CN109255070A - Recommendation information processing method, device, computer equipment and storage medium - Google Patents
Recommendation information processing method, device, computer equipment and storage medium Download PDFInfo
- Publication number
- CN109255070A CN109255070A CN201810865257.1A CN201810865257A CN109255070A CN 109255070 A CN109255070 A CN 109255070A CN 201810865257 A CN201810865257 A CN 201810865257A CN 109255070 A CN109255070 A CN 109255070A
- Authority
- CN
- China
- Prior art keywords
- recommendation information
- vector
- recommendation
- candidate
- information
- 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
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
- G06V10/464—Salient features, e.g. scale invariant feature transforms [SIFT] using a plurality of salient features, e.g. bag-of-words [BoW] representations
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The application proposes a kind of recommendation information processing method, device, computer equipment and storage medium, wherein method comprises determining that the corresponding primary vector of i-th of recommendation information in recommendation list, wherein i is positive integer;According to the primary vector, the corresponding secondary vector of i recommendation information before determining;According to the primary vector and the secondary vector, determine that Candidate Recommendation information concentrates the click probability of each Candidate Recommendation information;According to the click probability of each Candidate Recommendation information, is concentrated from the Candidate Recommendation information and choose i+1 recommendation information.Pass through this method, it realizes according to the relevance between recommendation information and recommendation information is ranked up, be conducive to improve the clicking rate of recommendation information in recommendation list, it solves and recommendation list is determined according to single marking result in the prior art, have ignored the low technical problem of relevance, the clicking rate of recommendation information between recommendation information.
Description
Technical field
This application involves Internet technical field more particularly to a kind of recommendation information processing methods, device, computer equipment
And storage medium.
Background technique
With the development of intelligent terminal and Internet technology, information recommendation has become the major way that user obtains information
One of.
Currently, being mostly to treat using scoring model and recommend the multiple of user when information recommendation system carries out information recommendation
Information is given a mark, and is then ranked up according to marking result to information to be recommended, the forward predetermined number of selected and sorted
Information forms recommendation list, and recommendation list is fed back to user.
However, the above- mentioned information way of recommendation, only accounts for the optimization of single recommendation information, and have ignored in recommendation list
Relevance between each recommendation information causes the clicking rate of each recommendation information in recommendation list lower.
Summary of the invention
The application is intended to solve at least some of the technical problems in related technologies.
For this purpose, the application's proposes a kind of recommendation information processing method, device, computer equipment and storage medium, it is used for
Solution determines recommendation list according to single marking result in the prior art, has ignored the relevance between recommendation information, recommends
The low technical problem of the clicking rate of information.
In order to achieve the above object, the application first aspect embodiment proposes a kind of recommendation information processing method, comprising:
Determine the corresponding primary vector of i-th of recommendation information in recommendation list, wherein i is positive integer;
According to the primary vector, the corresponding secondary vector of i recommendation information before determining;
According to the primary vector and the secondary vector, determine that Candidate Recommendation information concentrates each Candidate Recommendation information
Click probability;
According to the click probability of each Candidate Recommendation information, is concentrated from the Candidate Recommendation information and choose i+1 recommendation
Information.
The recommendation information processing method of the embodiment of the present application, by determining, i-th of recommendation information is corresponding in recommendation list
Primary vector, and the corresponding secondary vector of preceding i recommendation information is determined according to primary vector, and then according to primary vector and second
Vector determines that Candidate Recommendation information concentrates the click probability of each Candidate Recommendation information, according to the point of each Candidate Recommendation information
Probability is hit, is concentrated from Candidate Recommendation information and chooses i+1 recommendation information.As a result, by according to previous recommendation information to
Measure and be completed the vector of all recommendation informations of sequence, to determine current recommendation information, realize according to recommendation information it
Between relevance recommendation information is ranked up, be conducive to the clicking rate for improving recommendation information in recommendation list, improve use
Family experience.
In order to achieve the above object, the application second aspect embodiment proposes a kind of recommendation information processing unit, comprising:
First determining module, for determining the corresponding primary vector of i-th of recommendation information in recommendation list, wherein i is positive
Integer;
Second determining module, for determining the corresponding secondary vector of preceding i recommendation information according to the primary vector;
Probability determination module, for determining that Candidate Recommendation information is concentrated according to the primary vector and the secondary vector
The click probability of each Candidate Recommendation information;
Selecting module is concentrated from the Candidate Recommendation information and is selected for the click probability according to each Candidate Recommendation information
Take i+1 recommendation information.
The recommendation information processing unit of the embodiment of the present application, by determining, i-th of recommendation information is corresponding in recommendation list
Primary vector, and the corresponding secondary vector of preceding i recommendation information is determined according to primary vector, and then according to primary vector and second
Vector determines that Candidate Recommendation information concentrates the click probability of each Candidate Recommendation information, according to the point of each Candidate Recommendation information
Probability is hit, is concentrated from Candidate Recommendation information and chooses i+1 recommendation information.As a result, by according to previous recommendation information to
Measure and be completed the vector of all recommendation informations of sequence, to determine current recommendation information, realize according to recommendation information it
Between relevance recommendation information is ranked up, be conducive to the clicking rate for improving recommendation information in recommendation list, improve use
Family experience.
In order to achieve the above object, the application third aspect embodiment proposes a kind of computer equipment, comprising: processor and deposit
Reservoir;Wherein, the processor is held to run with described by reading the executable program code stored in the memory
The corresponding program of line program code, for realizing the recommendation information processing method as described in first aspect embodiment.
In order to achieve the above object, the application fourth aspect embodiment proposes a kind of non-transitory computer-readable storage medium
Matter is stored thereon with computer program, and the recommendation as described in first aspect embodiment is realized when which is executed by processor
Cease processing method.
In order to achieve the above object, the 5th aspect embodiment of the application proposes a kind of computer program product, when the calculating
The recommendation information processing method as described in first aspect embodiment is realized when instruction in machine program product is executed as processor.
The additional aspect of the application and advantage will be set forth in part in the description, and will partially become from the following description
It obtains obviously, or recognized by the practice of the application.
Detailed description of the invention
The application is above-mentioned and/or additional aspect and advantage will become from the following description of the accompanying drawings of embodiments
Obviously and it is readily appreciated that, in which:
Fig. 1 is a kind of flow diagram of recommendation information processing method provided by the embodiment of the present application;
Fig. 2 is the positional relationship exemplary diagram of each recommendation information in recommendation list;
Fig. 3 is the flow diagram of another kind recommendation information processing method provided by the embodiment of the present application;
Fig. 4 is the flow diagram of another recommendation information processing method provided by the embodiment of the present application;
Fig. 5 is the flow diagram of another recommendation information processing method provided by the embodiment of the present application;
Fig. 6 is the example architecture figure for realizing the recommendation information order models of recommendation information processing method of the application;
Fig. 7 is a kind of structural schematic diagram of recommendation information processing unit provided by the embodiment of the present application;
Fig. 8 is the structural schematic diagram of another kind recommendation information processing unit provided by the embodiment of the present application;
Fig. 9 is the structural schematic diagram of another recommendation information processing unit provided by the embodiment of the present application;
Figure 10 is the structural schematic diagram of another recommendation information processing unit provided by the embodiment of the present application;And
Figure 11 is the structural schematic diagram of computer equipment provided by the embodiment of the present application.
