CN109063104A - Method for refreshing, device, storage medium and the terminal device of recommendation information - Google Patents
Method for refreshing, device, storage medium and the terminal device of recommendation information Download PDFInfo
- Publication number
- CN109063104A CN109063104A CN201810846642.1A CN201810846642A CN109063104A CN 109063104 A CN109063104 A CN 109063104A CN 201810846642 A CN201810846642 A CN 201810846642A CN 109063104 A CN109063104 A CN 109063104A
- Authority
- CN
- China
- Prior art keywords
- sequence
- information
- recommendation information
- recommendation
- information sequence
- 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
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The present invention proposes method for refreshing, device, storage medium and the terminal device of a kind of recommendation information, wherein the described method includes: the refresh requests at respond request end, extract recommendation information from recommendation information set and be combined into initial information sequence；Wherein, the initial information sequence includes M recommendation information；Probability is clicked according to the prediction of each recommendation information in the initial information sequence, adjusts putting in order for the initial information sequence, obtains preference information sequence；Recommendation information sequence is extracted from the preference information sequence, the recommendation information sequence includes being sorted in the preference information sequence in the recommendation information of top n, wherein M is greater than N；And the recommendation information sequence is sent to the request end, it is shown with carrying out refreshing in the request end.Using the present invention, under the premise of the recommendation information sequence that this refreshing issues is optimal, it is contemplated that following refresh the recommendation information sequence issued.
Description
Technical field
The present invention relates to field of computer technology more particularly to a kind of method for refreshing of recommendation information, device, storage medium
And terminal device.
Background technique
With the development of internet, diversified recommendation platform is emerged, to Internet user's push or recommendation information.
Such as the push of commodity favor information, the recommendation of the search information of search engine, news information or each field the information such as article
Push.
The recommender system of such recommendation platform is all that issuing for recommendation information is carried out as unit of refreshing.For example, when using
Family sends to server-side and requests when client triggers refresh operation.Server-side respond request simultaneously pushes away n recommendation information composition
It recommends list and is issued to client.Client shows the recommendation list received.Wherein, server-side passes through proposed algorithm, so that under
It issues the n recommendation information of user and puts in order and match the interest of user as far as possible, to attract user to click and read.But
Be, recommender system usually only consider to refresh this in information sequence it is as optimal as possible, and the information sequence between adjacent refreshing
The optimization of column but seldom considers.For example, user requests 3 refreshings, list (1), list (2) and list (3) are respectively obtained.If
Refresh interior recommendation information referring only to this, may be the best list that this refreshes, i.e. local optimum list.But such as
3 list are spliced together composed list by fruit, and the list of composition is not necessarily global best list.In particular, when using
Family carries out pull-up when refreshing, and can be stitched together and is shown with the list that issues of history repeatedly refreshed.At this point, and after connecing
The list of displaying is not necessarily the optimal list of arrangement.Thus, the list that list forms that issues by repeatedly refreshing how is made also to be
The technical issues of arranging optimal list, being current urgent need to resolve.
Traditional technical solution is usually: the proposed algorithm by being based on session (session), in current refresh under
The recommendation information of hair carries out selection arrangement.In the process, by it is past it is multiple refresh issued recommendation information and information it
Between put in order as feature, be added in proposed algorithm and calculated.To, obtained calculated result, can make by
The column shown together combined by the recommendation information list that the recommendation information list and client history that this refreshing issues show
The arrangement of table is optimal.
Although what above-mentioned technical proposal when issuing the recommendation information list of this refreshing, only considered that history issues pushes away
The feature of information list is recommended, then mutually splices displaying in the list for refreshing this refreshing with future, is not necessarily optimal list.?
User continues to influence the viewing experience of user in refresh process.
Summary of the invention
The embodiment of the present invention provides method for refreshing, device, storage medium and the terminal device of a kind of recommendation information, to solve
Or alleviate above one or more technical problems in the prior art.
