CN108388563A - Information output method and device - Google Patents
Information output method and device Download PDFInfo
- Publication number
- CN108388563A CN108388563A CN201710063291.2A CN201710063291A CN108388563A CN 108388563 A CN108388563 A CN 108388563A CN 201710063291 A CN201710063291 A CN 201710063291A CN 108388563 A CN108388563 A CN 108388563A
- Authority
- CN
- China
- Prior art keywords
- information
- sample data
- disaggregated model
- mentioned
- classification
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/332—Query formulation
- G06F16/3329—Natural language query formulation or dialogue systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
- G06F16/353—Clustering; Classification into predefined classes
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Human Computer Interaction (AREA)
- Computational Linguistics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
This application discloses information output methods and device.One specific implementation mode of this method includes：Receive the information that client is sent；Feature vector is extracted from the information, wherein this feature vector is used to characterize the content of the information；This feature vector is imported information disaggregated model trained in advance to be classified to obtain the classification belonging to the information；Object feedback information associated with the information is matched from the feedback information set that prestores indicated by the category, and the object feedback information is exported to the client.This embodiment improves information delivery efficiencies.
Description
Technical field
This application involves field of computer technology, and in particular to Internet technical field more particularly to information output method
And device.
Background technology
In artificial intelligence field, much automaticmanual intelligent response systems based on natural language processing have been emerged at present
System.The information delivery efficiency of artificial intelligence response system is largely dependent upon the accuracy of disaggregated model.
However, the accuracy of existing disaggregated model is usually relatively low, the information so as to cause artificial intelligence response system is defeated
Go out less efficient.
Invention content
The purpose of the application is to propose a kind of improved information output method and device, to solve background above technology department
Divide the technical issues of mentioning.
In a first aspect, this application provides a kind of information output method, this method includes：Receive the letter that client is sent
Breath；Feature vector is extracted from above- mentioned information, wherein features described above vector is used to characterize the content of above- mentioned information；It will be above-mentioned
Feature vector imports information disaggregated model trained in advance and is classified to obtain the classification belonging to above- mentioned information；From above-mentioned classification institute
Object feedback information associated with above- mentioned information is matched in the feedback information set that prestores indicated, and by above-mentioned object feedback
Information is exported to above-mentioned client.
In some embodiments, the above method further includes the steps that establishing information disaggregated model, above-mentioned to establish information classification
The step of model includes:It obtains sample data sets and distinguishes to each sample data in above-mentioned sample data sets related
The classification of connection；For each candidate information disaggregated model in pre-stored candidate information disaggregated model, by above-mentioned candidate
Information disaggregated model as candidate information disaggregated model to be assessed, based on above-mentioned sample data sets and with above-mentioned sample data set
Each sample data in conjunction distinguishes associated classification, determines the accuracy rate of above-mentioned candidate information disaggregated model to be assessed,
Wherein, above-mentioned pre-stored candidate information disaggregated model is unbred model；Based on identified accuracy rate, above-mentioned
Target information disaggregated model is determined in pre-stored candidate information disaggregated model；Using machine learning method, it is based on above-mentioned sample
Notebook data set and classification associated with each sample data difference in above-mentioned sample data sets believe above-mentioned target
Breath disaggregated model is trained to obtain information disaggregated model.
In some embodiments, the accuracy rate of the above-mentioned candidate information disaggregated model to be assessed of above-mentioned determination, including：Using friendship
Verification method is pitched to determine the accuracy rate of above-mentioned candidate information disaggregated model to be assessed.
In some embodiments, the accuracy rate of the above-mentioned candidate information disaggregated model to be assessed of above-mentioned determination, including：For upper
Each classification sample data in sample data sets is stated, determines weighted value associated with above-mentioned classification sample data, wherein
The total number for the sample data that above-mentioned weighted value includes with above-mentioned sample data sets by the number of above-mentioned classification sample data
Ratio；According to identified weighted value, the accuracy rate of above-mentioned candidate information disaggregated model to be assessed is determined.
In some embodiments, weighted value determined by above-mentioned basis determines above-mentioned candidate information disaggregated model to be assessed
Accuracy rate, including：Cycle executes following processing step predetermined number contents：By above-mentioned sample data sets be divided into training set and
Test set is divided using machine learning method based on above-mentioned training set and with each training sample data in above-mentioned training set
Not associated classification is trained the classification mould of the candidate information after being trained to above-mentioned candidate information disaggregated model to be assessed
Type, using the candidate information disaggregated model after above-mentioned training to the classifications of each test sample data in above-mentioned test set into
Row prediction obtains prediction result, true based on above-mentioned prediction result for each classification test sample data in above-mentioned test set
Fixed predictablity rate associated with above-mentioned classification test sample data, will be included with above-mentioned sample data sets with it is above-mentioned
There is classification test sample data the associated weighted value of sample data of the same category to make with the product of above-mentioned predictablity rate
For weight estimation accuracy rate associated with above-mentioned classification test sample data, will be tested with each classification in above-mentioned test set
The numerical value that sample data distinguishes associated weight estimation accuracy rate addition gained is classified as with above-mentioned candidate information to be assessed
The associated first weight estimation accuracy rate of model restores above-mentioned candidate information disaggregated model to be assessed to unbred shape
State, wherein above-mentioned training set and above-mentioned test set include the sample data of the same category, with above-mentioned classification test sample data phase
Associated predictablity rate is that the candidate information disaggregated model after above-mentioned training is pre- to the classification of above-mentioned classification test sample data
Survey the ratio of correct number and the number of above-mentioned classification test sample data；By the first weighting of the above-mentioned predetermined number of gained
Accuracy rate of the average value of predictablity rate as above-mentioned candidate information disaggregated model to be assessed.
In some embodiments, above-mentioned based on identified accuracy rate, in above-mentioned pre-stored candidate information classification mould
Target information disaggregated model is determined in type, including：By the accuracy rate highest in above-mentioned pre-stored candidate information disaggregated model
Candidate information disaggregated model as target information disaggregated model.
Second aspect, this application provides a kind of information output apparatus, which includes：Receiving unit is configured to connect
Receive the information that client is sent；Extraction unit is configured to extract feature vector from above- mentioned information, wherein features described above
Content of the vector for characterizing above- mentioned information；Taxon is configured to importing features described above vector into information trained in advance
Disaggregated model is classified to obtain the classification belonging to above- mentioned information；Output unit is configured to pre- indicated by the above-mentioned classification
It deposits and matches object feedback information associated with above- mentioned information in feedback information set, and above-mentioned object feedback information is exported
To above-mentioned client.
In some embodiments, above-mentioned apparatus further includes：Information disaggregated model establishes unit, is configured to establish information point
Class model, including：Obtain subelement, be configured to obtain sample data sets and with each in above-mentioned sample data sets
Sample data distinguishes associated classification；Accuracy rate determination subelement is configured to classify for pre-stored candidate information
Each candidate information disaggregated model in model, using above-mentioned candidate information disaggregated model as candidate information to be assessed classification mould
Type, based on above-mentioned sample data sets and class associated with each sample data difference in above-mentioned sample data sets
Not, the accuracy rate of above-mentioned candidate information disaggregated model to be assessed is determined, wherein above-mentioned pre-stored candidate information disaggregated model
It is unbred model；Target information disaggregated model determination subelement, is configured to based on identified accuracy rate, above-mentioned
Target information disaggregated model is determined in pre-stored candidate information disaggregated model；Training subelement is configured to utilize machine
Learning method, it is based on above-mentioned sample data sets and associated respectively with each sample data in above-mentioned sample data sets
Classification, above-mentioned target information disaggregated model is trained to obtain information disaggregated model.