Specific embodiment
Embodiments herein is described below in detail, examples of the embodiments are shown in the accompanying drawings, wherein from beginning to end
Same or similar label indicates same or similar element or element with the same or similar functions.Below with reference to attached
The embodiment of figure description is exemplary, it is intended to for explaining the application, and should not be understood as the limitation to the application.
Below with reference to the accompanying drawings the recommendation information processing method, device, computer equipment and storage of the embodiment of the present application are described
Medium.
Fig. 1 is a kind of flow diagram of recommendation information processing method provided by the embodiment of the present application.
As shown in Figure 1, the recommendation information processing method may comprise steps of:
Step 101, the corresponding primary vector of i-th of recommendation information in recommendation list is determined, wherein i is positive integer.
Wherein, recommendation list is for showing the recommendation information that sequence is completed, when carrying out recommendation information prediction, every prediction
A recommendation information out just shows the recommendation information in recommendation list;I-th of recommendation information is newest exhibition in recommendation list
The recommendation information shown.
In the present embodiment, when predicting recommendation information, i-th of recommendation information corresponding first in recommendation list can be first determined
Vector.
It,, can be to Candidate Recommendation when determining corresponding vector for text information as a kind of possible implementation
The word for including in information is counted, and according to the unduplicated word for including in Candidate Recommendation information, constructs bag of words list.
For example, it is assumed that Candidate Recommendation information is as follows:
Text 1:My dog ate my homework.
Text 2:My cat ate the sandwich.
Text 3:A dolphin ate the homework.
Then according to above-mentioned Candidate Recommendation information, bag of words list can be constructed are as follows: [a ate cat dolphin dog
Homework my sandwich the], it altogether include 9 unduplicated words in the bag of words list.
In turn, it can be encoded based on one-hot, according to i-th of recommendation information and constructed bag of words list, determine the
The corresponding primary vector of i recommendation information.Wherein, the dimension of primary vector is consistent with the length of bag of words list.
Example still is exemplified as with above-mentioned, it is assumed that i-th of recommendation information is text 1, then primary vector corresponding with text 1 is [0
1 0 0 1 1 1 0 0]。
As a kind of possible implementation, existing model can also be utilized, for example utilizes word2vec, i-th is pushed away
It recommends information and is converted to corresponding vector, obtain primary vector.
Herein it should be noted that in recommendation information other than including text information, some fortunately includes pictorial information, right
In comprising pictorial information, picture feature can be extracted, corresponding picture vector is determined according to the picture feature of extraction, in turn
According to the corresponding text vector of text information and the corresponding picture vector of pictorial information, obtain recommendation information corresponding first to
Amount.For example, text vector can be added with picture vector, obtain primary vector by the way of summation.
Step 102, according to primary vector, the corresponding secondary vector of i recommendation information before determining.
In the present embodiment, it is determined that, can be further according to primary vector, really after the primary vector of i-th of recommendation information
Determine the corresponding secondary vector of preceding i recommendation information in recommendation list.
It, can be first according to recommendation each in recommendation list when determining secondary vector as a kind of possible implementation
Breath and positional relationship of the i+1 recommendation information in recommendation list, determine the corresponding weight of each recommendation information.For example, root
According to the pattern that shows of recommendation list, position of the i+1 recommendation information in recommendation list is first determined, and then push away according to preceding i
The positional relationship between information and i+1 recommendation information is recommended, the corresponding weight of i recommendation information before determining.Wherein, with i-th+
The distance of 1 recommendation information is closer, and corresponding weight is bigger.
For example, it is assumed that the capacity of recommendation list is 8, can show that 8 recommendation informations, 8 recommendation informations are being recommended
Positional relationship in list is as shown in Figure 2.Assuming that i=3, i.e., current 4th recommendation information to be predicted, from figure 2 it can be seen that
2nd recommendation information and the 3rd recommendation information are adjacent with the 4th region of recommendation information is shown, the 1st recommendation information and displaying
The distance between region of 4th recommendation information farther out, then can distribute lesser weight for the 1st recommendation information, be the 2nd
Recommendation information and the 3rd recommendation information distribute biggish weight, for example, the 1st corresponding weight of recommendation information is the 0.2, the 2nd
Recommendation information and the 3rd corresponding weight of recommendation information are 0.4.
In turn, according to the corresponding weight of each recommendation information and vector, before determining i recommendation information corresponding second to
Amount.Wherein, the corresponding vector of each recommendation information in recommendation list, can be using corresponding with i-th of recommendation information is determined the
The identical mode of one vector determines that, for i-th of recommendation information, corresponding vector is primary vector.Preceding i are pushed away
Information is recommended, can first calculate the product of each recommendation information corresponding vector and weight, then resulting i result is subjected to phase
Add, obtains secondary vector.
As a kind of possible implementation, before determining in i recommendation information, the corresponding weight of each recommendation information
When, different weights can also be distributed for each recommendation information according to the sequence of each recommendation information.For example, the sequence of recommendation information
More forward, corresponding weight is smaller.For example, it is assumed that i=3, then for the 1st recommendation information, corresponding weight can be 0.2,
For the 2nd recommendation information, corresponding weight can be 0.3, and for the 3rd recommendation information, corresponding weight can be 0.5.
In turn, secondary vector is calculated by weighted summation according to the corresponding weight of each recommendation information and vector.
Step 103, according to primary vector and secondary vector, determine that Candidate Recommendation information concentrates each Candidate Recommendation information
Click probability.
Wherein, Candidate Recommendation information collection, the search term that can be inputted according to user determines, searches what is recalled according to search term
Hitch fruit is as Candidate Recommendation information collection;It is determined alternatively, can also be recorded according to the historical viewings of user, such as according to user couple
The operation of previous recommendation list, history recommendation list determines multiple relevant information, constitutes Candidate Recommendation information collection.It needs to illustrate
, Candidate Recommendation information concentration does not include the recommendation information having shown that in recommendation list.
In the present embodiment, it is determined that, can be according to primary vector and secondary vector, really after primary vector and secondary vector
Determine the click probability that Candidate Recommendation information concentrates each Candidate Recommendation information.
It, can be using preparatory trained deep neural network model (Deep as a kind of possible implementation
Neural Networks, DNN), primary vector and secondary vector are input in DNN model, and input Candidate Recommendation letter one by one
The corresponding vector of each Candidate Recommendation information concentrated is ceased, concentrates each Candidate Recommendation information difference to obtain Candidate Recommendation information
Corresponding click probability.
Step 104, it according to the click probability of each Candidate Recommendation information, is concentrated from Candidate Recommendation information and chooses i+1
Recommendation information.
In the present embodiment, it is determined that, can be according to each candidate after the corresponding click probability of each Candidate Recommendation information
The corresponding click probability of recommendation information selects i+1 recommendation information from Candidate Recommendation information concentration, and by determining i-th+
1 recommendation information is shown in recommendation list.For example, will click on the highest Candidate Recommendation information of probability, it is determined as i+1 and pushes away
Recommend information.