In a first aspect, the embodiment of the invention provides a kind of method for refreshing of recommendation information, comprising:
The refresh requests at respond request end extract recommendation information from recommendation information set and are combined into initial information sequence；
Wherein, the initial information sequence includes M recommendation information；
Probability is clicked according to the prediction of each recommendation information in the initial information sequence, adjusts the initial information sequence
It puts in order, obtains preference information sequence；
Recommendation information sequence is extracted from the preference information sequence, the recommendation information sequence includes the preference information
It is sorted in sequence in the recommendation information of top n, wherein M is greater than N；And
The recommendation information sequence is sent to the request end, is shown with carrying out refreshing in the request end.
With reference to first aspect, in the first embodiment of first aspect, the refresh requests at the respond request end, from
Recommendation information is extracted in recommendation information set is combined into initial information sequence, comprising:
According to the user information in the refresh requests, corresponding each recommendation information is searched from recommendation information library, is obtained
Recommendation information set；
According to preset proposed algorithm, the recommendation weighted value of each recommendation information in the recommendation information set is calculated；With
And
By each recommendation information in the recommendation information set according to recommendation weighted value sequence；And
Sequence is chosen in preceding M recommendation informations, is combined into initial information sequence.
With reference to first aspect, described according in the initial information sequence in second of embodiment of first aspect
Probability is clicked in the prediction of each recommendation information, adjusts putting in order for the initial information sequence, obtains preference information sequence, packet
It includes:
Putting in order for recommendation information in the initial information sequence is adjusted at random, obtains candidate information sequence；
Probability is clicked by the prediction that sequence estimation model calculates each recommendation information of the candidate information sequence；
Probability is clicked according to the prediction of each recommendation information of the candidate information sequence, calculates the candidate information sequence
Recommender score；And
From the candidate information sequence obtained, the highest candidate information sequence of recommender score is chosen as preference information sequence
Column.
Second of embodiment with reference to first aspect, in the third embodiment of first aspect, the random tune
Recommendation information puts in order in the whole initial information sequence, obtains candidate information sequence, comprising:
Mutation operation is carried out to the arrangement position of recommendation information in the initial information sequence, obtains candidate information sequence；
Wherein, the mutation operation includes randomly selecting two recommendation informations from the initial information sequence to carry out place-exchange.
Second of embodiment with reference to first aspect, in the 4th kind of embodiment of first aspect, the refreshing side
Method further include:
Obtain the training data of the sequence estimation model；Wherein, the training data includes the sample of the request end
The click probability of each recommendation information in recommendation information sequence and the sample recommendation information sequence；And
By the generation time of the sample recommendation information sequence, the training data is inputted into the sequence estimation model,
To be trained update to the sequence estimation model.
Second aspect, the embodiment of the present invention also provide a kind of refreshing apparatus of recommendation information, comprising:
Initiation sequence generation module extracts recommendation from recommendation information set for the refresh requests at respond request end
Breath is combined into initial information sequence；Wherein, the initial information sequence includes M recommendation information；
Preferred sequence generation module, for clicking probability according to the prediction of each recommendation information in the initial information sequence,
Putting in order for the initial information sequence is adjusted, preference information sequence is obtained；
Recommend sequential extraction procedures module, for extracting recommendation information sequence, the recommendation from the preference information sequence
Breath sequence includes being sorted in the preference information sequence in the recommendation information of top n, wherein M is greater than N；And
Sequence sending module, for sending the recommendation information sequence to the request end, to be carried out in the request end
Refresh display.
In conjunction with second aspect, in the first embodiment of second aspect, the initiation sequence generation module includes:
Recommendation information searching unit, for being searched from recommendation information library according to the user information in the refresh requests
Corresponding each recommendation information, obtains recommendation information set；
Information weight calculation unit, for calculating respectively pushing away in the recommendation information set according to preset proposed algorithm
Recommend the recommendation weighted value of information；And
Recommendation information sequencing unit, for arranging each recommendation information in the recommendation information set according to recommendation weighted value
Sequence；And
Recommendation information selection unit is combined into initial information sequence for choosing sequence in preceding M recommendation informations.