In some embodiments, above-mentioned accuracy rate determination subelement includes：First accuracy rate determining module, is configured to adopt
The accuracy rate of above-mentioned candidate information disaggregated model to be assessed is determined with cross validation method.
In some embodiments, above-mentioned accuracy rate determination subelement includes：Weighted value determining module, is configured to for upper
Each classification sample data in sample data sets is stated, determines weighted value associated with above-mentioned classification sample data, wherein
The total number for the sample data that above-mentioned weighted value includes with above-mentioned sample data sets by the number of above-mentioned classification sample data
Ratio；Second accuracy rate determining module is configured to, according to identified weighted value, determine above-mentioned candidate information to be assessed point
The accuracy rate of class model.
In some embodiments, above-mentioned second accuracy rate determining module includes：Submodule is handled, cycle is configured to and executes
Following processing step predetermined number contents：Above-mentioned sample data sets are divided into training set and test set, utilize machine learning side
Method, based on above-mentioned training set and with each training sample data in above-mentioned training set, associated classification is waited for above-mentioned respectively
Assessment candidate information disaggregated model is trained the candidate information disaggregated model after being trained, and utilizes the candidate after above-mentioned training
Information disaggregated model is predicted to obtain prediction result to the classification of each test sample data in above-mentioned test set, for
Each classification test sample data in above-mentioned test set are determined and above-mentioned classification test sample data based on above-mentioned prediction result
Associated predictablity rate will be included with above-mentioned classification test sample data with above-mentioned sample data sets with identical
The associated weighted value of sample data of classification and the product of above-mentioned predictablity rate as with above-mentioned classification test sample data
Associated weight estimation accuracy rate, by weighting associated with each classification test sample data difference in above-mentioned test set
Predictablity rate is added the numerical value of gained as first weight estimation associated with above-mentioned candidate information disaggregated model to be assessed
Accuracy rate restores above-mentioned candidate information disaggregated model to be assessed to unbred state, wherein above-mentioned training set and above-mentioned
Test set includes the sample data of the same category, and predictablity rate associated with above-mentioned classification test sample data is above-mentioned instruction
Candidate information disaggregated model after white silk surveys the correct number of class prediction of above-mentioned classification test sample data with above-mentioned classification
The ratio of the number of sample notebook data；Accuracy rate determination sub-module is configured to add the above-mentioned predetermined number of gained first
Weigh accuracy rate of the average value of predictablity rate as above-mentioned candidate information disaggregated model to be assessed.
In some embodiments, above-mentioned target information disaggregated model determination subelement includes：Target information disaggregated model is true
Cover half block is configured to the highest candidate information classification mould of accuracy rate in above-mentioned pre-stored candidate information disaggregated model
Type is as target information disaggregated model.
The third aspect, this application provides a kind of server, which includes：One or more processors；Storage dress
It sets, for storing one or more programs；When said one or multiple programs are executed by said one or multiple processors so that
Said one or multiple processors realize the method as described in any realization method in first aspect.
Fourth aspect, this application provides a kind of computer readable storage mediums, are stored thereon with computer program, special
Sign is, the method as described in any realization method in first aspect is realized when above procedure is executed by processor.
Information output method and device provided by the present application, by extracting feature vector from the information received, with
Just this feature vector information disaggregated model trained in advance is imported to be classified to obtain the classification belonging to the information.Then from this
Object feedback information associated with the information is matched in the feedback information set that prestores indicated by classification, so as to by the target
Feedback information is exported to client.To which the information disaggregated model for being effectively utilized trained in advance classifies to the information, carry
The high efficiency of information output.
Description of the drawings
By reading a detailed description of non-restrictive embodiments in the light of the attached drawings below, the application's is other
Feature, objects and advantages will become more apparent upon：
Fig. 1 is that this application can be applied to exemplary system architecture figures therein；
Fig. 2 is the flow chart according to one embodiment of the information output method of the application；
Fig. 3 is the schematic diagram of an application scenarios corresponding with embodiment shown in Fig. 2；
Fig. 4 is the accuracy rate that weighted value determines candidate information disaggregated model to be assessed determined by basis according to the application
One embodiment flow chart.
Fig. 5 is the structural schematic diagram according to one embodiment of the information output apparatus of the application；
Fig. 6 is adapted for the structural schematic diagram of the computer system of the server for realizing the embodiment of the present application.
Specific implementation mode
The application is described in further detail with reference to the accompanying drawings and examples.It is understood that this place is retouched
The specific embodiment stated is used only for explaining related invention, rather than the restriction to the invention.It also should be noted that in order to
Convenient for description, is illustrated only in attached drawing and invent relevant part with related.
It should be noted that in the absence of conflict, the features in the embodiments and the embodiments of the present application can phase
Mutually combination.The application is described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.
Fig. 1 shows the exemplary system of the embodiment of the information output method or information output apparatus that can apply the application
System framework 100.
As shown in Figure 1, system architecture 100 may include terminal device 101,102,103, network 104 and server 105.
Network 104 between terminal device 101,102,103 and server 105 provide communication link medium.Network 104 can be with
Including various connection types, such as wired, wireless communication link or fiber optic cables etc..
User can be interacted by network 104 with server 105 with using terminal equipment 101,102,103, to receive or send out
Send message etc..Various telecommunication customer end applications can be installed on terminal device 101,102,103, such as support automaticmanual intelligence
The application of energy response, web browser applications, the application of shopping class, searching class application, instant messaging tools, mailbox client, society
Hand over platform software etc..
Terminal device 101,102,103 can be the various electronic equipments for having display screen, including but not limited to intelligent hand
Machine, tablet computer, E-book reader, MP3 player (Moving Picture Experts Group Audio Layer
III, dynamic image expert's compression standard audio level 3), MP4 (Moving Picture Experts Group Audio
Layer IV, dynamic image expert's compression standard audio level 4) player, pocket computer on knee and desktop computer etc.
Deng.
Server 105 can be to provide the server of various services, such as the letter to the transmission of terminal device 101,102,103
Breath analyze etc. the background server of processing, which can also be (such as related to above- mentioned information by handling result
The object feedback information of connection) feed back to terminal device.
It should be noted that the information output method that the embodiment of the present application is provided generally is executed by server 105, accordingly
Ground, information output apparatus are generally positioned in server 105.
It should be understood that the number of the terminal device, network and server in Fig. 1 is only schematical.According to realization need
It wants, can have any number of terminal device, network and server.
With continued reference to Fig. 2, the flow 200 of one embodiment of the information output method according to the application is shown.The letter
Output method is ceased, is included the following steps：
Step 201, the information that client is sent is received.
In the present embodiment, the electronic equipment (such as server 105 shown in FIG. 1) of information output method operation thereon
Can be received by wired connection mode or radio connection client (such as terminal device shown in FIG. 1 101,102,
103) information sent.Wherein, which can be the consulting that user is sent by above-mentioned client with text or voice mode
Information, for example, for seek advice from can deliver goods to the customers, place an order after whether can change order, which the means of payment has, big household electrical appliances peace
The information how dress expense is calculated etc..