The recommendation information processing method of the present embodiment is the process repeated, it is determined that i+1 recommendation information
Later, it can repeat the above process, continue to determine the i-th+2 recommendation informations, and so on, until showing the show area in list
Domain is occupied or Candidate Recommendation information concentrates that there is no Candidate Recommendation information.
It should be noted that can be determined in recommendation list in different ways when predicting first recommendation information
First recommendation information.
As an example, when recommendation list is empty (i.e. i=0), two vectors can be generated at random, respectively as the
One vector sum secondary vector.For determining primary vector by the way of bag of words list and one-hot coding, it can give birth at random
At primary vector and secondary vector, wherein the dimension of primary vector and secondary vector, it is consistent with the length of bag of words list.In turn,
According to primary vector and secondary vector, determine that Candidate Recommendation information concentrates the click probability of each Candidate Recommendation information, Jin Ergen
Strong point hits determine the probability and goes out first recommendation information.
As an example, when recommendation list is empty (i.e. i=0), traditional scoring model can also be used, to candidate
Each Candidate Recommendation information that recommendation information is concentrated is given a mark, and the highest Candidate Recommendation information that will give a mark is determined as first and pushes away
Recommend information.
The recommendation information processing method of the present embodiment, by determining i-th of recommendation information corresponding first in recommendation list
Vector, and i recommendation information corresponding secondary vector before being determined according to primary vector, so according to primary vector and second to
Amount, determines that Candidate Recommendation information concentrates the click probability of each Candidate Recommendation information, according to the click of each Candidate Recommendation information
Probability is concentrated from Candidate Recommendation information and chooses i+1 recommendation information.Pass through the vector according to previous recommendation information as a result,
It is realized to determine current recommendation information according between recommendation information with the vector for all recommendation informations that sequence is completed
Relevance recommendation information is ranked up, be conducive to the clicking rate for improving recommendation information in recommendation list, improve user
Experience.
In practical application, user will not usually leave immediately after having browsed the information currently recommended, and in most cases, use
Family is often repeatedly interacted with product (such as browser) when browsing information, to obtain more information.At one
Between in section, such as half an hour, one hour etc., enter a product from user and leave the product to the user, occur in the meantime
All information browse behaviors, a session can be referred to as.It cannot understand, the information for including in session is able to reflect out and uses
Interest in the short time of family is conducive to targetedly to user's recommendation information.Based on this, the application also proposed another kind and push away
Recommend information processing method.Fig. 3 is the flow diagram of another kind recommendation information processing method provided by the embodiment of the present application.
As shown in figure 3, the recommendation information processing method, may comprise steps of:
Step 201, the corresponding primary vector of i-th of recommendation information in recommendation list is determined, wherein recommendation list is and pushes away
Corresponding j-th of the recommendation list of object is recommended, i and j are positive integer.
Wherein, recommended is the user using terminal corresponding to recommendation list.
Step 202, according to primary vector, the corresponding secondary vector of i recommendation information before determining.
In the present embodiment, description to step 201- step 202 be may refer in previous embodiment to step 101- step
102 description, details are not described herein again.
Step 203, recommended currently corresponding session vector is determined.
Session refers to all information browse behaviors that recommended occurs in a certain period of time, is able to reflect out recommendation pair
As interest in a short time.Include the information that recommended is clicked, browsed in session, can determine that recommended is clicked, browsed
The corresponding vector of the information crossed, and then session vector is determined according to identified vector.
It, can be using pre- when determining recommended currently corresponding session vector as a kind of possible implementation
First trained Recognition with Recurrent Neural Network (Recurrent Neural Networks, RNN) model, it will user couple during words
The click of recommendation information in each recommendation list, browsing information input into RNN model, with obtain each recommendation list to
Amount, according to the vector of each recommendation list, that is, can determine recommended currently corresponding session vector.Wherein, RNN model can be adopted
With unidirectional RNN model, to describe the sequential relationship in session between each information, to preferably portray recommended in a short time
Interest and reading model.
It, can first basis when determining recommended currently corresponding session vector as a kind of possible implementation
The corresponding click information of each recommendation information in preceding j recommendation list, each recommendation list correspondence in j recommendation list before determining
Click vector, and then according to corresponding vector and the preceding j recommendation list clicked of recommendation list each in preceding j recommendation list
Recommendation order determines current corresponding session vector.For example, it is directed to each recommendation list, it can be using described in previous embodiment
Bag of words list and one-hot coding mode, determine user clicked in recommendation list each recommendation information it is corresponding to
Amount, and then sum to resulting vector, obtain the corresponding click vector of the recommendation list.Then, it is arranged according to preceding j recommendation
The recommendation order of table distributes corresponding weight for each recommendation list, and then by the way of weighted sum, determines current corresponding
Session vector.
As a kind of possible implementation, when determining recommended currently corresponding session vector, can first determine
In preceding j recommendation list, all corresponding vectors of recommendation information clicked of user, and then resulting institute's directed quantity is asked
With obtain recommended currently corresponding session vector.
Step 204, according to current corresponding session vector, primary vector and secondary vector, Candidate Recommendation information collection is determined
In each Candidate Recommendation information click probability.
In the present embodiment, it is determined that recommended, can be according to current corresponding meeting currently after corresponding session vector
Vector, primary vector and secondary vector are talked about, determines that Candidate Recommendation information concentrates the click probability of each Candidate Recommendation information.
For example, session vector, primary vector and secondary vector can be input in DNN model, and input is candidate one by one
The corresponding vector of each Candidate Recommendation information that recommendation information is concentrated obtains Candidate Recommendation information and concentrates each Candidate Recommendation information
Corresponding click probability.
Step 205, it according to the click probability of each Candidate Recommendation information, is concentrated from Candidate Recommendation information and chooses i+1
Recommendation information.
The highest Candidate Recommendation information of probability is clicked for example, Candidate Recommendation information can be concentrated, is determined as i+1 and pushes away
Recommend information.
The recommendation information processing method of the present embodiment, by determining i-th of recommendation information corresponding first in recommendation list
Vector, the corresponding secondary vector of preceding i recommendation information, and determine recommended currently corresponding session vector, and then basis
Session vector, primary vector and secondary vector determine Candidate Recommendation information and in each Candidate Recommendation information click probability, root
Strong point hits probability and concentrates selection i+1 recommendation information from Candidate Recommendation information, not only allows between recommendation information as a result,
Relevance, it is also contemplated that the interest in user's short time, so as to be more able to satisfy user current for recommendation information in recommendation list
Demand, further the user experience is improved.
When user is in different scenes, the information content of browsing difference.For example, leisure time at night, user
It is more to the browsing of video resource, and under conditions of Outdoor Network condition is poor, user can be more to the browsing of graph text information
It is some.Therefore, when carrying out information recommendation, the scene in conjunction with locating for user, can further satisfaction user under different scenes
Individual demand.Based on this, present applicant proposes another recommendation information processing method, Fig. 4 is mentioned by the embodiment of the present application
The flow diagram of another the recommendation information processing method supplied.
As shown in figure 4, the recommendation information processing method, may include a kind of step:
Step 301, the corresponding primary vector of i-th of recommendation information in recommendation list is determined, wherein i is positive integer.
