported after the forecast of regression model as described to be evaluated
Public's account mass fraction.
In some embodiments of the invention, it please refers to shown in Fig. 3-b, the quality evaluation device 300 of public's account,
Further include：Data acquisition module 304 and model training module 305, wherein
The data acquisition module 304 obtains for the model acquisition module 301 and passes through regression algorithm from sample data
Before the regression model that middle school's acquistion is arrived, multiple public's accounts and the multiple public's account pair are got from public platform
The indicator-specific statistics data answered；
The model training module 305, for corresponding with the multiple public's account and the multiple public's account
Indicator-specific statistics data are trained study as sample data, by regression algorithm to regression model, and what output study obtained returns
Return model.
In some embodiments of the invention, it please refers to shown in Fig. 3-c, the model training module 305, including：
Data categorization module 3051 obtains training sample data and test sample number for sample data to be divided into two classes
According to the training sample data include：Training public's account and the corresponding indicator-specific statistics data of trained public's account, institute
Stating test sample data includes：Test public's account and the corresponding indicator-specific statistics data of test public's account, wherein from
The multiple public's account that public platform is got is divided into two classes：Trained public's account and test public's account；
First model processing modules 3052, for using the training sample data, by regression algorithm to the recurrence
Model carries out Feature Engineering and analyzes to obtain the fisrt feature data of regression model, according to the fisrt feature data of the regression model
The regression model trained and completed is exported after carrying out prediction optimization to the regression model；
Second model processing modules 3053, for using the test sample data, by the regression algorithm to described
The regression model that training is completed carries out Feature Engineering and analyzes to obtain the second feature data of regression model, according to the regression model
Second feature data the regression model that output study obtains after evaluation and test optimization is carried out to the regression model.
In some embodiments of the invention, it please refers to shown in Fig. 3-d, shown in Fig. 3-c, the model training mould
Block 300 further includes：Data labeling module 3054 and data screening module 3055, wherein
The data labeling module 3054 uses the number of training for first model processing modules 3052
According to, before analyzing to obtain the fisrt feature data of regression model to regression model progress Feature Engineering by regression algorithm,
Data mark is carried out respectively to the training sample data and the test sample data, the training sample data after being marked
With the test sample data after mark；
The data screening module 3055, for the training sample data after the mark and the test after the mark
Sample data carries out data screening respectively.
In some embodiments of the invention, it please refers to shown in Fig. 3-e, first model processing modules 3052, including：
Importance analysis module 30521, for the corresponding finger of training public's account described in the training sample data
Mark statistical data carries out screening analysis according to feature importance, and qualitative character list is written in the characteristic screened；
Stability analysis module 30522, for judging the characteristic in the qualitative character list in historical time section
Inside whether change, is returned using the characteristic exported from high to low according to stability in the qualitative character list as described
Return the fisrt feature data of model.
In some embodiments of the invention, the model training module 305 is specifically used for the multiple public's account
And the corresponding indicator-specific statistics data of the multiple public's account are as sample data, by regression algorithm respectively to multiple recurrence
Model is trained study, multiple regression models that output study obtains；
The model acquisition module 301, specifically for carrying out evaluation respectively to the multiple regression model, from
The best regression model of Evaluated effect is selected in the multiple regression model as the regression model got.
In some embodiments of the invention, the indicator-specific statistics data, including：Operation indicator statistical data, bean vermicelli refer to
Mark statistical data, article indicator-specific statistics data and interactive indicator-specific statistics data.
In some embodiments of the invention, the indicator-specific statistics data, including：Uplink behavior of the bean vermicelli to public's account
Indicator-specific statistics data, public's account payment indicator-specific statistics data, public's account are to the downstream message indicator-specific statistics data of bean vermicelli.