In conjunction with second aspect, in second of embodiment of second aspect, the preferred sequence generation module, comprising:
Candidate sequence generation unit is obtained for adjusting putting in order for recommendation information in the initial information sequence at random
Obtain candidate information sequence；
Probability calculation unit is clicked, for calculating each recommendation information of the candidate information sequence by sequence estimation model
Prediction click probability；
Probability is clicked in recommender score computing unit, the prediction for each recommendation information according to the candidate information sequence,
Calculate the recommender score of the candidate information sequence；And
Sequence selection unit, for choosing the highest candidate information sequence of recommender score from the candidate information sequence obtained
Column are used as preference information sequence.
In conjunction with second of embodiment of second aspect, in the third embodiment of second aspect, the candidate sequence
Generation unit is specifically used for: carrying out mutation operation to the arrangement position of recommendation information in the initial information sequence, obtains candidate
Information sequence；Wherein, the mutation operation includes randomly selecting two recommendation informations from the initial information sequence to carry out position
Set exchange.
In conjunction with second of embodiment of second aspect, in the 4th kind of embodiment of second aspect, the refreshing apparatus
Further include:
Training data module, for obtaining the training data of the sequence estimation model；Wherein, the training data includes
The click probability of each recommendation information in the sample recommendation information sequence of the request end and the sample recommendation information sequence；And
Model training module inputs the training data for pressing the generation time of the sample recommendation information sequence
The sequence estimation model, to be trained update to the sequence estimation model.
The function of described device can also execute corresponding software realization by hardware realization by hardware.It is described
Hardware or software include one or more modules corresponding with above-mentioned function.
It include processor and memory, the memory in the refreshing structure of recommendation information in a possible design
Refreshing apparatus for recommendation information executes the refurbishing procedure of recommendation information in above-mentioned first aspect, the processor is configured to
For executing the program stored in the memory.The refreshing apparatus of the recommendation information can also include communication interface, be used for
The refreshing apparatus and other equipment or communication of recommendation information.
The third aspect, the embodiment of the present invention also provide a kind of computer readable storage medium, the refreshing for recommendation information
Computer software instructions used in device, involved by the method for refreshing including the recommendation information for executing above-mentioned first aspect
And program.
One of technical solution in above-mentioned technical proposal have the following advantages that or the utility model has the advantages that
The embodiment of the present invention generates the initial of the recommendation information sequence relatively issued a length when responding this refresh requests
Information sequence, and the prediction in initial information sequence in each recommendation information is combined to click probability, it adjusts putting in order for sequence and obtains
Preference information sequence is obtained, and forms recommendation from the N number of recommendation information for extracting foremost in the column face in preference information sequence
Sequence is ceased, is handed down to request end to be shown in request end.It therefore, is optimal in the recommendation information sequence that this refreshing issues
Under the premise of, it is contemplated that following refresh the recommendation information sequence issued.So that continuing in refresh process in user, refresh institute
The content of display more attracts user, and user obtains better viewing experience.
Above-mentioned general introduction is merely to illustrate that the purpose of book, it is not intended to be limited in any way.Except foregoing description
Schematical aspect, except embodiment and feature, by reference to attached drawing and the following detailed description, the present invention is further
Aspect, embodiment and feature, which will be, to be readily apparent that.
Detailed description of the invention
In the accompanying drawings, unless specified otherwise herein, otherwise indicate the same or similar through the identical appended drawing reference of multiple attached drawings
Component or element.What these attached drawings were not necessarily to scale.It should be understood that these attached drawings depict only according to the present invention
Disclosed some embodiments, and should not serve to limit the scope of the present invention.
Fig. 1 is the flow diagram of one embodiment of the method for refreshing of recommendation information provided by the invention；
Fig. 2 is the flow diagram of one embodiment of the generating process of initial information sequence provided by the invention；
Fig. 3 is the flow diagram of one embodiment of the acquisition process of preference information sequence provided by the invention；
Fig. 4 is the schematic diagram of one embodiment of sequence estimation model provided by the invention；
Fig. 5 is that one of the method for refreshing provided by the invention for recommending article applies exemplary flow chart；
Fig. 6 is a reality of the difference of the method for refreshing of the method for refreshing provided by the invention for recommending article and the prior art
Apply the schematic diagram of example；
Fig. 7 is the structural schematic diagram of one embodiment of the refreshing apparatus of recommendation information provided by the invention；
Fig. 8 is the knot of one embodiment of the initiation sequence generation module of the refreshing apparatus of recommendation information provided by the invention
Structure schematic diagram；
Fig. 9 is the knot of one embodiment of the preferred sequence generation module of the refreshing apparatus of recommendation information provided by the invention
Structure schematic diagram；
Figure 10 is the structural schematic diagram of another embodiment of the refreshing apparatus of recommendation information provided by the invention；
Figure 11 is the structural schematic diagram of one embodiment of terminal device provided by the invention.