Step 302, according to primary vector, the corresponding secondary vector of i recommendation information before determining.
In the present embodiment, description to step 301- step 302 be may refer in previous embodiment to step 101- step
102 description, details are not described herein again.
Step 303, the corresponding scene vector of recommendation list is determined.
Wherein, scene information includes but is not limited to temporal information, the attribute information of terminal device, network state, position letter
Breath, refreshing frequency etc..
When recommended is under different scenes, the information that user is browsed is different.For example, recommended institute
When the network state of the terminal device used is poor, recommended is tended to browse text information;It is network-like when terminal device
When state is preferably and recommended is in leisure state, recommended is tended to browse video information.Thus, can in the present embodiment
To predict recommendation information in conjunction with current scene information, so that recommendation information and current scene matching.Specifically, it can obtain
Current scene information is taken, and then determines the corresponding scene vector of recommendation list.
It, can be with when determining the corresponding scene vector of recommendation list in a kind of possible implementation of the embodiment of the present application
According to the corresponding temporal information of recommendation list, refreshing frequency, the attribute of terminal device, the position of terminal device and terminal device
At least one of network state, determine the corresponding scene vector of recommendation list.
As an example, it can determine and recommend according to the bag of words list used when determining primary vector and secondary vector
The corresponding scene vector of list.When scene information is multiple, the corresponding vector of each scene information can be determined respectively, then right
Multiple vectors are summed or are averaging, to obtain scene vector.
As an example, possible scene information can be counted in advance, and corresponding for the setting of each scene information
Vector stores the mapping relations between each scene information and corresponding vector, in turn, determine the corresponding scene of recommendation list to
When amount, corresponding scene vector is determined by inquiring mapping relations according to scene information.It, can be with when scene information is multiple
The corresponding vector of each scene information is determined respectively, then multiple vectors are summed or are averaging, to obtain scene vector.
Step 304, according to scene vector, primary vector and secondary vector, determine that Candidate Recommendation information concentrates each candidate
The click probability of recommendation information.
In the present embodiment, it is determined that, can be according to primary vector, after primary vector, secondary vector and scene vector
Two vector sum scene vectors determine that Candidate Recommendation information concentrates the click probability of each Candidate Recommendation information.
As a kind of possible implementation, determine that Candidate Recommendation information concentrates the click probability of each Candidate Recommendation information
When, scene vector, primary vector and secondary vector can be input in preparatory trained DNN model, and input is waited one by one
The corresponding vector of each Candidate Recommendation information for selecting recommendation information to concentrate obtains Candidate Recommendation information and concentrates each Candidate Recommendation letter
Cease corresponding click probability.
Further, in a kind of possible implementation of the embodiment of the present application, preparatory trained DNN model can be set
Setting can first basis when determining that Candidate Recommendation information concentrates the click probability of each Candidate Recommendation information there are two prediction interval
The corresponding third vector of each recommendation information, primary vector and secondary vector, determine the corresponding primary of each Candidate Recommendation information
Description vectors, wherein the corresponding third vector of each recommendation information can be obtained using mode identical with primary vector is determined
?.Then, according to primary description vectors and scene vector, determine that Candidate Recommendation information concentrates the click of each Candidate Recommendation information
Probability.As a result, current scene information is just combined in the final click probability for determining Candidate Recommendation information, to improve
Influence degree of the scene information to prediction result, makes recommendation information be more in line with current scene.
Step 305, it according to the click probability of each Candidate Recommendation information, is concentrated from Candidate Recommendation information and chooses i+1
Recommendation information.
In the present embodiment, it is determined that after the click probability of each Candidate Recommendation information, can be believed according to each Candidate Recommendation
The click probability of breath determines i+1 recommendation information from Candidate Recommendation information concentration.
The recommendation information processing method of the present embodiment, by determining i-th of recommendation information corresponding first in recommendation list
Vector, the corresponding secondary vector of preceding i recommendation information, and determine the corresponding scene vector of recommendation list, and then according to scene
Vector, primary vector and secondary vector determine Candidate Recommendation information and in each Candidate Recommendation information click probability, according to point
It hits probability and concentrates selection i+1 recommendation information from Candidate Recommendation information, not only allow for the association between recommendation information as a result,
Property, it is also contemplated that current scene information uses so that the recommendation information in recommendation list is more able to satisfy the current demand of user
The recommendation list that family obtains meets current usage scenario, and further the user experience is improved.
The preference that different user browses information is different, for example, some users like browsing the information of military class, some users
Like browsing the relevant information of automobile.In order to realize targetedly to user's recommendation information, meet the personalization of different user
It needs, in a kind of possible implementation of the embodiment of the present application, can be combined with the relevant information of user, carry out letter to user
Breath is recommended.To which present applicant proposes another recommendation information processing method, Fig. 5 is another provided by the embodiment of the present application
The flow diagram of kind recommendation information processing method.
As shown in figure 5, the recommendation information processing method, may comprise steps of:
Step 401, the corresponding primary vector of i-th of recommendation information in recommendation list is determined, wherein i is positive integer.
Step 402, according to primary vector, the corresponding secondary vector of i recommendation information before determining.
In the present embodiment, description to step 401- step 402 be may refer in previous embodiment to step 101- step
102 description, details are not described herein again.
Step 403, the portrait vector of the corresponding recommended of recommendation list is determined.
Because the occupation of recommended, hobby are different, different recommendeds information of interest is also different.For needle
It realizes personalized information recommendation to different recommended, in the present embodiment, can also be determined according to the information of recommended
Recommendation list.Wherein, the information of recommended includes but is not limited to the essential informations such as age, gender, the occupation of recommended, with
And historical viewings information of recommended etc..
In the present embodiment, the information of recommended can be first obtained, and then determine according to the information of recommended and recommend column
The portrait vector of the corresponding recommended of table.
As an example, it may be predetermined that the essential informations such as gender, age, hobby of recommended are corresponding
Vector, and establish the mapping relations between each essential information and corresponding vector, when including recommending in the information of recommended
When the essential information of object, by inquiring mapping relations, the corresponding vector of essential information is determined.It is wrapped when in the information of recommended
When including the historical viewings information of recommended, historical viewings information can be determined using mode identical with primary vector is determined
Corresponding vector.In turn, by the corresponding vector of essential information and the corresponding vector of historical viewings information, collectively as recommendation pair
The portrait vector of elephant.
Step 404, according to portrait vector, primary vector and the secondary vector of recommended, Candidate Recommendation information collection is determined
In each Candidate Recommendation information click probability.
For example, when determining that Candidate Recommendation information concentrates the click probability of each Candidate Recommendation information, it can be by recommended
Portrait vector, primary vector and secondary vector be input in preparatory trained DNN model, and input Candidate Recommendation letter one by one
The corresponding vector of each Candidate Recommendation information concentrated is ceased, Candidate Recommendation information is obtained and concentrates each Candidate Recommendation information right respectively
The click probability answered.
Step 405, it according to the click probability of each Candidate Recommendation information, is concentrated from Candidate Recommendation information and chooses i+1
Recommendation information.