Regression algorithm is passed through from sample data it is found that obtaining first to the description of the embodiment of the present invention by above example
The regression model that middle school's acquistion is arrived, sample data include：Multiple public's accounts in public platform and multiple public's accounts pair
Then public's account to be evaluated is input in regression model, by regression model to be evaluated by the indicator-specific statistics data answered
Public's account carry out fractional value prediction, the fractional value that exports is as public's account to be evaluated after finally obtaining forecast of regression model
Number mass fraction.Sample data can be extracted in the embodiment of the present invention from public platform to train to obtain regression model,
Fractional value prediction can be carried out to public's account to be evaluated, by regression model to obtain the matter of public's account to be evaluated
Measure score.Since regression model is to train to obtain by the sample data extracted from public platform, sample data derives from
Public platform, public's account in public platform can carry out quality evaluation by the regression model, so as to realize to public affairs
The quality of many accounts makes effectively evaluating.
Fig. 4 is a kind of server architecture schematic diagram provided in an embodiment of the present invention, which can be because of configuration or property
Energy is different and generates bigger difference, may include one or more central processing units (central processing
Units, CPU) 1122 (for example, one or more processors) and memory 1132, one or more storage applications
The storage medium 1130 (such as one or more mass memory units) of program 1142 or data 1144.Wherein, memory
1132 and storage medium 1130 can be of short duration storage or persistent storage.The program for being stored in storage medium 1130 may include one
A or more than one module (diagram does not mark), each module may include to the series of instructions operation in server.More into
One step, central processing unit 1122 could be provided as communicating with storage medium 1130, and storage medium is executed on server 1100
Series of instructions operation in 1130.
Server 1100 can also include one or more power supplys 1126, one or more wired or wireless nets
Network interface 1150, one or more input/output interfaces 1158, and/or, one or more operating systems 1141, example
Such as Windows ServerTM, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM etc..
Fig. 4 institutes can be based on by the quality evaluating method step of public's account performed by server in above-described embodiment
The server architecture shown.
In addition it should be noted that, the apparatus embodiments described above are merely exemplary, wherein described as separation
The unit of part description may or may not be physically separated, the component shown as unit can be or
It can not be physical unit, you can be located at a place, or may be distributed over multiple network units.It can be according to reality
Border needs to select some or all of module therein to achieve the purpose of the solution of this embodiment.In addition, provided by the invention
In device embodiment attached drawing, the connection relation between module indicates there is communication connection between them, specifically can be implemented as one
Item or a plurality of communication bus or signal wire.Those of ordinary skill in the art are without creative efforts, you can with
Understand and implements.
Through the above description of the embodiments, it is apparent to those skilled in the art that the present invention can borrow
Help software that the mode of required common hardware is added to realize, naturally it is also possible to by specialized hardware include application-specific integrated circuit, specially
It is realized with CPU, private memory, special components and parts etc..Under normal circumstances, all functions of being completed by computer program can
It is easily realized with corresponding hardware, moreover, for realizing that the particular hardware structure of same function can also be a variety of more
Sample, such as analog circuit, digital circuit or special circuit etc..But it is more for the purpose of the present invention in the case of software program it is real
It is now more preferably embodiment.Based on this understanding, technical scheme of the present invention substantially in other words makes the prior art
The part of contribution can be expressed in the form of software products, which is stored in the storage medium that can be read
In, such as the floppy disk of computer, USB flash disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory
Device (RAM, Random Access Memory), magnetic disc or CD etc., including some instructions are with so that a computer is set
Standby (can be personal computer, server or the network equipment etc.) executes the method described in each embodiment of the present invention.
In conclusion the above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations；Although with reference to upper
Stating embodiment, invention is explained in detail, it will be understood by those of ordinary skill in the art that：It still can be to upper
The technical solution recorded in each embodiment is stated to modify or equivalent replacement of some of the technical features；And these
Modification or replacement, the spirit and scope for various embodiments of the present invention technical solution that it does not separate the essence of the corresponding technical solution.