Specific embodiment
Hereinafter, certain exemplary embodiments are simply just described.As one skilled in the art will recognize that
Like that, without departing from the spirit or scope of the present invention, described embodiment can be modified by various different modes.
Therefore, attached drawing and description are considered essentially illustrative rather than restrictive.
Referring to Fig. 1, can be applied to server the embodiment of the invention provides a kind of method for refreshing of recommendation information and set
It is standby.Server apparatus may include computer, microcomputer etc..The present embodiment includes step S100 to step S400, specific as follows:
S100, the refresh requests at respond request end extract recommendation information from recommendation information set and are combined into initial information
Sequence；Wherein, initial information sequence includes M recommendation information.
In the present embodiment, request end may include smart phone, plate, computer etc..Request end can install and take
The client application for device system matches of being engaged in is requested for sending to server apparatus, and shows the information received.Client
Holding application program may include the APP such as Baidu, Google, Taobao, Jingdone district (Application, application program).Recommendation information can be with
Including commodity favor information, the search information of search engine, news information or the article in each field etc..If refresh requests are to use
Family request refreshes the retrieval page of product, then recommendation information may include the description information of product, such as picture, the valence of commodity
Lattice, producer etc..If refresh requests are that user requests to refresh the retrieval page of a certain retrieval sentence, recommendation information may include
Article relevant to the retrieval sentence or Domestic News etc..Initial information sequence can be believed by the user of recommendation information and request end
The degree of correlation of breath or the prediction clicking rate of recommendation information are arranged.The length of initial information sequence is initial information sequence packet
The quantity of the recommendation information contained.Compared to the length of the recommendation information sequence sent to request end, the length of initial information sequence compared with
It is long.
S200 clicks probability according to the prediction of recommendation information each in initial information sequence, adjusts the row of initial information sequence
Column sequence, obtains preference information sequence.
In the present embodiment, since putting in order for initial information sequence only considers each recommendation information in recommendation information set
In weighted value, be frequently not optimal sequence, it is then desired to consider sequence integrally to the influence of user, such as prediction user
To the click probability of each recommendation information.
S300 extracts recommendation information sequence from preference information sequence, and recommendation information sequence includes in preference information sequence
It sorts in the recommendation information of top n, wherein M is greater than N.
In the present embodiment, the numerical value of M can be the integral multiple of N.For example, by taking recommendation information is article as an example, if right
Sending N number of article in any refresh requests, then preference information sequence or initial information sequence may include M=N*S article,
Wherein, it is the following refreshing frequency in need of consideration including this refreshing that S, which is greater than 1, S,.If N is numerical value 5, M is numerical value 20,
Then recommendation information sequence is to sort in preference information sequence in preceding 5 recommendation informations.
S400 sends recommendation information sequence to request end, is shown with carrying out refreshing in request end.
In the present embodiment, during this refreshing issues recommendation information sequence, pre-generated one longer first
Beginning sequence (initial information sequence).Then putting in order in initiation sequence is adjusted, obtains preferred sequence (preference information
Sequence).Finally, extracting sequence from preferred sequence in top n recommendation information, as the recommendation information sequence issued.Due to excellent
Sequence length is selected to be longer than the sequence issued, so this refreshes issued sequence it is contemplated that subsequent refresh issued sequence
Column.
In one possible implementation, as shown in Fig. 2, the generation of the initial information sequence of above-mentioned steps S100
Journey may include:
S110 searches corresponding each recommendation information from recommendation information library, obtains according to the user information in refresh requests
Recommendation information set.