For example, Candidate Recommendation information can be concentrated, the highest Candidate Recommendation information of probability is clicked, is determined as i+1
Recommendation information.
The recommendation information processing method of the present embodiment, by determining i-th of recommendation information corresponding first in recommendation list
Vector, the corresponding secondary vector of preceding i recommendation information, and determine recommended portrait vector, and then according to portrait vector,
Primary vector and secondary vector determine Candidate Recommendation information and in each Candidate Recommendation information click probability, according to click probability
It is concentrated from Candidate Recommendation information and chooses i+1 recommendation information, not only allow for the relevance between recommendation information as a result, also
The information for considering user itself realizes targetedly information recommendation, improves the personalization of information recommendation, further mentions
User experience is risen.
It should be noted that the recommendation information processing method of previous embodiment, can individually implement, reality can also be combined
It applies, for example, the information and scene information in conjunction with user carry out information recommendation, alternatively, carrying out in conjunction with session information and scene information
Information recommendation.Foregoing embodiments are only as an example, and cannot function as the limitation to the application.
Fig. 6 is the example architecture figure for realizing the recommendation information order models of recommendation information processing method of the application.Such as Fig. 6
Shown, which includes two full articulamentums and two full convolutional network (Fully Convolutional
Networks layer, FCN) layer, wherein FCN layers, for predicting recommendation information, can receive the defeated of arbitrary dimension
Enter, so as to avoid the limitation of input size, improves flexibility.From fig. 6, it can be seen that when carrying out recommendation information prediction,
First by user information vector, Candidate Recommendation information vector, i-th of recommendation information vector, preceding i recommendation information vector and meeting
Words vector is input to first full articulamentum, and the output result of first full articulamentum is input to first FCN layers, with true
Determine initial recommendation information.Wherein, session vector can according to the session information in user's short time, using unidirectional RNN model come
It determines.By first full articulamentum, it may learn the point of interest distribution for including in session information, resource type, show sample
The customized information of user is arrived in the information such as formula, and study.Then, determining initial recommendation information and scene vector are input to
Second full articulamentum, and the output result of second full articulamentum is input to second FCN layers, available i+1
The click probability of recommendation information.As shown in Figure 6, combine scene information progress recommendation information pre- at second full articulamentum
It surveys, influence of the scene information to prediction result is enhanced, more to embody individual demand of the user under different scenes.Using
When the recommendation information order models are ranked up recommendation information, the relevance between recommendation information had both been considered, it is contemplated that
Individual demand, the demand under different scenes and the short-term interest of user, improve recommendation information prediction accuracy and
Clicking rate, the user experience is improved.
In order to realize above-described embodiment, the application also proposes a kind of recommendation information processing unit.
Fig. 7 is a kind of structural schematic diagram of recommendation information processing unit provided by the embodiment of the present application.
As shown in fig. 7, the recommendation information processing unit 50 include: the first determining module 510, it is the second determining module 520, general
Rate determining module 530 and selecting module 540.Wherein,
First determining module 510, for determining the corresponding primary vector of i-th of recommendation information in recommendation list, wherein i
For positive integer.
Second determining module 520, for determining the corresponding secondary vector of preceding i recommendation information according to primary vector.
In a kind of possible implementation of the embodiment of the present application, the second determining module 520 is specifically used for basis and each pushes away
Information and positional relationship of the i+1 recommendation information in recommendation list are recommended, determines the corresponding weight of each recommendation information;According to
The corresponding weight of each recommendation information and vector, the corresponding secondary vector of i recommendation information before determining.
Probability determination module 530, for determining that Candidate Recommendation information concentrates each time according to primary vector and secondary vector
Select the click probability of recommendation information.
Selecting module 540 is concentrated from Candidate Recommendation information and is chosen for the click probability according to each Candidate Recommendation information
I+1 recommendation information.
In a kind of possible implementation of the embodiment of the present application, recommendation list pushes away for j-th corresponding with recommended
Recommend list, wherein j is positive integer.As shown in figure 8, on the basis of embodiment as shown in Figure 7, the recommendation information processing unit
50 can also include:
Session vector determining module 550, for determining recommended currently corresponding session vector.
Specifically, session vector determining module 550 is used for according to the corresponding point of recommendation information each in preceding j recommendation list
Hit information, the corresponding click vector of each recommendation list in j recommendation list before determining;According to each in preceding j recommendation list
The corresponding recommendation order for clicking vector and preceding j recommendation list of recommendation list, determines current corresponding session vector.
To which in the present embodiment, probability determination module 530 is also used to according to current corresponding session vector, primary vector
And secondary vector, determine that Candidate Recommendation information concentrates the click probability of each Candidate Recommendation information.
By when determining i+1 recommendation information, determining the click probability of Candidate Recommendation information in conjunction with session vector,
And then according to determine the probability i+1 recommendation information is clicked, the relevance between recommendation information is not only allowed for, it is also contemplated that use
Interest in the short time of family further improves so that the recommendation information in recommendation list is more able to satisfy the current demand of user
User experience.
In a kind of possible implementation of the embodiment of the present application, as shown in figure 9, on the basis of embodiment as shown in Figure 7
On, which can also include:
Scene vector determining module 560, for determining the corresponding scene vector of recommendation list.
Specifically, scene vector determining module 560 is used for according to the corresponding temporal information of recommendation list, refreshing frequency, end
At least one of the attribute of end equipment, the position of terminal device and network state of terminal device determine that recommendation list is corresponding
Scene vector.
To, in the present embodiment, probability determination module 530 is also used to according to scene vector, primary vector and secondary vector,
Determine that Candidate Recommendation information concentrates the click probability of each Candidate Recommendation information.
Specifically, probability determination module 530 is used for according to the corresponding third vector of each recommendation information, primary vector and the
Two vectors determine the corresponding primary description vectors of each Candidate Recommendation information;According to primary description vectors and scene vector, determine
Candidate Recommendation information concentrates the click probability of each Candidate Recommendation information.
By determining the click probability of Candidate Recommendation information in conjunction with the corresponding scene vector of recommendation list, and then according to point
Probability is hit to determine i+1 recommendation information, so that not only allowing for the association between recommendation information when predicting recommendation information
Property, it is also contemplated that current scene information uses so that the recommendation information in recommendation list is more able to satisfy the current demand of user
The recommendation list that family obtains meets current usage scenario, and further the user experience is improved.
In a kind of possible implementation of the embodiment of the present application, as shown in Figure 10, on the basis of embodiment as shown in Figure 7
On, which can also include:
Portrait vector determining module 570, for determining the portrait vector of the corresponding recommended of recommendation list.
To which in the present embodiment, probability determination module 530 is also used to portrait vector, primary vector according to recommended
And secondary vector, determine that Candidate Recommendation information concentrates the click probability of each Candidate Recommendation information.
The click probability of Candidate Recommendation information is determined by the portrait vector in conjunction with recommended, and then general according to clicking
Rate determines i+1 recommendation information, so that when predicting recommendation information the relevance between recommendation information is not only allowed for,
The information for also contemplating user itself realizes targetedly information recommendation, improves the personalization of information recommendation, further
The user experience is improved.