Claims (6)
1. a kind of quality evaluating method of public's account, which is characterized in that including：
Multiple public's accounts and the corresponding indicator-specific statistics data of the multiple public's account are got from public platform；
Sample data is divided into two classes, obtains training sample data and test sample data, the training sample data include：Instruction
Practice public's account and the corresponding indicator-specific statistics data of trained public's account, the test sample data include：Test is public
Many accounts and the corresponding indicator-specific statistics data of test public's account, wherein what is got from public platform is the multiple
Public's account is divided into two classes：Trained public's account and test public's account；Wherein, the indicator-specific statistics data, packet
It includes：Operation indicator statistical data, bean vermicelli indicator-specific statistics data, article indicator-specific statistics data and interactive indicator-specific statistics data；Or, institute
Indicator-specific statistics data are stated, including：Bean vermicelli pays indicator-specific statistics to uplink behavioral indicator statistical data, the public's account of public's account
Data, public's account are to the downstream message indicator-specific statistics data of bean vermicelli；
The corresponding indicator-specific statistics data of training public's account described in the training sample data are carried out according to feature importance
The characteristic screened is written qualitative character list, judges the characteristic in the qualitative character list by screening analysis
Whether change according in historical time section, the feature that will from high to low be exported according to stability in the qualitative character list
Fisrt feature data of the data as the regression model；According to the fisrt feature data of the regression model to the recurrence mould
Type exports the regression model that training is completed after carrying out prediction optimization；
Using the test sample data, the regression model completed to the training by the regression algorithm carries out Feature Engineering
Analysis obtain the second feature data of regression model, according to the second feature data of the regression model to the regression model into
The regression model that output study obtains after row evaluation and test optimization；
Obtain by regression algorithm from the acquistion of sample data middle school to regression model, the sample data includes：Public platform
In multiple public's accounts and the corresponding indicator-specific statistics data of the multiple public's account；
Public's account to be evaluated is input in the regression model, by the regression model to the public to be evaluated
Account carries out fractional value prediction；
Obtain mass fraction of the fractional value exported after the forecast of regression model as public's account to be evaluated.
2. according to the method described in claim 1, it is characterized in that, described use the training sample data, by returning calculation
Before method is analyzed to obtain the fisrt feature data of regression model to regression model progress Feature Engineering, the method is also wrapped
It includes：
Data mark is carried out respectively to the training sample data and the test sample data, the training sample after being marked
Test sample data after data and mark；
Data screening is carried out respectively to the training sample data after the mark and the test sample data after the mark.
3. according to the method described in claim 1, it is characterized in that, described with the multiple public's account and the multiple public affairs
The corresponding indicator-specific statistics data of many accounts are trained study by regression algorithm as sample data to regression model, export
Learn obtained regression model, including：
Using the multiple public's account and the corresponding indicator-specific statistics data of the multiple public's account as sample data, pass through
Regression algorithm is trained study, multiple regression models that output study obtains to multiple regression models respectively；
It is described obtain by regression algorithm from the acquistion of sample data middle school to regression model, including：
Evaluation is carried out respectively to the multiple regression model, Evaluated effect is selected from the multiple regression model
Best regression model is as the regression model got.
4. a kind of quality evaluation device of public's account, which is characterized in that including：
Data acquisition module, for model acquisition module obtain by regression algorithm from the acquistion of sample data middle school to recurrence mould
Before type, multiple public's accounts and the corresponding indicator-specific statistics data of the multiple public's account are got from public platform；
Data categorization module obtains training sample data and test sample data, the instruction for sample data to be divided into two classes
Practicing sample data includes：Training public's account and the corresponding indicator-specific statistics data of trained public's account, the test specimens