In the present embodiment, the refresh requests that request end is sent would generally carry user information.In order to make recommendation information more
Letter can be extracted from recommendation information library in conjunction with user information while meeting refresh requests for the demand for matching user
Breath.If same user continuously refreshes repeatedly in the short time, when this refreshes and searches recommendation information, this user can be existed
Refreshed searched recommendation information in certain time in past to exclude.
S120 calculates the recommendation weighted value of each recommendation information in recommendation information set according to preset proposed algorithm.
In the present embodiment, proposed algorithm may include method, method neural network based etc. based on matrix decomposition.
Use proposed algorithm can with the click probability of each recommendation information in set of computations or with the degree of correlation of user information.Advowson
Weight values may rely on the degree of correlation etc. for clicking probability and user information of recommendation information.For example, the click of recommendation information is general
Rate is higher, recommends weighted value higher；The degree of correlation of user information is higher, recommends weighted value higher.
S130, by each recommendation information in recommendation information set according to recommendation weighted value sequence.
S140 chooses sequence in preceding M recommendation informations, is combined into initial information sequence.
Exemplary, if the length of recommendation information sequence is N, the refreshing frequency of the system setting of server is S, and S is greater than
1, then it can choose the recommendation information to be sorted in recommendation information set at preceding N*S.At this point, M=N*S.
The sequence of initial information sequence is based only upon recommendation weighted value, and the recommendation weighted value of each recommendation information is all based on certainly
Body recommendation information is calculated, and is failed to consider the connection between recommendation information, that is, is failed to consider to be formed by multiple recommendation informations
Influence of the sequence to user.Thus, it is not necessarily that this refreshes optimal sequence if initial information sequence is handed down to user
Column, and sequence is too long.For this purpose, current embodiment require that obtain preference information sequence based on initial information sequence, and from excellent
It selects and intercepts the sequence of appropriate length in information sequence as the sequence for being handed down to request end.
In one possible implementation, as shown in figure 3, the acquisition of the preference information sequence of above-mentioned steps S200
Journey may include:
S210, it is random to adjust putting in order for recommendation information in initial information sequence, obtain candidate information sequence.
In the present embodiment, in initial information sequence recommendation information put in order be adjusted sequence obtained with
Initial sequence is close but difference, can be convenient and subsequent therefrom finds preferred sequence.It specifically, can be to initial information sequence
The arrangement position of recommendation information carries out mutation operation in column, and mutation operation includes randomly selecting two from initial information sequence to push away
It recommends information and carries out place-exchange.And one or many mutation operations are executed, the available one new sequence different from initiation sequence
Column.It is multiple to execute the above process, and excludes identical sequence, available multiple sequences form candidate information arrangement set.
It is exemplary, if initial information sequence is { information A1, information A2, information A3, information A4, the letter being arranged in order
Cease A5 }, random selection information carries out place-exchange.For example, information A1 and information A3 carries out place-exchange.One can be executed at random
It is secondary or multiple, for example, (carrying out place-exchange for the first time twice for information A1 and information A3, second is information A3 and information A5
Carry out place-exchange).By the example above, can obtain candidate information sequence is { information A5, information A2, the letter being arranged in order
Cease A1, information A4, information A3 }.
S220 clicks probability by the prediction that sequence estimation model calculates each recommendation information of candidate information sequence.
In the present embodiment, sequence estimation model can be built-up based on neural network, such as: RNN (Recurrent
Neural Network, Recognition with Recurrent Neural Network), the deformation of RNN network etc..As shown in figure 4, sequence estimation model can use
The mode of RNN is modeled.Neurode in figure is chronologically successively unfolded from left to right, and x is input layer, and h is hidden layer, and y is
Output layer.ht-n... ..., ht-2, ht-1, htIndicate the neurode of hidden state, xt-n,……,xt-2,xt-1,xtIt indicates in sequence
The feature of each recommendation information, yt-n,……,yt-2,yt-1,ytIndicate the probability that whether each recommendation information can be clicked in sequence.