It should be noted that the aforementioned explanation to recommendation information processing method embodiment is also applied for the embodiment
Recommendation information processing unit, realization principle is similar, and details are not described herein again.
The recommendation information processing unit of the embodiment of the present application, by determining, i-th of recommendation information is corresponding in recommendation list
Primary vector, and the corresponding secondary vector of preceding i recommendation information is determined according to primary vector, and then according to primary vector and second
Vector determines that Candidate Recommendation information concentrates the click probability of each Candidate Recommendation information, according to the point of each Candidate Recommendation information
Probability is hit, is concentrated from Candidate Recommendation information and chooses i+1 recommendation information.As a result, by according to previous recommendation information to
Measure and be completed the vector of all recommendation informations of sequence, to determine current recommendation information, realize according to recommendation information it
Between relevance recommendation information is ranked up, be conducive to the clicking rate for improving recommendation information in recommendation list, improve use
Family experience.
In order to realize above-described embodiment, the application also proposes a kind of computer equipment, comprising: processor and memory.Its
In, processor runs journey corresponding with executable program code by reading the executable program code stored in memory
Sequence, for realizing recommendation information processing method as in the foregoing embodiment.
Figure 11 is the structural schematic diagram of computer equipment provided by the embodiment of the present application, shows and is suitable for being used to realizing this
Apply for the block diagram of the exemplary computer device 90 of embodiment.The computer equipment 90 that Figure 11 is shown is only an example,
Should not function to the embodiment of the present application and use scope bring any restrictions.
As shown in figure 11, computer equipment 90 is showed in the form of general purpose computing device.The component of computer equipment 90
Can include but is not limited to: one or more processor or processing unit 906, system storage 910 connect not homologous ray
The bus 908 of component (including system storage 910 and processing unit 906).
Bus 908 indicates one of a few class bus structures or a variety of, including memory bus or Memory Controller,
Peripheral bus, graphics acceleration port, processor or the local bus using any bus structures in a variety of bus structures.It lifts
For example, these architectures include but is not limited to industry standard architecture (Industry Standard
Architecture;Hereinafter referred to as: ISA) bus, microchannel architecture (Micro Channel Architecture;Below
Referred to as: MAC) bus, enhanced isa bus, Video Electronics Standards Association (Video Electronics Standards
Association;Hereinafter referred to as: VESA) local bus and peripheral component interconnection (Peripheral Component
Interconnection;Hereinafter referred to as: PCI) bus.
Computer equipment 90 typically comprises a variety of computer system readable media.These media can be it is any can be by
The usable medium that computer equipment 90 accesses, including volatile and non-volatile media, moveable and immovable medium.
System storage 910 may include the computer system readable media of form of volatile memory, such as deposit at random
Access to memory (Random Access Memory;Hereinafter referred to as: RAM) 911 and/or cache memory 912.Computer is set
Standby 90 may further include other removable/nonremovable, volatile/non-volatile computer system storage mediums.Only
As an example, storage system 913 can be used for reading and writing immovable, non-volatile magnetic media (Figure 11 do not show, commonly referred to as
" hard disk drive ").Although being not shown in Figure 11, can provide for reading removable non-volatile magnetic disk (such as " floppy disk ")
The disc driver write, and to removable anonvolatile optical disk (such as: compact disc read-only memory (Compact Disc Read
Only Memory;Hereinafter referred to as: CD-ROM), digital multi CD-ROM (Digital Video Disc Read Only
Memory;Hereinafter referred to as: DVD-ROM) or other optical mediums) read-write CD drive.In these cases, each driving
Device can be connected by one or more data media interfaces with bus 908.System storage 910 may include at least one
Program product, the program product have one group of (for example, at least one) program module, these program modules are configured to perform this
Apply for the function of each embodiment.
Computer-readable signal media may include in a base band or as carrier wave a part propagate data-signal,
Wherein carry computer-readable program code.The data-signal of this propagation can take various forms, including --- but
It is not limited to --- electromagnetic signal, optical signal or above-mentioned any appropriate combination.Computer-readable signal media can also be
Any computer-readable medium other than computer readable storage medium, which can send, propagate or
Transmission is for by the use of instruction execution system, device or device or program in connection.
The program code for including on computer-readable medium can transmit with any suitable medium, including --- but it is unlimited
In --- wireless, electric wire, optical cable, RF etc. or above-mentioned any appropriate combination.
Can with one or more programming languages or combinations thereof come write for execute the application operation computer
Program code, described program design language include object oriented program language-such as Java, Smalltalk, C++,
It further include conventional procedural programming language-such as " C " language or similar programming language.Program code can be with
It fully executes, partly execute on the user computer on the user computer, being executed as an independent software package, portion
Divide and partially executes or executed on a remote computer or server completely on the remote computer on the user computer.
Program/utility 914 with one group of (at least one) program module 9140, can store and deposit in such as system
In reservoir 910, such program module 9140 includes but is not limited to operating system, one or more application program, Qi Tacheng
It may include the realization of network environment in sequence module and program data, each of these examples or certain combination.Program
Module 9140 usually executes function and/or method in embodiments described herein.
Computer equipment 90 can also be with one or more external equipments 10 (such as keyboard, sensing equipment, display 100
Deng) communication, can also be enabled a user to one or more equipment interact with the terminal device 90 communicate, and/or with make
Any equipment (such as network interface card, the modulation /demodulation that the computer equipment 90 can be communicated with one or more of the other calculating equipment
Device etc.) communication.This communication can be carried out by input/output (I/O) interface 902.Also, computer equipment 90 can be with
Pass through network adapter 900 and one or more network (such as local area network (Local Area Network;Hereinafter referred to as:
LAN), wide area network (Wide Area Network;Hereinafter referred to as: WAN) and/or public network, for example, internet) communication.Such as figure
Shown in 11, network adapter 900 is communicated by bus 908 with other modules of computer equipment 90.Although should be understood that Figure 11
In be not shown, can in conjunction with computer equipment 90 use other hardware and/or software module, including but not limited to: microcode is set
Standby driver, redundant processing unit, external disk drive array, RAID system, tape drive and data backup storage system
System etc..
Processing unit 906 by the program that is stored in system storage 910 of operation, thereby executing various function application with
And data processing, such as realize the recommendation information processing method referred in previous embodiment.
In order to realize above-described embodiment, the application also proposes a kind of non-transitorycomputer readable storage medium, deposits thereon
Computer program is contained, when which is executed by processor, realizes recommendation information processing method as in the foregoing embodiment.
In order to realize above-described embodiment, the application also proposes a kind of computer program product, when the computer program produces
When instruction in product is executed by processor, recommendation information processing method as in the foregoing embodiment is realized.
In the description of this specification, reference term " one embodiment ", " some embodiments ", " example ", " specifically show
The description of example " or " some examples " etc. means specific features, structure, material or spy described in conjunction with this embodiment or example
Point is contained at least one embodiment or example of the application.In the present specification, schematic expression of the above terms are not
It must be directed to identical embodiment or example.Moreover, particular features, structures, materials, or characteristics described can be in office
It can be combined in any suitable manner in one or more embodiment or examples.In addition, without conflicting with each other, the skill of this field
Art personnel can tie the feature of different embodiments or examples described in this specification and different embodiments or examples
It closes and combines.