Notebook data includes：Test public's account and the corresponding indicator-specific statistics data of test public's account, wherein from public platform
The multiple public's account got is divided into two classes：Trained public's account and test public's account；Wherein, described
Indicator-specific statistics data, including：Operation indicator statistical data, bean vermicelli indicator-specific statistics data, article indicator-specific statistics data and interaction refer to
Mark statistical data；Or, the indicator-specific statistics data, including：Bean vermicelli is to the uplink behavioral indicator statistical data of public's account, the public
Account pays indicator-specific statistics data, public's account to the downstream message indicator-specific statistics data of bean vermicelli；
First model processing modules carry out the regression model by regression algorithm for using the training sample data
Feature Engineering is analyzed to obtain the fisrt feature data of regression model, according to the fisrt feature data of the regression model to described time
The regression model that training is completed is exported after returning model to carry out prediction optimization；First model processing modules, including：Importance point
Module is analysed, for important according to feature to the corresponding indicator-specific statistics data of training public's account described in the training sample data
Property carry out screening analysis, by the characteristic screened be written qualitative character list；Stability analysis module, for judging
Whether the characteristic stated in qualitative character list changes in historical time section, by the qualitative character list according to
Fisrt feature data of the characteristic that stability exports from high to low as the regression model；
Second model processing modules complete the training by the regression algorithm for using the test sample data
Regression model carry out Feature Engineering and analyze to obtain the second feature data of regression model, it is special according to the second of the regression model
Output learns obtained regression model after sign data carry out evaluation and test optimization to the regression model；Model acquisition module, for obtaining
Take by regression algorithm from the acquistion of sample data middle school to regression model, the sample data includes：It is more in public platform
A public's account and the corresponding indicator-specific statistics data of the multiple public's account；
Model prediction module passes through the regression model for public's account to be evaluated to be input in the regression model
Fractional value prediction is carried out to public's account to be evaluated；
Quality assessment module, for obtaining the fractional value exported after the forecast of regression model as public's account to be evaluated
Number mass fraction.
5. device according to claim 4, which is characterized in that the model training module further includes：Data labeling module
With data screening module, wherein
The data labeling module uses the training sample data for first model processing modules, is calculated by returning
Before method is analyzed to obtain the fisrt feature data of regression model to regression model progress Feature Engineering, to the training sample
Data and the test sample data carry out data mark respectively, the training sample data after being marked and the test after mark
Sample data；
The data screening module, for the training sample data after the mark and the test sample data after the mark
Data screening is carried out respectively.
6. device according to claim 4, which is characterized in that the model training module is specifically used for the multiple
Public's account and the corresponding indicator-specific statistics data of the multiple public's account are right respectively by regression algorithm as sample data
Multiple regression models are trained study, multiple regression models that output study obtains；
The model acquisition module, specifically for carrying out evaluation respectively to the multiple regression model, from described more
The best regression model of Evaluated effect is selected in a regression model as the regression model got.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610420186.5A CN105824806B (en) | 2016-06-13 | 2016-06-13 | A kind of quality evaluating method and device of public's account |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610420186.5A CN105824806B (en) | 2016-06-13 | 2016-06-13 | A kind of quality evaluating method and device of public's account |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105824806A CN105824806A (en) | 2016-08-03 |
| CN105824806B true CN105824806B (en) | 2018-10-23 |
Family
ID=56532199
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610420186.5A Active CN105824806B (en) | 2016-06-13 | 2016-06-13 | A kind of quality evaluating method and device of public's account |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105824806B (en) |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106657016A (en) * | 2016-11-10 | 2017-05-10 | 北京奇艺世纪科技有限公司 | Illegal user name recognition method and system |
| CN108416684B (en) * | 2017-02-10 | 2021-09-07 | 腾讯科技（深圳）有限公司 | Method and device for evaluating credibility of account main body and server |