It in the present embodiment, can preparatory training sequence assessment models.For example, can using BPTT (back-propagation algorithm,
Back Propagation Trough Time) carry out model training.Generally, the training process of model may include: to obtain
The training data of sequence estimation model；Wherein, training data includes the sample recommendation information sequence and sample recommendation of request end
Cease the click probability of each recommendation information in sequence.For example, if the recommendation information in sample is clicked in samples show process,
The click probability of this recommendation information is 1；If the recommendation information in sample shows that process is not clicked in sample screen, this recommendation information
Click probability be 0.Then, by the generation time of sample recommendation information sequence, training data is sequentially input into sequence estimation mould
Type, to be trained update to sequence estimation model.It can be noticed that the hidden layer of RNN includes between hidden state neural network
From connection weight W.Thus, when the feature of recommendation information is sequentially placed into learnt in RNN network when, RNN can lead to
Cross the update iteration of W, so that it may the relationship between study to adjacent recommendation information, to realize the modeling to sequence.
S230 clicks probability according to the prediction of each recommendation information of candidate information sequence, calculates pushing away for candidate information sequence
Recommend score.
In the present embodiment, the prediction of each recommendation information in sequence can be clicked probability to be added, obtains this sequence
The recommender score of column.Or each recommendation information in sequence assigns a weighted value.For example, the arrangement by recommendation information is suitable
Sequence assigns each recommendation information corresponding weighted value, and probability is clicked in the prediction of each recommendation information and is multiplied with the weighted value of itself,
And product is summed, obtain the recommender score of this sequence.
S240 chooses the highest candidate information sequence of recommender score as preferred letter from the candidate information sequence obtained
Cease sequence.
In the present embodiment, for the recommendation information sequence arbitrarily inputted, using sequence estimation model to this sequence
Each of column recommendation information predicts a click probability.Due in RNN e-learning sequence between recommendation information
Relationship, and later use sequence estimation model calculates obtained click probability to each recommendation information, carrys out sequence calculating thus
One recommender score, thus this recommender score is it can be considered that connection between recommendation information in sequence.In turn, according to recommended hour
The sequence that number is chosen from candidate information sequence is often better than initial information sequence.
Referring to figs. 5 and 6, Fig. 5 is a kind of method for refreshing for recommending article provided in an embodiment of the present invention using exemplary
Flow chart.Fig. 6 is the difference of the method for refreshing of the method for refreshing provided in an embodiment of the present invention for recommending article and the prior art
Schematic diagram.
The exemplary basic principle of this application is: the article collection that server-side can all be recalled in each refresh from article library
It closes, is issued with therefrom extracting article and forming article sequence (list).It should be noted that also may be used in subsequent refreshing
It can recall, shoot off this and refresh after issued article remaining article in article set.It is especially continuous within a short period of time to brush
Under news.Because user draws a portrait variation less in a short time, the change in resources in article library is little, remains in article set
Under article a possibility that being called back it is larger.Therefore, when this refreshes generation and issues sequence, if this is refreshed not
The article issued takes into account, and it is whole that this refreshing can be made, which to issue sequence and issue together with sequence assembly with subsequent refreshing,
Body optimal sequence.Wherein, a feasible program is: issuing the process of article sequence in this refreshing, generates a longer sequence
Column, and this refreshing only issues and is arranged in top n article in this sequence.In this way, this refreshing issue with regard to article sequence consider
It is following to refresh issued article sequence, so that when this refreshes issued sequence and following the issued sequence assembly of refreshing exists
This sequence is total optimization sequence when together.
Referring to Fig. 5, this application example includes 4 steps, specific as follows:
(1) basis recalls article set and generates initiation sequence.
The main purpose of this step is to generate a preferably initiation sequence according to the article set recalled.
The specific method is as follows:
1. using any one proposed algorithm, a weight w is calculated separately to the every article recalled from article library, is weighed
The probability that the user information degree of correlation or this article for being reused in expression and request end are clicked by user.
2. sorting from large to small by every article weight w, and sequence is chosen in preceding m*n article.Wherein, n indicates primary
Refresh the quantity for issuing article；M > 1 indicates the refreshing number for needing to be considered in total.
It, can be using a variety of classical proposed algorithms for calculating the proposed algorithm of weight w.For example, based on matrix decomposition
Method, method neural network based etc..