Step 202, feature vector is extracted from above- mentioned information.
In the present embodiment, above-mentioned electronic equipment can extract feature vector from above- mentioned information.Wherein, this feature to
Amount can be used for characterizing the content of above- mentioned information.Here, if above- mentioned information is voice messaging, above-mentioned electronic equipment can utilize
Above- mentioned information is converted to text message by speech recognition technology.Above-mentioned electronic equipment can by above- mentioned information or to above- mentioned information into
The Feature Selection Model for extracting Feature Words that the text message of row conversion gained imports training in advance carries out Feature Words extraction,
And the Feature Words based on gained generate feature vector.As an example, for text message " how big household electrical appliances mounting cost is calculated ", lead to
The Feature Words for crossing the text information of features described above extraction model extraction can be " big household electrical appliances " and " mounting cost ", then can be with
Obtain the two-dimensional feature vector (" big household electrical appliances ", " mounting cost ") of text information.
In some optional realization methods of the present embodiment, above-mentioned electronic equipment can also be to extracting from above- mentioned information
Feature Words quantified, obtain the numerical value for characteristic feature word, and be based on the numerical generation feature vector.Here, above-mentioned
Numerical value can be characterized the weights of word, and weights can be characterized word in the number occurred in the information.
Step 203, feature vector information disaggregated model trained in advance is imported to be classified to obtain belonging to above- mentioned information
Classification.
In the present embodiment, the feature vector extracted from above- mentioned information can be imported above-mentioned letter by above-mentioned electronic equipment
Cease disaggregated model, above- mentioned information disaggregated model can by features described above vector according to advance trained correspondence, find with it is upper
The corresponding classification of feature vector is stated, and using the category as the classification belonging to above- mentioned information.
In some optional realization methods of the present embodiment, above- mentioned information output method can also include establishing information point
The step of class model may include：First, above-mentioned electronic equipment can obtain sample data sets and with the sample data sets
In each sample data distinguish associated classification, wherein with each sample data in above-mentioned sample data sets
Associated classification can be marked artificially in advance respectively.Then, in pre-stored candidate information disaggregated model
Each candidate information disaggregated model, using the candidate information disaggregated model as candidate information disaggregated model to be assessed, above-mentioned electricity
Sub- equipment can be based on above-mentioned sample data sets and related respectively to each sample data in above-mentioned sample data sets
The classification of connection determines the accuracy rate of above-mentioned candidate information disaggregated model to be assessed.Later, above-mentioned electronic equipment can be based on institute really
Fixed accuracy rate determines target information disaggregated model in above-mentioned pre-stored candidate information disaggregated model.Finally, above-mentioned electricity
Sub- equipment can utilize machine learning method, based on above-mentioned sample data sets and with each in above-mentioned sample data sets
Sample data distinguishes associated classification, is trained to obtain information disaggregated model to above-mentioned target information disaggregated model.Here,
Cross validation method may be used to determine the accuracy rate of above-mentioned candidate information disaggregated model to be assessed in above-mentioned electronic equipment.And
Above-mentioned electronic equipment can classify the highest candidate information of accuracy rate in above-mentioned pre-stored candidate information disaggregated model
Model is as target information disaggregated model.If the highest candidate information disaggregated model more than one of accuracy rate, above-mentioned electronic equipment
A candidate information disaggregated model can be randomly selected out from the highest candidate information disaggregated model of the accuracy rate as target
Information disaggregated model.It should be noted that above-mentioned cross validation method is the known technology studied and applied extensively at present, herein
It repeats no more.
In some optional realization methods of the present embodiment, for each classification sample in above-mentioned sample data sets
Data, above-mentioned electronic equipment can determine weighted value associated with category sample data, wherein the weighted value can be should
The ratio of the total number for the sample data that the number of classification sample data is included with above-mentioned sample data sets.For example, above-mentioned
The total number for the sample data that sample data is included is 1000, the sample for a certain classification that above-mentioned sample data sets are included
The number of data is 100, then the ratio that weighted value associated with category sample data is 100 and 1000, i.e., 10%.So
Afterwards, above-mentioned electronic equipment can determine the accuracy rate of above-mentioned candidate information disaggregated model to be assessed according to identified weighted value.
Here, above-mentioned electronic equipment can determine above-mentioned wait for by executing flow 400 as shown in Figure 4 according to identified weighted value
Assess the accuracy rate of candidate information disaggregated model.
Step 204, it is matched from the feedback information set that prestores indicated by above-mentioned classification associated with above- mentioned information
Object feedback information, and object feedback information is exported to client.
In the present embodiment, each classification that can be identified for above- mentioned information disaggregated model, above-mentioned electronic equipment
Locally or there can be the feedback information set that prestores associated with the category with the server that above-mentioned electronic equipment remotely connects, it should
The feedback information set that prestores may include and the relevant feedback information of the information of the category.For example, above- mentioned information disaggregated model institute
Some classification that can recognize that is " order consulting ", then the feedback information set that prestores associated with " order consulting " classification
Can include to seek advice from the relevant feedback information of category information with order, such as " you can Click here checks and accordingly order feedback information
Subsequent ' deletion ' option of odd numbers, deletes the order purchaser record.Deleting order can be in my order page order recycle bin
The inside views ".Moreover, for each classification that above- mentioned information disaggregated model can be identified, above-mentioned electronic equipment is local
Or it can be prestored with the server that above-mentioned electronic equipment remotely connects for characterizing the category and associated with the category
The feedback information set that prestores correspondence information list.Each information in the information list may include classification and
The mark of the feedback information set that prestores associated with the category.
In the present embodiment, when above-mentioned electronic equipment the classification belonging to above- mentioned information is obtained by executing above-mentioned steps 203
Afterwards, the feedback information set that prestores indicated by above-mentioned classification, and then above-mentioned electricity can be determined by reading above- mentioned information list
Sub- equipment can match object feedback information associated with above- mentioned information in the prestored information feedback set, and by the mesh
Mark feedback information is exported to above-mentioned client.Here, above-mentioned electronic equipment can be by calculating above- mentioned information and above-mentioned classification institute
The matching degree of each feedback information in the feedback information set that prestores indicated, by the sequence that matching degree is descending, from this
The second predetermined number feedback information is selected in the feedback information set that prestores as object feedback information.
In some optional realization methods of the present embodiment, for prestoring instead indicated by the classification belonging to above- mentioned information
The matching degree of each feedback information in feedforward information set, this feedback information and above- mentioned information can be this feedback information
Including above- mentioned information Feature Words number and the above- mentioned information Feature Words that are included total number ratio.As an example,
Above- mentioned information is " how big household electrical appliances mounting cost is calculated ", and the Feature Words of above- mentioned information are " big household electrical appliances " and " mounting cost "；It is above-mentioned
A certain feedback information in the feedback information set that prestores indicated by classification is that " different manufacturers mounting cost fee-collection standard is not
Together, it is proposed that the customer service of consulting producer understands, and air-conditioning mounting cost can be checked in the buyer's guide page ".Due to this bar feedback information packet
The Feature Words " mounting cost " of above- mentioned information are included, therefore the matching degree of this feedback information and above- mentioned information can be 1 and 2 ratio
Value, i.e., 50%.
It is an application scenarios corresponding with embodiment shown in Fig. 2 with continued reference to Fig. 3, Fig. 3.In the application scenarios of Fig. 3
In, as shown in label 301, user passes through client sending information information " how order is cancelled " first.Later, such as label 302
Shown, server can extract feature vector (3,1) in above-mentioned text message, wherein 3 indicate Feature Words " order ", 1 table
Show Feature Words " cancellation ".Then, as shown in label 303, features described above vector can be imported information trained in advance by server
Disaggregated model is classified to obtain the classification " order consulting " belonging to above-mentioned text message.Finally, as shown in label 304, service
Device can match target associated with above-mentioned text message from the feedback information set that prestores indicated by " order consulting "
Feedback information " please Click here and cancel in respective orders application.If cancelling failure because commodity have been sent out, it is proposed that reject
(specialty goods is such as：It is fresh, luxury goods, except customization class etc.) ", and the object feedback information is exported to above-mentioned client.
Information output method shown in the present embodiment, be effectively utilized in advance trained information disaggregated model to the information into
Row classification, improves the efficiency of information output.
With further reference to Fig. 4, it illustrates weighted values determined by the basis according to the application to determine candidate letter to be assessed
Cease the flow 400 of the accuracy rate of disaggregated model.The flow 400 includes the following steps：
Step 401, sample data sets are divided into training set and test set.
In the present embodiment, above-mentioned sample data sets can be divided into training set and test set by above-mentioned electronic equipment,
Wherein, above-mentioned training set and above-mentioned test set include the sample data of the same category.Here, the training that above-mentioned training set is included
Sample data and the test sample data that above-mentioned test set is included are typically mutually different.And above-mentioned training set is included
The numbers of training sample data be typically more than the numbers of the test sample data that above-mentioned test set is included.On as an example,
It includes sample data a1, a2, a3, b1, b2, b3 to state sample data sets, and the wherein classification belonging to a1, a2, a3 is noted as a,
Classification belonging to b1, b2, b3 is noted as b, then it is pre- can to randomly select first in a classification sample datas for above-mentioned electronic equipment
Fixed number mesh sample data is as test sample, if a1 is by as test sample, above-mentioned electronic equipment can make a2 and a3
For training sample；Similarly, if b2 in b classification sample datas is by as test sample, above-mentioned electronic equipment can by b1 and
B3 is as training sample；Above-mentioned electronic equipment can be using test sample a1 and b2 as test set, by training sample a2, a3, b1
With b3 as training set.
Step 402, using machine learning method, divide based on training set and with each training sample data in training set
Not associated classification is trained candidate information disaggregated model to be assessed candidate information disaggregated model after being trained.
In the present embodiment, above-mentioned electronic equipment can utilize machine learning method, by above-mentioned training set and with it is above-mentioned
Associated classification is trained candidate information disaggregated model to be assessed to each training sample data in training set respectively,
Obtain to establish each training sample data in above-mentioned training set and classification associated with the training sample data it
Between accurate correspondence training after candidate information disaggregated model.
Step 403, using the candidate information disaggregated model after training to each test sample data in test set
Classification is predicted to obtain prediction result.
In the present embodiment, for each test sample data in above-mentioned test set, above-mentioned electronic equipment can be adopted
With method identical with above-mentioned steps 206, the feature vector of the test sample data is gone out from the test sample extracting data, and
This feature vector is imported into the candidate information disaggregated model after above-mentioned training and carries out class prediction, obtains prediction result.
Step 404, it for each classification test sample data in test set, is determined based on prediction result and is surveyed with the category
The associated predictablity rate of sample notebook data will be included to have with category test sample data with sample data sets
The associated weighted value of sample data of the same category and the product of above-mentioned predictablity rate as with category test sample number
According to associated weight estimation accuracy rate.
In the present embodiment, for each classification test sample data in above-mentioned test set, above-mentioned electronic equipment can be with
Predictablity rate associated with category test sample data is determined based on above-mentioned prediction result.Wherein, the predictablity rate
Can be candidate information disaggregated model after above-mentioned training to the correct numbers of class prediction of category test sample data with
The ratio of the number of category test sample data.Here, above-mentioned electronic equipment can will be with category test sample data phase
Associated prediction result is compared with the concrete class of category test sample data, to obtain the candidate after above-mentioned training
Class prediction correct number of the information disaggregated model to category test sample data.Later, above-mentioned electronic equipment can incite somebody to action
Included with above-mentioned sample data sets has the sample data of the same category associated with category test sample data
Weighted value is with the product of the predictablity rate as weight estimation accuracy rate associated with category test sample data.
As an example, above-mentioned sample data sets include 1000 sample datas, wherein having belonging to 200 sample datas
Classification is noted as a, has the classification of 400 sample datas to be noted as b, and the classification of remaining 400 sample data is noted as
c.Weighted value associated with a classification sample datas is 20%.It is related respectively to b classifications sample data and c classification sample datas
The weighted value of connection is 40%.It is assumed that above-mentioned test set includes each 50 of the test sample data of a, b, c classification, and above-mentioned electricity
Sub- equipment has determined that the candidate information disaggregated model after above-mentioned training correctly counts the class prediction of a classification test sample data
Mesh is 45, and the correct number of class prediction to b classification test sample data is 46, to the classification of c classification test sample data
Predict that correct number is 44.So for each classification test sample data in above-mentioned test set, with a classification test samples
The associated predictablity rate of data can be 45 and 50 ratio, i.e., 90%, it is associated with a classification test sample data add
Weighing predictablity rate can be by 20% (weighted value associated with a classification sample datas that above-mentioned sample data sets include)
With 90% product, i.e., 16%；Predictablity rate associated with b classification test sample data can be 46 and 50 ratio,
I.e. 92%, weight estimation accuracy rate associated with b classification test sample data can be 40% (with above-mentioned sample data sets
Including the associated weighted value of b classification sample datas) with 92% product, i.e., 36.8%；With c classification test sample data
Associated predictablity rate can be 44 and 50 ratio, i.e., 88%, it is associated with c classification test sample data weighting in advance
Survey accuracy rate can be 40% (weighted value associated with the c classification sample datas that above-mentioned sample data sets include) and
88% product, i.e., 35.2%.
Step 405, by weight estimation accuracy rate associated with each classification test sample data difference in test set
The numerical value of gained is added as first weight estimation accuracy rate associated with candidate information disaggregated model to be assessed.
In the present embodiment, above-mentioned electronic equipment can will divide with each classification test sample data in above-mentioned test set
Not associated weight estimation accuracy rate is added the numerical value of gained as associated with above-mentioned candidate information disaggregated model to be assessed
The first weight estimation accuracy rate.Continue by taking the example in step 404 as an example, it is known that with a classification test samples in test set
The associated weight estimation accuracy rate of data is 16%, and weighting associated with the b classification test sample data in test set is pre-
It is 36.8% to survey accuracy rate, and weight estimation accuracy rate associated with the c classification test sample data in test set is 35.2%,
So 16%, 36.8% and 35.2% be added gained numerical value 88% can as with above-mentioned candidate information disaggregated model to be assessed
Associated first weight estimation accuracy rate.
In some optional realization methods of the present embodiment, above-mentioned electronic equipment can record above-mentioned first weight estimation
In accuracy rate to local memory or hard disk.
Step 406, candidate information disaggregated model to be assessed is restored to unbred state.
In the present embodiment, to obtain first weight estimation associated with above-mentioned candidate information disaggregated model to be assessed accurate
After true rate, above-mentioned electronic equipment can restore above-mentioned candidate information disaggregated model to be assessed to unbred state, in case
Subsequent training uses.
In some optional realization methods of the present embodiment, above-mentioned electronic equipment divides by above-mentioned candidate information to be assessed
Class model restores to unbred state, can be executed to the cycle of above-mentioned steps 401 to above-mentioned steps 406 with recording needle secondary
Number.
Step 407, determine that the cycle for above-mentioned steps 401 to above-mentioned steps 406 executes whether number reaches predetermined number
Contents.
In the present embodiment, above-mentioned electronic equipment can will be recorded for above-mentioned after having executed above-mentioned steps 406
The cycle of step 401 to above-mentioned steps 406 executes number and is compared with predetermined number, if the cycle executes number less than predetermined
Number, above-mentioned electronic equipment can go to step 401；If the ring, which executes number, reaches predetermined number contents, above-mentioned electronic equipment can
To go to step 408.
Step 408, using the average value of the first weight estimation accuracy rate of the predetermined number of gained as candidate letter to be assessed
Cease the accuracy rate of disaggregated model.
In the present embodiment, above-mentioned electronic equipment has executed above-mentioned steps 401 to 406 predetermined number of above-mentioned steps in cycle
It, can be using the average value of the first weight estimation accuracy rate of obtained above-mentioned predetermined number as above-mentioned candidate to be assessed after secondary
The accuracy rate of information disaggregated model.
Embodiment shown in Fig. 4, it is associated with each classification sample data in above-mentioned sample data sets by combining
Weighted value, to determine the accuracy rate of above-mentioned candidate information disaggregated model to be assessed, to improve to above-mentioned candidate to be assessed
The accuracy that the accuracy rate of information disaggregated model is assessed.
With further reference to Fig. 5, as the realization to method shown in above-mentioned each figure, this application provides a kind of outputs of information to fill
The one embodiment set, the device embodiment is corresponding with embodiment of the method shown in Fig. 2, which specifically can be applied to respectively
In kind electronic equipment.
As shown in figure 5, information output apparatus 500 shown in the present embodiment includes：Receiving unit 501, extraction unit 502,
Taxon 503 and output unit 504.Wherein, receiving unit 501 is configured to receive the information that client is sent；Extraction is single
Member 502 is configured to extract feature vector from above- mentioned information, wherein features described above vector is for characterizing the interior of above- mentioned information
Hold；Taxon 503 is configured to be classified to obtain by features described above vector importing information disaggregated model trained in advance
State the classification belonging to information；Output unit 504 is configured to match from the feedback information set that prestores indicated by above-mentioned classification
Go out object feedback information associated with above- mentioned information, and above-mentioned object feedback information is exported to above-mentioned client.
In the present embodiment, in information output apparatus 500：Receiving unit 501, extraction unit 502,503 and of taxon
The specific processing of output unit 504 and its caused technique effect can respectively with reference in 2 corresponding embodiment of figure step 201,
The related description of step 202, step 203 and step 204, details are not described herein.
In some optional realization methods of the present embodiment, above-mentioned apparatus 500 can also include：Information disaggregated model is built
Vertical unit (not shown), is configured to establish information disaggregated model, may include：Subelement (not shown) is obtained,
It is configured to obtain sample data sets and class associated with each sample data difference in above-mentioned sample data sets
Not；Accuracy rate determination subelement (not shown) is configured to for every in pre-stored candidate information disaggregated model
One candidate information disaggregated model, using the candidate information disaggregated model as candidate information disaggregated model to be assessed, based on above-mentioned
Sample data sets and classification associated with each sample data difference in above-mentioned sample data sets, determine above-mentioned wait for
Assess the accuracy rate of candidate information disaggregated model, wherein above-mentioned pre-stored candidate information disaggregated model is unbred
Model；Target information disaggregated model determination subelement (not shown), is configured to based on identified accuracy rate, above-mentioned
Target information disaggregated model is determined in pre-stored candidate information disaggregated model；Training subelement (not shown), configuration
For utilizing machine learning method, based on above-mentioned sample data sets and with each sample number in above-mentioned sample data sets
According to associated classification respectively, above-mentioned target information disaggregated model is trained to obtain information disaggregated model.
In some optional realization methods of the present embodiment, above-mentioned accuracy rate determination subelement may include：First is accurate
Rate determining module (not shown) is configured to determine above-mentioned candidate information classification mould to be assessed using cross validation method
The accuracy rate of type.
In some optional realization methods of the present embodiment, above-mentioned accuracy rate determination subelement may include：Weighted value
Determining module (not shown), is configured to for each classification sample data in above-mentioned sample data sets, determine with
The associated weighted value of category sample data, wherein above-mentioned weighted value is the number of category sample data and above-mentioned sample
The ratio of the total number for the sample data that data acquisition system is included；Second accuracy rate determining module (not shown), configuration are used
According to identified weighted value, the accuracy rate of above-mentioned candidate information disaggregated model to be assessed is determined.
In some optional realization methods of the present embodiment, above-mentioned second accuracy rate determining module may include：Processing
Submodule (not shown) is configured to cycle and executes following processing step predetermined number contents：By above-mentioned sample data sets
Be divided into training set and test set, using machine learning method, based on above-mentioned training set and with each in above-mentioned training set
Training sample data are after associated classification is trained above-mentioned candidate information disaggregated model to be assessed and is trained respectively
Candidate information disaggregated model, using the candidate information disaggregated model after above-mentioned training to each test specimens in above-mentioned test set
The classification of notebook data is predicted to obtain prediction result, for each classification test sample data in above-mentioned test set, is based on
Above-mentioned prediction result determines predictablity rate associated with category test sample data, will be with above-mentioned sample data sets institute
Including with category test sample data have the associated weighted value of the sample data of the same category and above-mentioned prediction it is accurate
The product of rate as weight estimation accuracy rate associated with category test sample data, by with it is each in above-mentioned test set
Classification test sample data distinguish associated weight estimation accuracy rate be added the numerical value of gained as with above-mentioned candidate to be assessed
The associated first weight estimation accuracy rate of information disaggregated model, by above-mentioned candidate information disaggregated model to be assessed restore to without
Trained state, wherein above-mentioned training set and above-mentioned test set include the sample data of the same category, in above-mentioned test set
Each classification test sample data, predictablity rate associated with category test sample data be above-mentioned training after time
Select class prediction correct number and category test sample data of the information disaggregated model to category test sample data
The ratio of number；Accuracy rate determination sub-module (not shown) is configured to add the above-mentioned predetermined number of gained first
Weigh accuracy rate of the average value of predictablity rate as above-mentioned candidate information disaggregated model to be assessed.
In some optional realization methods of the present embodiment, above-mentioned target information disaggregated model determination subelement can wrap
It includes：Target information disaggregated model determining module (not shown) is configured to classify above-mentioned pre-stored candidate information
The highest candidate information disaggregated model of accuracy rate in model is as target information disaggregated model.
Below with reference to Fig. 6, it illustrates the computer systems 600 suitable for the server for realizing the embodiment of the present application
Structural schematic diagram.Server shown in Fig. 6 is only an example, should not be to the function and use scope band of the embodiment of the present application
Carry out any restrictions.
As shown in fig. 6, computer system 600 includes central processing unit (CPU) 601, it can be read-only according to being stored in
Program in memory (ROM) 602 or be loaded into the program in random access storage device (RAM) 603 from storage section 608 and
Execute various actions appropriate and processing.In RAM 603, also it is stored with system 600 and operates required various programs and data.
CPU 601, ROM 602 and RAM 603 are connected with each other by bus 604.Input/output (I/O) interface 605 is also connected to always
Line 604.
It is connected to I/O interfaces 605 with lower component：Importation 606 including keyboard, mouse etc.；It is penetrated including such as cathode
The output par, c 607 of spool (CRT), liquid crystal display (LCD) etc. and loud speaker etc.；Storage section 608 including hard disk etc.；
And the communications portion 609 of the network interface card including LAN card, modem etc..Communications portion 609 via such as because
The network of spy's net executes communication process.Driver 610 is also according to needing to be connected to I/O interfaces 605.Detachable media 611, such as
Disk, CD, magneto-optic disk, semiconductor memory etc. are mounted on driver 610, as needed in order to be read from thereon
Computer program be mounted into storage section 608 as needed.
Particularly, in accordance with an embodiment of the present disclosure, it may be implemented as computer above with reference to the process of flow chart description
Software program.For example, embodiment of the disclosure includes a kind of computer program product comprising be carried on computer-readable medium
On computer program, which includes the program code for method shown in execution flow chart.In such reality
It applies in example, which can be downloaded and installed by communications portion 609 from network, and/or from detachable media
611 are mounted.When the computer program is executed by central processing unit (CPU) 601, executes and limited in the system of the application
Above-mentioned function.
It should be noted that computer-readable medium shown in the application can be computer-readable signal media or meter
Calculation machine readable storage medium storing program for executing either the two arbitrarily combines.Computer readable storage medium for example can be --- but not
Be limited to --- electricity, magnetic, optical, electromagnetic, infrared ray or semiconductor system, device or device, or arbitrary above combination.Meter
The more specific example of calculation machine readable storage medium storing program for executing can include but is not limited to：Electrical connection with one or more conducting wires, just
It takes formula computer disk, hard disk, random access storage device (RAM), read-only memory (ROM), erasable type and may be programmed read-only storage
Device (EPROM or flash memory), optical fiber, portable compact disc read-only memory (CD-ROM), light storage device, magnetic memory device,
Or above-mentioned any appropriate combination.In this application, can be any include computer readable storage medium or storage journey
The tangible medium of sequence, the program can be commanded the either device use or in connection of execution system, device.And at this
In application, computer-readable signal media may include in a base band or as the data-signal that a carrier wave part is propagated,
Wherein carry computer-readable program code.Diversified forms may be used in the data-signal of this propagation, including but unlimited
In electromagnetic signal, optical signal or above-mentioned any appropriate combination.Computer-readable signal media can also be that computer can
Any computer-readable medium other than storage medium is read, which can send, propagates or transmit and be used for
By instruction execution system, device either device use or program in connection.Include on computer-readable medium
Program code can transmit with any suitable medium, including but not limited to：Wirelessly, electric wire, optical cable, RF etc. or above-mentioned
Any appropriate combination.
Flow chart in attached drawing and block diagram, it is illustrated that according to the system of the various embodiments of the application, method and computer journey
The architecture, function and operation in the cards of sequence product.In this regard, each box in flowchart or block diagram can generation
A part for a part for one module, program segment, or code of table, above-mentioned module, program segment, or code includes one or more
Executable instruction for implementing the specified logical function.It should also be noted that in some implementations as replacements, institute in box
The function of mark can also occur in a different order than that indicated in the drawings.For example, two boxes succeedingly indicated are practical
On can be basically executed in parallel, they can also be executed in the opposite order sometimes, this is depended on the functions involved.Also it wants
It is noted that the combination of each box in block diagram or flow chart and the box in block diagram or flow chart, can use and execute rule
The dedicated hardware based systems of fixed functions or operations is realized, or can use the group of specialized hardware and computer instruction
It closes to realize.
Being described in unit involved in the embodiment of the present application can be realized by way of software, can also be by hard
The mode of part is realized.Described unit can also be arranged in the processor, for example, can be described as：A kind of processor packet
Include receiving unit, extraction unit, taxon and output unit.Wherein, the title of these units not structure under certain conditions
The restriction of the pairs of unit itself, for example, receiving unit is also described as " receiving the unit for the information that client is sent ".
As on the other hand, present invention also provides a kind of computer-readable medium, which can be
Included in device described in above-described embodiment；Can also be individualism, and without be incorporated the device in.Above-mentioned calculating
Machine readable medium carries one or more program, when said one or multiple programs are executed by the device, makes
Obtaining the device includes：Receive the information that client is sent；Feature vector is extracted from above- mentioned information, wherein features described above to
Measure the content for characterizing above- mentioned information；Features described above vector is imported information disaggregated model trained in advance to be classified to obtain
Classification belonging to above- mentioned information；It is matched from the feedback information set that prestores indicated by above-mentioned classification associated with above- mentioned information
Object feedback information, and above-mentioned object feedback information is exported to above-mentioned client.
Above description is only the preferred embodiment of the application and the explanation to institute's application technology principle.People in the art
Member should be appreciated that invention scope involved in the application, however it is not limited to technology made of the specific combination of above-mentioned technical characteristic
Scheme, while should also cover in the case where not departing from foregoing invention design, it is carried out by above-mentioned technical characteristic or its equivalent feature
Other technical solutions of arbitrary combination and formation.Such as features described above has similar work(with (but not limited to) disclosed herein
Can technical characteristic replaced mutually and the technical solution that is formed.
Claims (10)
1. a kind of information output method, which is characterized in that the method includes：
Receive the information that client is sent；
Feature vector is extracted from described information, wherein described eigenvector is used to characterize the content of described information；
Described eigenvector is imported into information disaggregated model trained in advance and is classified to obtain the classification belonging to described information；
Object feedback information associated with described information is matched from the feedback information set that prestores indicated by the classification,
And the object feedback information is exported to the client.
2. according to the method described in claim 1, it is characterized in that, the method further includes establishing the step of information disaggregated model
Suddenly, described the step of establishing information disaggregated model, includes:
Obtain sample data sets and classification associated with each sample data difference in the sample data sets；
For each candidate information disaggregated model in pre-stored candidate information disaggregated model, by the candidate information point
Class model as candidate information disaggregated model to be assessed, based on the sample data sets and in the sample data sets
Each sample data distinguishes associated classification, determines the accuracy rate of the candidate information disaggregated model to be assessed, wherein institute
It is unbred model to state pre-stored candidate information disaggregated model；
Based on identified accuracy rate, target information classification mould is determined in the pre-stored candidate information disaggregated model
Type；
Using machine learning method, based on the sample data sets and with each sample number in the sample data sets
According to associated classification respectively, the target information disaggregated model is trained to obtain information disaggregated model.
3. according to the method described in claim 2, it is characterized in that, the determination candidate information disaggregated model to be assessed
Accuracy rate, including：
The accuracy rate of the candidate information disaggregated model to be assessed is determined using cross validation method.
4. according to the method described in claim 2, it is characterized in that, the determination candidate information disaggregated model to be assessed
Accuracy rate, including：
For each classification sample data in the sample data sets, power associated with the classification sample data is determined
Weight values, wherein the sample number that the weighted value includes with the sample data sets by the number of the classification sample data
According to total number ratio；
According to identified weighted value, the accuracy rate of the candidate information disaggregated model to be assessed is determined.
5. according to the method described in claim 4, it is characterized in that, weighted value determined by the basis, determines described to be evaluated
Estimate the accuracy rate of candidate information disaggregated model, including：
Cycle executes following processing step predetermined number contents：The sample data sets are divided into training set and test set, profit
It is based on the training set and associated with each training sample data difference in the training set with machine learning method
Classification is trained the candidate information disaggregated model after being trained to the candidate information disaggregated model to be assessed, using described
Candidate information disaggregated model after training is predicted to obtain to the classification of each test sample data in the test set
Prediction result determines and the class each classification test sample data in the test set based on the prediction result
The other associated predictablity rate of test sample data will be included and the classification test specimens with the sample data sets
Notebook data have the product of the associated weighted value of sample data and the predictablity rate of the same category as with the class
The other associated weight estimation accuracy rate of test sample data will divide with each classification test sample data in the test set
Not associated weight estimation accuracy rate is added the numerical value of gained as associated with the candidate information disaggregated model to be assessed
The first weight estimation accuracy rate, the candidate information disaggregated model to be assessed is restored to unbred state, wherein institute
It states training set and the test set includes the sample data of the same category, prediction associated with the classification test sample data
Accuracy rate is that the candidate information disaggregated model after the training correctly counts the class prediction of the classification test sample data
The ratio of mesh and the number of the classification test sample data；
Using the average value of the first weight estimation accuracy rate of the predetermined number of gained as the candidate information to be assessed point
The accuracy rate of class model.
6. according to the method described in one of claim 2-5, which is characterized in that it is described based on identified accuracy rate, described
Target information disaggregated model is determined in pre-stored candidate information disaggregated model, including：
Using the highest candidate information disaggregated model of accuracy rate in the pre-stored candidate information disaggregated model as target
Information disaggregated model.
7. a kind of information output apparatus, which is characterized in that described device includes：
Receiving unit is configured to receive the information that client is sent；
Extraction unit is configured to extract feature vector from described information, wherein described eigenvector is described for characterizing
The content of information；
Taxon is configured to be classified to obtain by described eigenvector importing information disaggregated model trained in advance described
Classification belonging to information；
Output unit is configured to match from the feedback information set that prestores indicated by the classification related to described information
The object feedback information of connection, and the object feedback information is exported to the client.
8. device according to claim 7, which is characterized in that described device further includes：Information disaggregated model establishes unit,
It is configured to establish information disaggregated model, including：
Obtain subelement, be configured to obtain sample data sets and with each sample data in the sample data sets
Associated classification respectively；
Accuracy rate determination subelement is configured to for each candidate information in pre-stored candidate information disaggregated model
Disaggregated model is based on the sample data set using the candidate information disaggregated model as candidate information disaggregated model to be assessed
Conjunction and classification associated with each sample data difference in the sample data sets determine the candidate letter to be assessed
Cease the accuracy rate of disaggregated model, wherein the pre-stored candidate information disaggregated model is unbred model；
Target information disaggregated model determination subelement is configured to based on identified accuracy rate, in the pre-stored time
It selects and determines target information disaggregated model in information disaggregated model；
Training subelement, be configured to utilize machine learning method, based on the sample data sets and with the sample data
Each sample data in set distinguishes associated classification, is trained to obtain information to the target information disaggregated model
Disaggregated model.
9. a kind of server, including：
One or more processors；
Storage device, for storing one or more programs；
When one or more of programs are executed by one or more of processors so that one or more of processors are real
The now method as described in any in claim 1-6.
10. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that described program is handled
The method as described in any in claim 1-6 is realized when device executes.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710063291.2A CN108388563B (en) | 2017-02-03 | 2017-02-03 | Information output method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710063291.2A CN108388563B (en) | 2017-02-03 | 2017-02-03 | Information output method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108388563A true CN108388563A (en) | 2018-08-10 |
| CN108388563B CN108388563B (en) | 2022-11-08 |
Family
ID=63075842
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710063291.2A Active CN108388563B (en) | 2017-02-03 | 2017-02-03 | Information output method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108388563B (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109146076A (en) * | 2018-08-13 | 2019-01-04 | 东软集团股份有限公司 | model generating method and device, data processing method and device |
| CN109376868A (en) * | 2018-09-30 | 2019-02-22 | 北京字节跳动网络技术有限公司 | Information management system |
| CN109409419A (en) * | 2018-09-30 | 2019-03-01 | 北京字节跳动网络技术有限公司 | Method and apparatus for handling data |
| CN109559239A (en) * | 2018-11-26 | 2019-04-02 | 泰康保险集团股份有限公司 | Generation method, device, electronic equipment, storage medium are suggested in complaint handling |
| CN109670111A (en) * | 2018-12-20 | 2019-04-23 | 北京字节跳动网络技术有限公司 | Method and apparatus for pushed information |
| CN109740156A (en) * | 2018-12-28 | 2019-05-10 | 北京金山安全软件有限公司 | Feedback information processing method and device, electronic equipment and storage medium |
| CN110909074A (en) * | 2019-10-21 | 2020-03-24 | 北京海益同展信息科技有限公司 | Method and device for processing social data, computer equipment and storage medium |
| CN111159465A (en) * | 2019-12-31 | 2020-05-15 | 杭州网易云音乐科技有限公司 | Song classification method and device |
Citations (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101127029A (en) * | 2007-08-24 | 2008-02-20 | 复旦大学 | Method for training SVM classifier in large scale data classification |