| CN107845408B (en) * | 2017-10-25 | 2020-10-27 | 医渡云（北京）技术有限公司 | Data evaluation method and device, storage medium and electronic device |
| CN108829750A (en) * | 2018-05-24 | 2018-11-16 | 国信优易数据有限公司 | A kind of quality of data determines system and method |
| CN108920617B (en) * | 2018-06-28 | 2022-07-12 | 中译语通科技股份有限公司 | Data acquisition judging system and method and information data processing terminal |
| CN110570025A (en) * | 2019-08-20 | 2019-12-13 | 精硕科技（北京）股份有限公司 | prediction method, device and equipment for real reading rate of WeChat seal |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101571934A (en) * | 2009-05-26 | 2009-11-04 | 北京航空航天大学 | Enterprise independent innovation ability prediction method based on support vector machine |
| CN102945279A (en) * | 2012-11-14 | 2013-02-27 | 清华大学 | Evaluating method and device of influence effect of microblog users |
| CN103761266A (en) * | 2014-01-02 | 2014-04-30 | 北京集奥聚合网络技术有限公司 | Click rate predicting method and system based on multistage logistic regression |
| CN104915397A (en) * | 2015-05-28 | 2015-09-16 | 国家计算机网络与信息安全管理中心 | Method and device for predicting microblog propagation tendencies |
| CN105224608A (en) * | 2015-09-06 | 2016-01-06 | 华南理工大学 | The hot news Forecasting Methodology analyzed based on microblog data and system |
| CN105512245A (en) * | 2015-11-30 | 2016-04-20 | 青岛智能产业技术研究院 | Enterprise figure building method based on regression model |
-
2016
- 2016-06-13 CN CN201610420186.5A patent/CN105824806B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101571934A (en) * | 2009-05-26 | 2009-11-04 | 北京航空航天大学 | Enterprise independent innovation ability prediction method based on support vector machine |
| CN102945279A (en) * | 2012-11-14 | 2013-02-27 | 清华大学 | Evaluating method and device of influence effect of microblog users |
| CN103761266A (en) * | 2014-01-02 | 2014-04-30 | 北京集奥聚合网络技术有限公司 | Click rate predicting method and system based on multistage logistic regression |
| CN104915397A (en) * | 2015-05-28 | 2015-09-16 | 国家计算机网络与信息安全管理中心 | Method and device for predicting microblog propagation tendencies |
| CN105224608A (en) * | 2015-09-06 | 2016-01-06 | 华南理工大学 | The hot news Forecasting Methodology analyzed based on microblog data and system |
| CN105512245A (en) * | 2015-11-30 | 2016-04-20 | 青岛智能产业技术研究院 | Enterprise figure building method based on regression model |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105824806A (en) | 2016-08-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105824806B (en) | A kind of quality evaluating method and device of public's account | |
| Nabizadeh et al. | Adaptive learning path recommender approach using auxiliary learning objects | |
| Li et al. | Mobile social media use intention in emergencies among Gen Y in China: An integrative framework of gratifications, task-technology fit, and media dependency | |
| CN110008399B (en) | Recommendation model training method and device, and recommendation method and device | |
| US11521221B2 (en) | Predictive modeling with entity representations computed from neural network models simultaneously trained on multiple tasks | |
| Garber et al. | From density to destiny: Using spatial dimension of sales data for early prediction of new product success | |
| CN109902708A (en) | A kind of recommended models training method and relevant apparatus | |
| Dong et al. | Multilevel modelling with spatial interaction effects with application to an emerging land market in Beijing, China | |
| CN109583904A (en) | Training method, impaired operation detection method and the device of abnormal operation detection model | |
| CN108804704A (en) | A kind of user's depth portrait method and device | |
| US11086754B2 (en) | Automated feedback-based application optimization | |
| CN110111139A (en) | Behavior prediction model generation method, device, electronic equipment and readable medium | |
| CN112070310A (en) | Loss user prediction method and device based on artificial intelligence and electronic equipment | |
| CN110503447A (en) | For determining the method and device of Sales Volume of Commodity predicted value | |
| Giupponi | Decision support for mainstreaming climate change adaptation in water resources management | |
| Al-Sharafi et al. | Cloud computing adoption in higher education: An integrated theoretical model | |
| Farmer et al. | Habitat models for land-use planning: assumptions and strategies for development | |
| CN110263250A (en) | A kind of generation method and device of recommended models | |
| CN107291722B (en) | Descriptor classification method and device | |
| Colbaugh et al. | Emerging topic detection for business intelligence via predictive analysis of'meme'dynamics | |
| CN113128739B (en) | Prediction method of user touch time, prediction model training method and related devices | |
| CN108132883A (en) | A kind of software test data analysis method, device, system and storage medium | |
| Hassouni et al. | Using generative adversarial networks to develop a realistic human behavior simulator | |
| JP2018097726A (en) | Determining apparatus, determining method, determining program, generating apparatus, generating method, and generating program | |
| KR101918359B1 (en) | A Data Processing System Using AI |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |