CN109446432A - A kind of information recommendation method and device - Google Patents
A kind of information recommendation method and device Download PDFInfo
- Publication number
- CN109446432A CN109446432A CN201811547986.9A CN201811547986A CN109446432A CN 109446432 A CN109446432 A CN 109446432A CN 201811547986 A CN201811547986 A CN 201811547986A CN 109446432 A CN109446432 A CN 109446432A
- Authority
- CN
- China
- Prior art keywords
- user
- information
- blog article
- prediction model
- recommendation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0631—Item recommendations
Landscapes
- Business, Economics & Management (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
This application discloses a kind of information-pushing methods, to solve using existing information recommendation technology, when carrying out information recommendation for specific user, and no normal direction potential user progress information recommendation, and the problem for causing information recommendation effect poor.Method includes: to receive real-time blog article publication data flow, and the account information of each blog article publication user is obtained from blog article publication data flow;Issue user for each blog article, the account information of user issued according to current blog article, based on the prediction model of pre-generated recommendation information, determine current blog article publication user and the recommendation information whether matching;When the current blog article publication user of determination matches with the recommendation information, Xiang Dangqian blog article issues user and pushes the recommendation information.Disclosed herein as well is a kind of information push-delivery apparatus.
Description
Technical field
This application involves field of computer technology more particularly to a kind of information recommendation methods and device.
Background technique
With the continuous development of Internet technology, the activity that people can carry out on the internet is also more and more abundant, because
And by internet channels to user carry out information recommendation become increasingly prevalent, for example, can by internet channels to
Family advertisement information, etc..
Generally, when carrying out information recommendation, information to be recommended often has certain specific aim, i.e., for information recommendation side
For certain class specific crowd (such as white collar, student, businessman or female group etc.), which can generate relatively good push away
Effect is recommended, recommendation effect can refer to whether recommendation information has an impact the user for receiving the information, for example, user is connecing
After receiving recommendation information, it may be checked by clicking the object that the recommendation information recommends recommendation message, Huo Zhehui
Access to recommendation information sender website, etc..
Currently, can mainly use following two mode when carrying out information recommendation:
Mode 1: information push is carried out to full platform user;
In order to enable more users to obtain the information of recommendation, information recommendation side often to the use of full platform per family into
Row information is recommended.
For example, by taking electric business website carries out advertisement recommendation as an example, it is assumed that it is expected the advertisement that a skirt is pushed to female user,
Based on the prior art, in order to reach the purpose, electric business website can be to all with sending the advertisement per family of the website.Due to the net
The user that stands simultaneously is not all female user, therefore when carrying out advertisement pushing, will certainly expend a large amount of resource of Website server
To realize that the other users outside female user carry out advertisement pushing;And male user is as receive this pushed information
And it is interfered.It can be seen that this kind of information push mode effect is very poor.
Mode 2: recommended just for satisfactory specific crowd;
In order to guarantee that specific aim and the information recommendation of recommendation information can generate preferable recommendation effect, letter is being carried out
When breath is recommended, often user is filtered as needed, and then guarantees only to carry out the user group for meeting particular requirement
Recommend.
For example, still by taking the advertisement that electric business website carries out skirt is recommended as an example, according to the recommendation side of mode 2 in the prior art
Formula, electric business website can select the user for buying skirt in all users of the website first, and then subsequent to those purchases
The user for buying skirt carries out advertisement recommendation.It is obvious that this kind of way of recommendation is due to carrying out user before carrying out information recommendation
Screening, thus largely can be by the potential user of purchaser record does not filter out before, to greatly affected letter
Cease the recommendation effect recommended.
It can be seen that needing a kind of information recommendation method with strong points and that potential user can be taken into account at present.
Summary of the invention
The embodiment of the present application provides a kind of information recommendation method, to solve using existing information recommendation technology, in needle
When carrying out information recommendation to specific user, no normal direction potential user carries out information recommendation, and causes information recommendation effect poor
Problem.
The embodiment of the present application also provides a kind of information recommending apparatus, uses existing information recommendation technology to solve,
When carrying out information recommendation for specific user, no normal direction potential user carries out information recommendation, and causes information recommendation effect poor
The problem of.
The embodiment of the present application adopts the following technical solutions:
A kind of information recommendation method, comprising:
Real-time blog article publication data flow is received, the account of each blog article publication user is obtained from blog article publication data flow
Number information;
User is issued for each blog article, the account information of user is issued according to current blog article, based on pre-generated
The prediction model of recommendation information, determine current blog article publication user and the recommendation information whether matching;
When the current blog article publication user of determination matches with the recommendation information, Xiang Dangqian blog article is issued described in user's push
Recommendation information.
A kind of information recommending apparatus, comprising:
Information acquisition unit is obtained from blog article publication data flow for receiving real-time blog article publication data flow
The account information of each blog article publication user;
Matching unit issues the account information of user, base according to current blog article for issuing user for each blog article
In the prediction model of pre-generated recommendation information, determine current blog article publication user and the recommendation information whether matching;
Information recommendation unit, for being won to current when the current blog article publication user of determination matches with the recommendation information
Text publication user pushes the recommendation information.
The embodiment of the present application use at least one above-mentioned technical solution can reach it is following the utility model has the advantages that
It, can be first according in real time before carrying out information push using information recommendation method provided by the embodiments of the present application
The blog article publication data flow received obtains the account information of each blog article publication user, and root from blog article publication data flow
According to those account informations, prediction model is issued based on pre-generated each blog article, current blog article publication user is determined respectively and pushes away
Recommend information whether matching just can be to be recommended only when determination current blog article publication user matches with the recommendation information
User pushes the recommendation information.By pre-generated prediction model, treats recommended user and screen, to predict wait push away
After recommending in user to the interested user of the recommendation information, and the matching user predicted to those is oriented information push,
To guarantee that recommendation information accurately pushes to the user in the presence of corresponding demand (or potential demand) on the one hand, to keep away
Exempt from that there is no the interference of corresponding demand user to other;Still further aspect can only be existed to determining to avoid using the prior art
The user of demand carries out information push, and ignores other potential users, and then the problem for causing information push effect poor.
Detailed description of the invention
The drawings described herein are used to provide a further understanding of the present application, constitutes part of this application, this Shen
Illustrative embodiments and their description please are not constituted an undue limitation on the present application for explaining the application.In the accompanying drawings:
Fig. 1 is a kind of idiographic flow schematic diagram of information-pushing method provided by the embodiments of the present application;
Fig. 2 is a kind of concrete structure schematic diagram of information push-delivery apparatus provided by the embodiments of the present application.
Specific embodiment
To keep the purposes, technical schemes and advantages of the application clearer, below in conjunction with the application specific embodiment and
Technical scheme is clearly and completely described in corresponding attached drawing.Obviously, described embodiment is only the application one
Section Example, instead of all the embodiments.Based on the embodiment in the application, those of ordinary skill in the art are not doing
Every other embodiment obtained under the premise of creative work out, shall fall in the protection scope of this application.
Below in conjunction with attached drawing, the technical scheme provided by various embodiments of the present application will be described in detail.
A kind of information-pushing method provided by the embodiments of the present application uses existing information recommendation technology to solve,
When carrying out information recommendation for specific user, no normal direction potential user carries out information recommendation, and causes information recommendation effect poor
The problem of.
Page generation method provided by the embodiment of the present application, the executing subject of this method, can be server, for example,
Advertisement Server, Information Push Server and micro blog server etc.;Furthermore the executing subject of this method is also possible in front end
The advertising service platform that is directly used for user, microblogging application program (Application, APP) or other with information
The application program of push function.The executing subject does not constitute the restriction to the application, and for ease of description, the present invention is real
Example is applied to be illustrated so that executing subject is microblogging service platform as an example.
The specific implementation flow schematic diagram of this method is as shown in Figure 1, mainly include the following steps:
Step 11, real-time blog article publication data flow is received, each blog article publication is obtained from blog article publication data flow
The account information of user;
Wherein, the blog article publication data flow refers to the data for user's transmission microblogging that micro blog server real-time reception arrives,
Include the content-data for every microblogging that current server real-time reception arrives in the blog article publication data flow, and issues this
The related data of the user of microblogging.And wherein, blog article publication user refers to that micro blog server can carry out the user of advertisement pushing,
In the embodiment of the present application, which can often refer to the user of the sending microblogging on current microblogging.
And in the embodiment of the present application, the account information that blog article issues user mainly may include following two:
1, user information;
User information mainly may include user identifier, user's gender, age of user, number of fans, attention number and account
The information such as grade.
Wherein, user identifier can refer to the account id that the user registers on microblogging;User's gender and age of user
It is user in register account number, voluntarily inputs and save information on the server;And number of fans, attention number and account grade
Refer to during user is using the account, the data that server is recorded and counted.
2, user behavior data;
The user behavior data mainly may include: to log in the frequency, send the blog article frequency and send blog article sum.Its
In, log in the frequency and send the blog article frequency be count to obtain in fixed time period, for example, can with one day for when
Between the period, to user the intraday login frequency and send the microblogging frequency count.
Under normal conditions, the system that most of website has dedicated for recording user behavior data, user is in microblogging APP
Or the various actions data executed on webpage can be uploaded to the system by the modes such as getting ready, which is receiving these
After behavioral data, it will usually be associated the account of those user behavior datas and the user in the form of journal file and deposit
Storage.
Then in the embodiment of the present application, micro blog server can obtain the behavioral data of user from above-mentioned journal file.
Step 12, user is issued for each blog article, is issued according to by executing the current blog article that step 11 is got
The account information of user determines that current blog article publication user pushes away with described based on the prediction model of pre-generated recommendation information
Recommend whether information matches;
In one embodiment, prediction model provided in the embodiment of the present application can be logic-based regression algorithm
What training generated.For the ease of the understanding to this programme, logistic regression algorithm is illustrated initially below.
Logistic regression algorithm (Logistic Regression algorithm, abbreviation LR algorithm), also referred to as Logit model, be from
Dissipate one of back-and-forth method model, belong to multiple variables analysis scope, commonly used in classification judgement or event occurrence rate it is pre-
It surveys.The algorithm is currently widely used a kind of machine learning calculation in search field, advertisement promotion field and recommendation application field
Method.
The process of Logit model is established, is just to solve for one group of weight W in fact0、W1、...、WmProcess, find out above-mentioned power
After value, in practical application Logit model, according to above-mentioned weight and real data, (that is: dimension enters each of parameter for M
Characteristic value X1-Xm) it is weighted summation, corresponding Z value is found out, shown in following following formula [1]:
Z=W0+W1×X1+......+Wm×Xm [1]
Z in above-mentioned formula [1] also often writes function g (x), then according to sigmoid letter shown in following formula [2]
Several forms finds out the value of variable P:
The codomain of variable P is [0,1], is represented in the case where currently entering parameter, (Y=1) occurs for certain event
Probability.A discrimination threshold is specified if it is calculated probability value, which can serve as classifier use.
It can be seen that the key of model is established, each coefficient W being just to solve in above-mentioned formula 10、W1、...、WmValue.
The solution procedure of above-mentioned coefficient is briefly described below.
Assuming that there is n independent training sample { (x in training set data1, y1), (x2, y2) ..., (xn, yn), y=0,
1}.X thereiniIt is above-mentioned dimension is that M enters parameter, each sample (x observedi, yi) occur the following formula of probability
[3] shown in:
P(yi, xi)=P (yi=1 | xi)y i(1-P(yi=1 | xi))1-y i [3]
Due to each sample be it is independent, n sample occur probability be exactly the probability multiplication respectively occurred, so as to
To obtain the likelihood function that n independent samples occur in entire training set data, then solve so that in following formula [4]
Each coefficient W when likelihood function value maximum0、W1、...、WmValue.
W in above-mentioned formula [4] is exactly to be solved comprising W0、W1、...、WmM dimensional vector, due in training set data
The x of each sampleiAnd yiAll it is known, therefore above-mentioned formula [4] can be directly substituted into, to obtains one group of non-linear expression
Formula solves above-mentioned likelihood function value then using mathematical methods such as gradient descent methods either newton-La Feisen iterative method
Each W when maximum0、W1、...、WmThe value of parameter, has obtained the value of this group of parameter, and the establishment process of model just completes.By
Establishment process in Logit model and the derivation algorithm that is related to, are all the prior arts of comparative maturity, therefore do not ask it
Solution details is further described.
Based on logistic regression algorithm described above, hereafter will construct for predicting whether microblog users are potential
Advertising service purchase user prediction model for, specific introduce is how to construct prediction model in this programme.
In one embodiment, this programme specifically can generate in accordance with the following steps prediction model:
Sub-step a: certain amount of and the matched sample of users of the recommendation information account information is obtained as positive sample
Notebook data;
In order to generation prediction model can be trained, it is necessary first to collect for training the prediction model to enter to join sample number
According to since the prediction model that this programme training generates is mainly used for predicting whether microblog users are potential advertising service purchase
User, thus can will buy in this programme advertising service user related data as training prediction model mistake
Sample data used in journey.
And since in actual use, the number of fans of user, sends microblogging frequency, microblogging item number, age at attention number
And whether the features such as gender may can buy advertising service to the user and cause a degree of influence, thus in one kind
In embodiment, this programme can by user identifier related to user, user's gender, age of user, number of fans, attention number with
And account grade, the frequency is logged in, the microblogging frequency is sent and sends those features such as microblogging sum as the training prediction model
Enter to join sample data.
In addition, in order to improve the accuracy for the prediction model prediction result that training obtains as much as possible, in a kind of embodiment party
In formula, this programme can collect the user data that can be used as training sample as much as possible.In the embodiment of the present application, often may be used
Using collect the related data of the user that at least 10,000 were bought advertising service as during training prediction model it is used just
Sample data.
Sub-step b: certain amount of and the unmatched sample of users of the recommendation information account information is obtained as negative
Sample data;
In the present solution, in addition to need according to above-mentioned sub-step a obtain bought advertising service user correlation
Data as positive sample data used in during training prediction model outside, it is also necessary to advertising service was not bought in acquisition
The related data of user is as used negative sample data during training prediction model.Equally, in the present solution, can incite somebody to action
With do not bought advertising service user user identifier, user's gender, age of user, number of fans, attention number and account grade,
The frequency is logged in, the blog article frequency is sent and sends negative sample number of those features such as blog article sum as the training prediction model
According to.
And the accuracy in order to guarantee trained obtained prediction model, in one embodiment, when training prediction model
The quantity of used positive sample data and negative sample data is generally consistent.
Sub-step c: according to the sample data got by executing sub-step a and sub-step b, training sample is generated
Collection;
In one embodiment, training sample set can be generated in the following way:
It will be by executing user identifier, user's gender, user in the sample data that sub-step a and sub-step b is obtained
Age, number of fans, attention number and account grade log in the frequency, send the microblogging frequency and send the features such as microblogging sum and make
For the x in training samplei, and by the y in wherein positive sample datai=1, and the y in negative sample datai=0, using above-mentioned side
The available sufficient amount of sample data of formula, and each sample data has determining xiAnd yi, that is, form training prediction model
Used training sample set.
Sub-step d: using logistic regression algorithm Logistic Regression, respectively with the positive sample data and
Whether the negative sample data match with the recommendation information as output as input data, using the sample of users, establish
The prediction model, the prediction model be used for according to the account information of user predict user to the recommendation information whether
Match.
The training set data that will have been generated substitutes into formula described above [4], and using gradient descent method or
It is the mathematical methods such as newton-La Feisen iterative method, solves each W when sening as an envoy to likelihood function value maximum0、W1、...、WmParameter
Value to get arrived in this programme for predict microblog users whether be potential advertising service purchase user prediction model.
In addition there is also the need to explanations, in order to guarantee that the result that the prediction model obtained using training is predicted is quasi-
True rate in one embodiment, can also be to using verifying number after obtaining prediction model by executing above-mentioned steps training
It is verified according to the prediction model.Specifically, the embodiment of the present application can with the following method to the prediction model of generation into
Row verifying: inputting the prediction model using corresponding verify data, obtains output result;According to the output as a result, calculating
The predictablity rate of the prediction model;Judge whether the predictablity rate is greater than preset accuracy rate threshold value;And root
It is judged that result determines the accuracy of the prediction model, to determine whether the prediction model can be used.
Wherein it is possible to using the related data of the fixed user for buying advertising service as verify data, by above-mentioned
After training process is it is found that input the prediction model for those verify datas, the result that normally obtains should be 1, so pass through by
Those verify datas are input in the prediction model, and whether to judge to export result, to be 1 calculate the accuracy rate of the prediction model.
In one embodiment, it is 80% that accuracy rate threshold value, which can be set, i.e., is being carried out using 100 verify datas to prediction model
After verifying, as long as there is 80 output results to meet expected results, then it is assumed that the prediction model accuracy rate is qualified, can be used.
And when the accuracy rate for buying model by verifying discovery prediction is less than preset accuracy rate threshold value, then illustrate the prediction
Model is undesirable, and then can be by adjusting establishing data used in prediction model, re-establish corresponding prediction mould
Type, until the predictablity rate of prediction model is greater than preset accuracy rate threshold value.
Prediction model is being generated by executing above-mentioned steps, and after verifying and determining that the prediction model is available, microblogging clothes
The prediction knot that the account information of user to be recommended can be inputted in the prediction model, and then be exported according to prediction model by business device
Fruit, the determining user to be recommended to match with recommendation information.
In addition, in the embodiment of the present application, the user due to having bought advertising service is clearly advertisement recommendation
Targeted user is ceased, thus before treating recommended user according to prediction model and screening, first choice can be from use to be recommended
The user for having bought advertising service is screened out in family, so that the number of users for needing to be screened by prediction model is reduced,
Improve the screening efficiency of user to be recommended.
After the prediction model for generating recommendation information through the above steps, it can will pass through and execute the account that step 11 be got
Number information inputs the prediction model as input, and with determination, whether current blog article publication user matches with recommendation information, specifically
Ground, method provided by the embodiments of the present application may include: using the prediction model of pre-generated recommendation information, with the user
Information and the user behavior data determine whether current blog article publication user matches with the recommendation information as input.
Step 13, when determining that the user to be recommended matches with the recommendation information by executing step 12, Xiang Suoshu
User to be recommended pushes the recommendation information.
I.e. after determining that user to be recommended is that user is bought in potential advertising service by executing step 13, micro blog server
The recommendation information of advertising service can be sent, to those users to prompt those users to buy advertising service.
What needs to be explained here is that in order to avoid frequently to those users transmission recommendation information and to the normal of those users
Using impacting, in one embodiment, determining that user to be recommended is potential advertising service by executing step 12
After buying user, micro blog server can also further determine that preceding once between the time of user's advertisement service recommendation information
Every whether being greater than preset time threshold (such as 7 days), only when the time interval apart from last time advertisement service recommendation information
After preset time threshold, otherwise it can just be pushed to the recommendation information of user's advertisement service without information, tool
Body, method provided by the embodiments of the present application may include: to determine to current blog article publication user to push a upper recommendation information
At the time of and current time time interval;When judging that the time interval is greater than prefixed time interval, Xiang Dangqian blog article hair
Cloth user pushes the recommendation information.
It, can be first according in real time before carrying out information push using information recommendation method provided by the embodiments of the present application
The blog article publication data flow received obtains the account information of each blog article publication user, and root from blog article publication data flow
According to those account informations, prediction model is issued based on pre-generated each blog article, current blog article publication user is determined respectively and pushes away
Recommend information whether matching just can be to be recommended only when determination current blog article publication user matches with the recommendation information
User pushes the recommendation information.By pre-generated prediction model, treats recommended user and screen, to predict wait push away
After recommending in user to the interested user of the recommendation information, and the matching user predicted to those is oriented information push,
To guarantee that recommendation information accurately pushes to the user in the presence of corresponding demand (or potential demand) on the one hand, to keep away
Exempt from that there is no the interference of corresponding demand user to other;Still further aspect can only be existed to determining to avoid using the prior art
The user of demand carries out information push, and ignores other potential users, and then the problem for causing information push effect poor.
In addition, a kind of information push-delivery apparatus provided by the embodiments of the present application, to solve using existing information recommendation skill
Art, when carrying out information recommendation for specific user, no normal direction potential user carries out information recommendation, and leads to information recommendation effect
Poor problem.The concrete structure schematic diagram of the device is as shown in Figure 2, comprising: information acquisition unit 21, matching unit 22 and
Information recommendation unit 23.
Wherein, data flow is issued from the blog article for receiving real-time blog article publication data flow in information acquisition unit 21
The middle account information for obtaining each blog article publication user;
Matching unit 22 obtains each for receiving real-time blog article publication data flow from blog article publication data flow
Blog article issues the account information of user, based on the prediction model of pre-generated recommendation information, determines current blog article publication user
With recommendation information whether matching;
Information recommendation unit 23 is used for when the current blog article publication user of determination matches with the recommendation information, Xiang Dangqian
Blog article issues user and pushes the recommendation information.
In one embodiment, the account information includes: user information and user behavior data;
The then matching unit 22, is specifically used for: using the prediction model of pre-generated recommendation information, with the user
Information and the user behavior data determine whether current blog article publication user matches with the recommendation information as input.
In one embodiment, the user information include: user identifier, user's gender, age of user, number of fans,
Attention number and account grade;The user behavior data includes: that the login frequency, the transmission blog article frequency and transmission blog article are total
Number.
In one embodiment, the information recommendation unit 23, is specifically used for: determining and pushes away to current blog article publication user
At the time of serving a recommendation information and the time interval at current time;When judge the time interval be greater than prefixed time interval
When, Xiang Dangqian blog article issues user and pushes the recommendation information.
In one embodiment, described device further includes prediction model generation unit, is specifically used for: obtaining specific quantity
, with the account information of the matched sample of users of the recommendation information as positive sample data;Obtain it is certain amount of, with it is described
The account information of the unmatched sample of users of recommendation information is as negative sample data;Using logistic regression algorithm Logistic
Regression, respectively using the positive sample data and the negative sample data as input data, with the sample of users
Whether match with the recommendation information as output, establish the prediction model, the prediction model is used for the account according to user
Whether number information prediction user matches the recommendation information.
In one embodiment, the prediction model generation unit, is also used to: inputting institute using corresponding verify data
Prediction model is stated, output result is obtained;According to the output as a result, calculating the predictablity rate of the prediction model;Judge institute
State whether predictablity rate is greater than preset accuracy rate threshold value;When the judgment result is yes, it is determined that the prediction model
It can use;When the judgment result is no, then adjustment establishes data used in prediction model, re-establishes corresponding prediction model,
Until the predictablity rate of prediction model is greater than preset accuracy rate threshold value.
It, can be first according in real time before carrying out information push using information recommending apparatus provided by the embodiments of the present application
The blog article publication data flow received obtains the account information of each blog article publication user, and root from blog article publication data flow
According to those account informations, prediction model is issued based on pre-generated each blog article, current blog article publication user is determined respectively and pushes away
Recommend information whether matching just can be to be recommended only when determination current blog article publication user matches with the recommendation information
User pushes the recommendation information.By pre-generated prediction model, treats recommended user and screen, to predict wait push away
After recommending in user to the interested user of the recommendation information, and the matching user predicted to those is oriented information push,
To guarantee that recommendation information accurately pushes to the user in the presence of corresponding demand (or potential demand) on the one hand, to keep away
Exempt from that there is no the interference of corresponding demand user to other;Still further aspect can only be existed to determining to avoid using the prior art
The user of demand carries out information push, and ignores other potential users, and then the problem for causing information push effect poor.
It should be understood by those skilled in the art that, the embodiment of the present invention can provide as method, system or computer program
Product.Therefore, complete hardware embodiment, complete software embodiment or reality combining software and hardware aspects can be used in the present invention
Apply the form of example.Moreover, it wherein includes the computer of computer usable program code that the present invention, which can be used in one or more,
The computer program implemented in usable storage medium (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.) produces
The form of product.
The present invention be referring to according to the method for the embodiment of the present invention, the process of equipment (system) and computer program product
Figure and/or block diagram describe.It should be understood that every one stream in flowchart and/or the block diagram can be realized by computer program instructions
The combination of process and/or box in journey and/or box and flowchart and/or the block diagram.It can provide these computer programs
Instruct the processor of general purpose computer, special purpose computer, Embedded Processor or other programmable data processing devices to produce
A raw machine, so that being generated by the instruction that computer or the processor of other programmable data processing devices execute for real
The device for the function of being specified in present one or more flows of the flowchart and/or one or more blocks of the block diagram.
These computer program instructions, which may also be stored in, is able to guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works, so that it includes referring to that instruction stored in the computer readable memory, which generates,
Enable the manufacture of device, the command device realize in one box of one or more flows of the flowchart and/or block diagram or
The function of being specified in multiple boxes.
These computer program instructions also can be loaded onto a computer or other programmable data processing device, so that counting
Series of operation steps are executed on calculation machine or other programmable devices to generate computer implemented processing, thus in computer or
The instruction executed on other programmable devices is provided for realizing in one or more flows of the flowchart and/or block diagram one
The step of function of being specified in a box or multiple boxes.
In a typical configuration, calculating equipment includes one or more processors (CPU), input/output interface, net
Network interface and memory.
Memory may include the non-volatile memory in computer-readable medium, random access memory (RAM) and/or
The forms such as Nonvolatile memory, such as read-only memory (ROM) or flash memory (flash RAM).Memory is computer-readable medium
Example.
Computer-readable medium includes permanent and non-permanent, removable and non-removable media can be by any method
Or technology come realize information store.Information can be computer readable instructions, data structure, the module of program or other data.
The example of the storage medium of computer includes, but are not limited to phase change memory (PRAM), static random access memory (SRAM), moves
State random access memory (DRAM), other kinds of random access memory (RAM), read-only memory (ROM), electric erasable
Programmable read only memory (EEPROM), flash memory or other memory techniques, read-only disc read only memory (CD-ROM) (CD-ROM),
Digital versatile disc (DVD) or other optical storage, magnetic cassettes, tape magnetic disk storage or other magnetic storage devices
Or any other non-transmission medium, can be used for storage can be accessed by a computing device information.As defined in this article, it calculates
Machine readable medium does not include temporary computer readable media (transitory media), such as the data-signal and carrier wave of modulation.
It should also be noted that, the terms "include", "comprise" or its any other variant are intended to nonexcludability
It include so that the process, method, commodity or the equipment that include a series of elements not only include those elements, but also to wrap
Include other elements that are not explicitly listed, or further include for this process, method, commodity or equipment intrinsic want
Element.In the absence of more restrictions, the element limited by sentence " including one ... ", it is not excluded that including described
There is also other identical elements in the process, method of element, commodity or equipment.
It will be understood by those skilled in the art that embodiments herein can provide as method, system or computer program product.
Therefore, complete hardware embodiment, complete software embodiment or embodiment combining software and hardware aspects can be used in the application
Form.It is deposited moreover, the application can be used to can be used in the computer that one or more wherein includes computer usable program code
The shape for the computer program product implemented on storage media (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.)
Formula.
The above description is only an example of the present application, is not intended to limit this application.For those skilled in the art
For, various changes and changes are possible in this application.All any modifications made within the spirit and principles of the present application are equal
Replacement, improvement etc., should be included within the scope of the claims of this application.
Claims (11)
1. a kind of information-pushing method characterized by comprising
Real-time blog article publication data flow is received, the account letter of each blog article publication user is obtained from blog article publication data flow
Breath;
User is issued for each blog article, the account information of user is issued according to current blog article, based on pre-generated recommendation
The prediction model of information, determines whether current blog article publication user matches with the recommendation information;
When the current blog article publication user of determination matches with the recommendation information, Xiang Dangqian blog article issues user and pushes the recommendation
Information.
2. the method according to claim 1, wherein the account information includes: user information and user's row
For data;
Then according to the account information, based on the prediction model of pre-generated recommendation information, determine the user to be recommended with
Whether recommendation information matches, and specifically includes:
Using the prediction model of pre-generated recommendation information, using the user information and the user behavior data as defeated
Enter, determines whether current blog article publication user matches with the recommendation information.
3. according to the method described in claim 2, it is characterized in that, the user information include: user identifier, user's gender,
Age of user, number of fans, attention number and account grade;
The user behavior data includes: to log in the frequency, send the blog article frequency and send blog article sum.
4. the method according to claim 1, wherein when determining current blog article publication user and the recommendation information
When matching, Xiang Dangqian blog article issues user and pushes the recommendation information, specifically includes:
It determines at the time of pushing a upper recommendation information to current blog article publication user and the time interval at current time;
When judging that the time interval is greater than prefixed time interval, Xiang Dangqian blog article issues user and pushes the recommendation information.
5. the method according to claim 1, wherein the prediction model is generated using following manner:
Certain amount of and the matched sample of users of the recommendation information account information is obtained as positive sample data;
Certain amount of and the unmatched sample of users of the recommendation information account information is obtained as negative sample data;
Using logistic regression algorithm Logistic Regression, respectively with the positive sample data and the negative sample number
According to as input data, whether is matched with the recommendation information as output using the sample of users, establishes the prediction model,
The prediction model is used to predict whether user matches the recommendation information according to the account information of user.
6. according to the method described in claim 5, it is characterized in that, using logistic regression algorithm Logistic Regression,
Respectively using the positive sample data and the negative sample data as input data, with the sample of users and the recommendation
Whether breath matches as output, after establishing the prediction model, the method also includes:
The prediction model is inputted using corresponding verify data, obtains output result;
According to the output as a result, calculating the predictablity rate of the prediction model;
Judge whether the predictablity rate is greater than preset accuracy rate threshold value;
When the judgment result is yes, it is determined that the prediction model is available;When the judgment result is no, then foundation prediction mould is adjusted
Data used in type re-establish corresponding prediction model, until the predictablity rate of prediction model is greater than preset
Accuracy rate threshold value.
7. a kind of information push-delivery apparatus characterized by comprising
Information acquisition unit obtains each rich for receiving real-time blog article publication data flow from blog article publication data flow
The account information of text publication user;
Matching unit issues the account information of user according to current blog article, based on pre- for issuing user for each blog article
The prediction model of the recommendation information first generated, determine current blog article publication user and the recommendation information whether matching;
Information recommendation unit, for when the current blog article publication user of determination matches with the recommendation information, Xiang Dangqian blog article to be sent out
Cloth user pushes the recommendation information.
8. device according to claim 7, which is characterized in that the account information includes: user information and user's row
For data;
The then matching unit, is specifically used for:
Using the prediction model of pre-generated recommendation information, using the user information and the user behavior data as defeated
Enter, determines whether current blog article publication user matches with the recommendation information.
9. device according to claim 7, which is characterized in that information recommendation unit is specifically used for:
It determines at the time of pushing a upper recommendation information to current blog article publication user and the time interval at current time;
When judging that the time interval is greater than prefixed time interval, Xiang Dangqian blog article issues user and pushes the recommendation information.
10. device according to claim 7, which is characterized in that further include prediction model generation unit, be used for:
Certain amount of and the matched sample of users of the recommendation information account information is obtained as positive sample data;
Certain amount of and the unmatched sample of users of the recommendation information account information is obtained as negative sample data;
Using logistic regression algorithm Logistic Regression, respectively with the positive sample data and the negative sample number
According to as input data, whether is matched with the recommendation information as output using the sample of users, establishes the prediction model,
The prediction model is used to predict whether user matches the recommendation information according to the account information of user.
11. device according to claim 10, which is characterized in that the prediction model generation unit is also used to:
The prediction model is inputted using corresponding verify data, obtains output result;
According to the output as a result, calculating the predictablity rate of the prediction model;
Judge whether the predictablity rate is greater than preset accuracy rate threshold value;
When the judgment result is yes, it is determined that the prediction model is available;When the judgment result is no, then foundation prediction mould is adjusted
Data used in type re-establish corresponding prediction model, until the predictablity rate of prediction model is greater than preset
Accuracy rate threshold value.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811547986.9A CN109446432A (en) | 2018-12-17 | 2018-12-17 | A kind of information recommendation method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811547986.9A CN109446432A (en) | 2018-12-17 | 2018-12-17 | A kind of information recommendation method and device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109446432A true CN109446432A (en) | 2019-03-08 |
Family
ID=65559275
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811547986.9A Pending CN109446432A (en) | 2018-12-17 | 2018-12-17 | A kind of information recommendation method and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109446432A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110399550A (en) * | 2019-03-14 | 2019-11-01 | 腾讯科技(深圳)有限公司 | A kind of information recommendation method and device |
CN111859238A (en) * | 2020-07-27 | 2020-10-30 | 平安科技(深圳)有限公司 | Method and device for predicting data change frequency based on model and computer equipment |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20080195715A1 (en) * | 2007-02-14 | 2008-08-14 | Tu-Hsin Tsai | System and method for asynchronous exchanging electronic messages |
CN103116589A (en) * | 2011-11-17 | 2013-05-22 | 腾讯科技(深圳)有限公司 | Method and device of sending recommendation information |
CN104281622A (en) * | 2013-07-11 | 2015-01-14 | 华为技术有限公司 | Information recommending method and information recommending device in social media |
CN105447730A (en) * | 2015-12-25 | 2016-03-30 | 腾讯科技(深圳)有限公司 | Target user orientation method and device |
CN105469263A (en) * | 2014-09-24 | 2016-04-06 | 阿里巴巴集团控股有限公司 | Commodity recommendation method and device |
CN107871244A (en) * | 2016-09-28 | 2018-04-03 | 腾讯科技(深圳)有限公司 | The detection method and device of a kind of advertising results |
-
2018
- 2018-12-17 CN CN201811547986.9A patent/CN109446432A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20080195715A1 (en) * | 2007-02-14 | 2008-08-14 | Tu-Hsin Tsai | System and method for asynchronous exchanging electronic messages |
CN103116589A (en) * | 2011-11-17 | 2013-05-22 | 腾讯科技(深圳)有限公司 | Method and device of sending recommendation information |
CN104281622A (en) * | 2013-07-11 | 2015-01-14 | 华为技术有限公司 | Information recommending method and information recommending device in social media |
CN105469263A (en) * | 2014-09-24 | 2016-04-06 | 阿里巴巴集团控股有限公司 | Commodity recommendation method and device |
CN105447730A (en) * | 2015-12-25 | 2016-03-30 | 腾讯科技(深圳)有限公司 | Target user orientation method and device |
CN107871244A (en) * | 2016-09-28 | 2018-04-03 | 腾讯科技(深圳)有限公司 | The detection method and device of a kind of advertising results |
Non-Patent Citations (2)
Title |
---|
WOUTER WEERKAMP 等: "Credibility-inspired ranking for blog post retrieval", 《INFORMATION RETRIEVAL》 * |
高俊波 等: "基于文本内容分析的微博广告过滤模型研究", 《计算机工程》 * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110399550A (en) * | 2019-03-14 | 2019-11-01 | 腾讯科技(深圳)有限公司 | A kind of information recommendation method and device |
CN110399550B (en) * | 2019-03-14 | 2023-08-15 | 腾讯科技(深圳)有限公司 | Information recommendation method and device |
CN111859238A (en) * | 2020-07-27 | 2020-10-30 | 平安科技(深圳)有限公司 | Method and device for predicting data change frequency based on model and computer equipment |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US11659050B2 (en) | Discovering signature of electronic social networks | |
CN106126558B (en) | A kind of public sentiment monitoring method and device | |
CN105224606B (en) | A kind of processing method and processing device of user identifier | |
CN110442790A (en) | Recommend method, apparatus, server and the storage medium of multi-medium data | |
CN110503206A (en) | A kind of prediction model update method, device, equipment and readable medium | |
US20160132904A1 (en) | Influence score of a brand | |
CN109561052B (en) | Method and device for detecting abnormal flow of website | |
CN110428298A (en) | A kind of shop recommended method, device and equipment | |
WO2018130201A1 (en) | Method for determining associated account, server and storage medium | |
KR102451099B1 (en) | Purchsase inducement system and method through inferencing user's needs based on artificial intelligence | |
CN108777701A (en) | A kind of method and device of determining receiver | |
CN112559896B (en) | Information recommendation method, device, equipment and computer readable storage medium | |
CN110880127B (en) | Consumption level prediction method and device, electronic equipment and storage medium | |
CN108632311A (en) | Information-pushing method and device | |
CN105590240A (en) | Discrete calculating method of brand advertisement effect optimization | |
CN111275503A (en) | Data processing method and device for acquiring lost user recall success rate | |
CN105608604A (en) | Continuous calculation method of brand advertisement effectiveness optimization | |
CN107436914A (en) | Recommend method and device | |
Jin et al. | Task-oriented web user modeling for recommendation | |
CN109446432A (en) | A kind of information recommendation method and device | |
KR20220137943A (en) | pattern-based classification | |
CN109710410A (en) | A kind of internet information resource distribution method and device | |
US20230316106A1 (en) | Method and apparatus for training content recommendation model, device, and storage medium | |
CN111310453A (en) | User theme vectorization representation method and system based on deep learning | |
Wu et al. | A hybrid approach based on collaborative filtering to recommending mobile apps |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190308 |