In addition, term " first ", " second " are used for descriptive purposes only and cannot be understood as indicating or suggesting relative importance
Or implicitly indicate the quantity of indicated technical characteristic.Define " first " as a result, the feature of " second " can be expressed or
Implicitly include at least one this feature.In the description of the present application, the meaning of " plurality " is at least two, such as two, three
It is a etc., unless otherwise specifically defined.
Any process described otherwise above or method description are construed as in flow chart or herein, and expression includes
It is one or more for realizing custom logic function or process the step of executable instruction code module, segment or portion
Point, and the range of the preferred embodiment of the application includes other realization, wherein can not press shown or discussed suitable
Sequence, including according to related function by it is basic simultaneously in the way of or in the opposite order, to execute function, this should be by the application
Embodiment person of ordinary skill in the field understood.
Expression or logic and/or step described otherwise above herein in flow charts, for example, being considered use
In the order list for the executable instruction for realizing logic function, may be embodied in any computer-readable medium, for
Instruction execution system, device or equipment (such as computer based system, including the system of processor or other can be held from instruction
The instruction fetch of row system, device or equipment and the system executed instruction) it uses, or combine these instruction execution systems, device or set
It is standby and use.For the purpose of this specification, " computer-readable medium ", which can be, any may include, stores, communicates, propagates or pass
Defeated program is for instruction execution system, device or equipment or the dress used in conjunction with these instruction execution systems, device or equipment
It sets.The more specific example (non-exhaustive list) of computer-readable medium include the following: there is the electricity of one or more wirings
Interconnecting piece (electronic device), portable computer diskette box (magnetic device), random access memory (RAM), read-only memory
(ROM), erasable edit read-only storage (EPROM or flash memory), fiber device and portable optic disk is read-only deposits
Reservoir (CDROM).In addition, computer-readable medium can even is that the paper that can print described program on it or other are suitable
Medium, because can then be edited, be interpreted or when necessary with it for example by carrying out optical scanner to paper or other media
His suitable method is handled electronically to obtain described program, is then stored in computer storage.
It should be appreciated that each section of the application can be realized with hardware, software, firmware or their combination.Above-mentioned
In embodiment, software that multiple steps or method can be executed in memory and by suitable instruction execution system with storage
Or firmware is realized.Such as, if realized with hardware in another embodiment, following skill well known in the art can be used
Any one of art or their combination are realized: have for data-signal is realized the logic gates of logic function from
Logic circuit is dissipated, the specific integrated circuit with suitable combinational logic gate circuit, programmable gate array (PGA), scene can compile
Journey gate array (FPGA) etc..
Those skilled in the art are understood that realize all or part of step that above-described embodiment method carries
It suddenly is that relevant hardware can be instructed to complete by program, the program can store in a kind of computer-readable storage medium
In matter, which when being executed, includes the steps that one or a combination set of embodiment of the method.
It, can also be in addition, can integrate in a processing module in each functional unit in each embodiment of the application
It is that each unit physically exists alone, can also be integrated in two or more units in a module.Above-mentioned integrated mould
Block both can take the form of hardware realization, can also be realized in the form of software function module.The integrated module is such as
Fruit is realized and when sold or used as an independent product in the form of software function module, also can store in a computer
In read/write memory medium.
Storage medium mentioned above can be read-only memory, disk or CD etc..Although having been shown and retouching above
Embodiments herein is stated, it is to be understood that above-described embodiment is exemplary, and should not be understood as the limit to the application
System, those skilled in the art can be changed above-described embodiment, modify, replace and become within the scope of application
Type.
Claims (11)
1. a kind of recommendation information processing method characterized by comprising
Determine the corresponding primary vector of i-th of recommendation information in recommendation list, wherein i is positive integer;
According to the primary vector, the corresponding secondary vector of i recommendation information before determining;
According to the primary vector and the secondary vector, determine that Candidate Recommendation information concentrates the click of each Candidate Recommendation information
Probability;
According to the click probability of each Candidate Recommendation information, is concentrated from the Candidate Recommendation information and choose i+1 recommendation information.
2. the method as described in claim 1, which is characterized in that the corresponding secondary vector of i recommendation information before the determination, packet
It includes:
According to each recommendation information and positional relationship of the i+1 recommendation information in recommendation list, each recommendation information is determined
Corresponding weight;
According to the corresponding weight of each recommendation information and vector, the corresponding secondary vector of i recommendation information before determining.
3. the method as described in claim 1, which is characterized in that the recommendation list pushes away for j-th corresponding with recommended
Recommend list, wherein j is positive integer;
The determining Candidate Recommendation information is concentrated before the click probability of each Candidate Recommendation information, further includes:
Determine the recommended currently corresponding session vector;
The determining Candidate Recommendation information concentrates the click probability of each Candidate Recommendation information, comprising:
According to the current corresponding session vector, the primary vector and the secondary vector, Candidate Recommendation information collection is determined
In each Candidate Recommendation information click probability.
4. method as claimed in claim 3, which is characterized in that the determination recommended currently corresponding session to
Amount, comprising:
According to the corresponding click information of recommendation information each in preceding j recommendation list, each recommendation in j recommendation list before determining
The corresponding click vector of list;
Pushing away for vector and the preceding j recommendation list is clicked according to recommendation list each in the preceding j recommendation list is corresponding
Sequence is recommended, determines current corresponding session vector.
5. the method as described in claim 1, which is characterized in that the determining Candidate Recommendation information concentrates each Candidate Recommendation letter
Before the click probability of breath, further includes:
Determine the corresponding scene vector of the recommendation list;
The determining Candidate Recommendation information concentrates the click probability of each Candidate Recommendation information, comprising:
According to the scene vector, the primary vector and the secondary vector, determine that Candidate Recommendation information concentrates each candidate
The click probability of recommendation information.
6. method as claimed in claim 5, which is characterized in that described according to the scene vector, the primary vector and institute
Primary vector is stated, determines that Candidate Recommendation information concentrates the click probability of each Candidate Recommendation information, comprising:
According to the corresponding third vector of each recommendation information, the primary vector and the secondary vector, determine that each candidate pushes away
Recommend the corresponding primary description vectors of information;
According to the primary description vectors and the scene vector, determine that Candidate Recommendation information concentrates each Candidate Recommendation information
Click probability.
7. method as claimed in claim 5, which is characterized in that the corresponding scene vector of the determination recommendation list, packet
It includes:
According to the corresponding temporal information of the recommendation list, refreshing frequency, the attribute of terminal device, terminal device position and
At least one of network state of terminal device determines the corresponding scene vector of the recommendation list.
8. the method according to claim 1 to 7, which is characterized in that the determining Candidate Recommendation information concentrates each time
Before the click probability for selecting recommendation information, further includes:
Determine the portrait vector of the corresponding recommended of the recommendation list;
The determining Candidate Recommendation information concentrates the click probability of each Candidate Recommendation information, comprising:
According to the portrait vector, the primary vector and the secondary vector of the recommended, Candidate Recommendation information collection is determined
In each Candidate Recommendation information click probability.
9. a kind of recommendation information processing unit characterized by comprising
First determining module, for determining the corresponding primary vector of i-th of recommendation information in recommendation list, wherein i is positive whole
Number;
Second determining module, for determining the corresponding secondary vector of preceding i recommendation information according to the primary vector;
Probability determination module, for it is each to determine that Candidate Recommendation information is concentrated according to the primary vector and the secondary vector
The click probability of Candidate Recommendation information;
Selecting module is concentrated from the Candidate Recommendation information for the click probability according to each Candidate Recommendation information and chooses i-th
+ 1 recommendation information.
10. a kind of computer equipment, which is characterized in that including processor and memory;
Wherein, the processor is run by reading the executable program code stored in the memory can be performed with described
The corresponding program of program code, for realizing recommendation information processing method such as of any of claims 1-8.
11. a kind of non-transitorycomputer readable storage medium, is stored thereon with computer program, which is characterized in that the program
Such as recommendation information processing method of any of claims 1-8 is realized when being executed by processor.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810865257.1A CN109255070B (en) | 2018-08-01 | 2018-08-01 | Recommendation information processing method and device, computer equipment and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810865257.1A CN109255070B (en) | 2018-08-01 | 2018-08-01 | Recommendation information processing method and device, computer equipment and storage medium |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109255070A true CN109255070A (en) | 2019-01-22 |
CN109255070B CN109255070B (en) | 2022-04-12 |
Family
ID=65049034
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810865257.1A Active CN109255070B (en) | 2018-08-01 | 2018-08-01 | Recommendation information processing method and device, computer equipment and storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109255070B (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112989276A (en) * | 2019-12-12 | 2021-06-18 | 北京沃东天骏信息技术有限公司 | Evaluation method and device of information push system |
CN112989198A (en) * | 2021-03-30 | 2021-06-18 | 北京三快在线科技有限公司 | Push content determination method, device, equipment and computer-readable storage medium |
CN113239175A (en) * | 2021-06-10 | 2021-08-10 | 中国平安人寿保险股份有限公司 | Method for displaying candidate sentence list and terminal equipment |
WO2023001093A1 (en) * | 2021-07-22 | 2023-01-26 | 北京字节跳动网络技术有限公司 | Article recommendation method and apparatus, and device and storage medium |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103778260A (en) * | 2014-03-03 | 2014-05-07 | 哈尔滨工业大学 | Individualized microblog information recommending system and method |
CN103885968A (en) * | 2012-12-20 | 2014-06-25 | 北京百度网讯科技有限公司 | Method and device for providing recommended information |
CN104572734A (en) * | 2013-10-23 | 2015-04-29 | 腾讯科技(深圳)有限公司 | Question recommendation method, device and system |
CN105678587A (en) * | 2016-01-12 | 2016-06-15 | 腾讯科技(深圳)有限公司 | Recommendation feature determination method and information recommendation method and device |
CN106339510A (en) * | 2016-10-28 | 2017-01-18 | 北京百度网讯科技有限公司 | The click prediction method and device based on artificial intelligence |
CN107424043A (en) * | 2017-06-15 | 2017-12-01 | 北京三快在线科技有限公司 | A kind of Products Show method and device, electronic equipment |
CN107463701A (en) * | 2017-08-15 | 2017-12-12 | 北京百度网讯科技有限公司 | Method and apparatus based on artificial intelligence pushed information stream |
-
2018
- 2018-08-01 CN CN201810865257.1A patent/CN109255070B/en active Active
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103885968A (en) * | 2012-12-20 | 2014-06-25 | 北京百度网讯科技有限公司 | Method and device for providing recommended information |
CN104572734A (en) * | 2013-10-23 | 2015-04-29 | 腾讯科技(深圳)有限公司 | Question recommendation method, device and system |
CN103778260A (en) * | 2014-03-03 | 2014-05-07 | 哈尔滨工业大学 | Individualized microblog information recommending system and method |
CN105678587A (en) * | 2016-01-12 | 2016-06-15 | 腾讯科技(深圳)有限公司 | Recommendation feature determination method and information recommendation method and device |
CN106339510A (en) * | 2016-10-28 | 2017-01-18 | 北京百度网讯科技有限公司 | The click prediction method and device based on artificial intelligence |
CN107424043A (en) * | 2017-06-15 | 2017-12-01 | 北京三快在线科技有限公司 | A kind of Products Show method and device, electronic equipment |
CN107463701A (en) * | 2017-08-15 | 2017-12-12 | 北京百度网讯科技有限公司 | Method and apparatus based on artificial intelligence pushed information stream |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112989276A (en) * | 2019-12-12 | 2021-06-18 | 北京沃东天骏信息技术有限公司 | Evaluation method and device of information push system |
CN112989198A (en) * | 2021-03-30 | 2021-06-18 | 北京三快在线科技有限公司 | Push content determination method, device, equipment and computer-readable storage medium |
CN113239175A (en) * | 2021-06-10 | 2021-08-10 | 中国平安人寿保险股份有限公司 | Method for displaying candidate sentence list and terminal equipment |
WO2023001093A1 (en) * | 2021-07-22 | 2023-01-26 | 北京字节跳动网络技术有限公司 | Article recommendation method and apparatus, and device and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN109255070B (en) | 2022-04-12 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109710805A (en) | Video interactive method and device based on interest cluster | |
CN104462364B (en) | Method and device is recommended in search | |
CN111061946B (en) | Method, device, electronic equipment and storage medium for recommending scenerized content | |
CN109255070A (en) | Recommendation information processing method, device, computer equipment and storage medium | |
CN109376309A (en) | Document recommendation method and device based on semantic label | |
CN104504133B (en) | The recommendation method and device of application program | |
CN110162693A (en) | A kind of method and server of information recommendation | |
CN108491540A (en) | Text message method for pushing, device and intelligent terminal | |
CN109389442A (en) | Method of Commodity Recommendation and device, storage medium and electric terminal | |
WO2016077255A1 (en) | Hybrid explanations in collaborative filter based recommendation system | |
CN105229631A (en) | The collection of reading content, follow the tracks of and present | |
CN104866522A (en) | Method and device for providing answer information for questioner | |
CN110275692A (en) | A kind of recommended method of phonetic order, device, equipment and computer storage medium | |
CN103955464A (en) | Recommendation method based on situation fusion sensing | |
CN109460512A (en) | Recommendation information processing method, device, equipment and storage medium | |
CN107230381A (en) | Recommend method, server and client in a kind of parking stall | |
CN107895056A (en) | A kind of information recommendation method, device, electronic equipment and storage medium | |
CN109241242A (en) | A kind of direct broadcasting room topic recommended method, device, server and storage medium | |
CN108173746A (en) | Friend recommendation method, apparatus and computer equipment | |
CN106874460A (en) | Information recommendation method, apparatus and system | |
CN107908641A (en) | A kind of method and system for obtaining picture labeled data | |
CN110413888A (en) | A kind of books recommended method and device | |
CN109960801A (en) | Data processing method and device | |
CN109726380A (en) | Table edit method and device | |
CN107220854A (en) | A kind of method and device for sending electronic ticket |
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 |