| CN102254193A (en) * | 2011-07-16 | 2011-11-23 | 西安电子科技大学 | Relevance vector machine-based multi-class data classifying method |
| CN102799635A (en) * | 2012-06-27 | 2012-11-28 | 天津大学 | Image set ordering method driven by user |
| CN103136377A (en) * | 2013-03-26 | 2013-06-05 | 重庆邮电大学 | Chinese text classification method based on evolution super-network |
| CN103258147A (en) * | 2013-05-24 | 2013-08-21 | 重庆邮电大学 | Parallel evolution super-network DNA micro array gene data sorting system and method based on GPU |
| CN103425677A (en) * | 2012-05-18 | 2013-12-04 | 阿里巴巴集团控股有限公司 | Method for determining classified models of keywords and method and device for classifying keywords |
| CN104537252A (en) * | 2015-01-05 | 2015-04-22 | 深圳市腾讯计算机系统有限公司 | User state single-classification model training method and device |
| CN104809434A (en) * | 2015-04-22 | 2015-07-29 | 哈尔滨工业大学 | Sleep staging method based on single-channel electroencephalogram signal ocular artifact removal |
| US20150302363A1 (en) * | 2013-01-06 | 2015-10-22 | Huawei Technologies Co., Ltd. | Meeting Scheduling Method, Device, and System |
| CN105447750A (en) * | 2015-11-17 | 2016-03-30 | 小米科技有限责任公司 | Information identification method, apparatus, terminal and server |
| US20160092790A1 (en) * | 2014-09-25 | 2016-03-31 | Samsung Eletrônica da Amazônia Ltda. | Method for multiclass classification in open-set scenarios and uses thereof |
| CN105630917A (en) * | 2015-12-22 | 2016-06-01 | 成都小多科技有限公司 | Intelligent answering method and intelligent answering device |
| CN105786969A (en) * | 2016-02-01 | 2016-07-20 | 百度在线网络技术（北京）有限公司 | Information display method and apparatus |
| CN105893465A (en) * | 2016-03-28 | 2016-08-24 | 北京京东尚科信息技术有限公司 | Automatic question answering method and device |
| CN105930723A (en) * | 2016-04-20 | 2016-09-07 | 福州大学 | Intrusion detection method based on feature selection |
| CN106156809A (en) * | 2015-04-24 | 2016-11-23 | 阿里巴巴集团控股有限公司 | For updating the method and device of disaggregated model |
-
2017
- 2017-02-03 CN CN201710063291.2A patent/CN108388563B/en active Active
Patent Citations (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101127029A (en) * | 2007-08-24 | 2008-02-20 | 复旦大学 | Method for training SVM classifier in large scale data classification |
| CN102254193A (en) * | 2011-07-16 | 2011-11-23 | 西安电子科技大学 | Relevance vector machine-based multi-class data classifying method |
| CN103425677A (en) * | 2012-05-18 | 2013-12-04 | 阿里巴巴集团控股有限公司 | Method for determining classified models of keywords and method and device for classifying keywords |
| CN102799635A (en) * | 2012-06-27 | 2012-11-28 | 天津大学 | Image set ordering method driven by user |
| US20150302363A1 (en) * | 2013-01-06 | 2015-10-22 | Huawei Technologies Co., Ltd. | Meeting Scheduling Method, Device, and System |
| CN103136377A (en) * | 2013-03-26 | 2013-06-05 | 重庆邮电大学 | Chinese text classification method based on evolution super-network |
| CN103258147A (en) * | 2013-05-24 | 2013-08-21 | 重庆邮电大学 | Parallel evolution super-network DNA micro array gene data sorting system and method based on GPU |
| US20160092790A1 (en) * | 2014-09-25 | 2016-03-31 | Samsung Eletrônica da Amazônia Ltda. | Method for multiclass classification in open-set scenarios and uses thereof |
| CN104537252A (en) * | 2015-01-05 | 2015-04-22 | 深圳市腾讯计算机系统有限公司 | User state single-classification model training method and device |
| CN104809434A (en) * | 2015-04-22 | 2015-07-29 | 哈尔滨工业大学 | Sleep staging method based on single-channel electroencephalogram signal ocular artifact removal |
| CN106156809A (en) * | 2015-04-24 | 2016-11-23 | 阿里巴巴集团控股有限公司 | For updating the method and device of disaggregated model |
| CN105447750A (en) * | 2015-11-17 | 2016-03-30 | 小米科技有限责任公司 | Information identification method, apparatus, terminal and server |
| CN105630917A (en) * | 2015-12-22 | 2016-06-01 | 成都小多科技有限公司 | Intelligent answering method and intelligent answering device |
| CN105786969A (en) * | 2016-02-01 | 2016-07-20 | 百度在线网络技术（北京）有限公司 | Information display method and apparatus |
| CN105893465A (en) * | 2016-03-28 | 2016-08-24 | 北京京东尚科信息技术有限公司 | Automatic question answering method and device |
| CN105930723A (en) * | 2016-04-20 | 2016-09-07 | 福州大学 | Intrusion detection method based on feature selection |
Non-Patent Citations (2)
| Title |
|---|
| 崔清亮等: "多核学习方法在分类中的应用研究", 《科学技术与工程》 * |
| 熊彪等: "基于高斯混合模型的遥感影像半监督分类", 《武汉大学学报(信息科学版)》 * |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109146076A (en) * | 2018-08-13 | 2019-01-04 | 东软集团股份有限公司 | model generating method and device, data processing method and device |
| CN109376868A (en) * | 2018-09-30 | 2019-02-22 | 北京字节跳动网络技术有限公司 | Information management system |
| CN109409419A (en) * | 2018-09-30 | 2019-03-01 | 北京字节跳动网络技术有限公司 | Method and apparatus for handling data |
| CN109409419B (en) * | 2018-09-30 | 2021-05-07 | 北京字节跳动网络技术有限公司 | Method and apparatus for processing data |
| CN109376868B (en) * | 2018-09-30 | 2021-06-25 | 北京字节跳动网络技术有限公司 | Information management system |
| CN109559239A (en) * | 2018-11-26 | 2019-04-02 | 泰康保险集团股份有限公司 | Generation method, device, electronic equipment, storage medium are suggested in complaint handling |
| CN109670111A (en) * | 2018-12-20 | 2019-04-23 | 北京字节跳动网络技术有限公司 | Method and apparatus for pushed information |
| CN109740156A (en) * | 2018-12-28 | 2019-05-10 | 北京金山安全软件有限公司 | Feedback information processing method and device, electronic equipment and storage medium |
| CN109740156B (en) * | 2018-12-28 | 2023-08-04 | 北京金山安全软件有限公司 | Feedback information processing method and device, electronic equipment and storage medium |
| CN110909074A (en) * | 2019-10-21 | 2020-03-24 | 北京海益同展信息科技有限公司 | Method and device for processing social data, computer equipment and storage medium |
| CN111159465A (en) * | 2019-12-31 | 2020-05-15 | 杭州网易云音乐科技有限公司 | Song classification method and device |
| CN111159465B (en) * | 2019-12-31 | 2023-09-29 | 杭州网易云音乐科技有限公司 | Song classification method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108388563B (en) | 2022-11-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108388563A (en) | Information output method and device | |
| CN109460513A (en) | Method and apparatus for generating clicking rate prediction model | |
| CN107908789A (en) | Method and apparatus for generating information | |
| CN107105031A (en) | Information-pushing method and device | |
| CN109460514A (en) | Method and apparatus for pushed information | |
| CN108520324A (en) | Method and apparatus for generating information | |
| CN108805594A (en) | Information-pushing method and device | |
| CN109922032A (en) | Method and apparatus for determining the risk of logon account | |
| CN108171191B (en) | Method and apparatus for detecting face | |
| CN108776692A (en) | Method and apparatus for handling information | |
| CN109634833A (en) | A kind of Software Defects Predict Methods and device | |
| CN109376268A (en) | Video classification methods, device, electronic equipment and computer readable storage medium | |
| CN108921323A (en) | Method and apparatus for generating information | |
| CN109388548A (en) | Method and apparatus for generating information | |
| CN109784352A (en) | A kind of method and apparatus for assessing disaggregated model | |
| CN108959087A (en) | test method and device | |
| CN110084317A (en) | The method and apparatus of image for identification | |
| CN110473042B (en) | Method and device for acquiring information | |
| CN107451785A (en) | Method and apparatus for output information | |
| CN107517251A (en) | Information-pushing method and device | |
| CN108182472A (en) | For generating the method and apparatus of information | |
| CN108171275A (en) | For identifying the method and apparatus of flowers | |
| CN109961328A (en) | The method and apparatus for determining order cooling off period | |
| CN109684624A (en) | A kind of method and apparatus in automatic identification Order Address road area | |
| CN109614327A (en) | Method and apparatus for output information |
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 |