(2) a variety of combined sequences are generated at random according to initiation sequence
In the present embodiment, initiation sequence but is not optimal often preferably.Because article is arranged in initiation sequence
Column sequence only considers the weight of article, and the calculating of the weight of article is only the information based on article itself, does not consider and other texts
Relationship between chapter.That is, the initiation sequence formed fails to consider the article between sequence.The main purpose of this step is according to just
Beginning sequence generates an arrangement set close but differentiated with initiation sequence, therefrom to find more preferably sequence.
The specific method is as follows:
For a sequence, the present embodiment defines a kind of mutation operation: selecting position i and the j (i in two sequences at random
≠ j), the article of i-th of position and j-th of position is exchanged.For initiation sequence, x times (x is a random number) can be carried out and become
ETTHER-OR operation, to obtain a new sequence.The present embodiment execution above process is multiple, obtains multiple and different candidate sequences.
(3) optimal sequence is chosen according to assessment models
The main purpose of this step is that optimal sequence is selected from multiple and different candidate sequences.According to sequence estimation mould
Type gives a mark to the article in each candidate sequence, and the progress for then calculating each candidate sequence according to this marking numerical value is again beaten
Point, the sequence of highest scoring is optimal sequence.
Optionally, assessment models can be modeled by the way of RNN, can be as shown in Figure 4.Neurode in figure
Chronologically successively it is unfolded from left to right, x is input layer, and h is hidden layer, and y is output layer.ht-n... ..., ht-2, ht-1, htIndicate hidden
The neurode of state, xt-n,……,xt-2,xt-1,xtIndicate the feature of each article in sequence, yt-n,……,yt-2,yt-1,ytTable
Show the probability that whether each article can be clicked in sequence.
In this example, RNN network can be trained in advance.It can use a large amount of user and request the page article number refreshed
According to as training sample.Each sample includes that the click of each article in the article sequence and sequence shown to user is general
Rate.Wherein, the article clicked for the displaying process in article sequence by user, clicking probability is 1；For in article sequence
Displaying process not by user click article, click probability be 0.The training process of RNN network can be using classical RNN
Training method, such as BPTT etc..
In forecast period, for each candidate sequence of input, assessment models can predict every text in this sequence
The click probability of chapter.Then, server-side will predict in this candidate sequence carrys out all click probability and is added, and obtains this candidate
Overall marking in sequence.
It can be noticed that the hidden layer of RNN includes between hidden state neural network from connection weight W, it can be by sequence
In the feature of adjacent article connect.Thus, learn when the feature of recommendation information to be sequentially placed into RNN network
When, RNN can learn the connection between adjacent article by the update iteration of W.When predicting the click probability of article, no
The feature of article itself is based only on independently to predict the click probability of this article, but other articles in meeting contact sequence
Feature predicts the click probability of this article.So, the overall marking of candidate article sequence can comprehensively consider each article itself
Feature and each article between connection.In this way, according to the sequence that the overall marking of sequence is chosen from candidate article sequence, it is past
It is past to be better than initial article sequence.
(4) this refreshes article and issues
In this step, interception is arranged in front sequence corresponding to n article from the optimal article sequence of selection, as
This refreshes the article sequence issued.
In practical applications, when refresh next time, operation as the aforementioned can be still executed, obtains new article sequence
To be issued.
As shown in fig. 6, the new method that this application example provides is compared with original method, it is known that: this application example
In the article sequence that this refreshing issues, it is contemplated that future refreshes the article sequence that may be issued, and shows this brush to user
It is still total optimization sequence when refreshing and being combined together display with the following refreshing this while new optimal sequence.In this way,
So that continuing to refresh in refresh process shown content in user more attracts user, user obtains better viewing experience.
Referring to Fig. 7, the embodiment of the present invention also provides a kind of refreshing apparatus of recommendation information, comprising:
Initiation sequence generation module 100 extracts from recommendation information set and recommends for the refresh requests at respond request end
Information is combined into initial information sequence；Wherein, the initial information sequence includes M recommendation information；
Preferred sequence generation module 200 is clicked general for the prediction according to each recommendation information in the initial information sequence
Rate adjusts putting in order for the initial information sequence, obtains preference information sequence；
Recommend sequential extraction procedures module 300, for extracting recommendation information sequence, the recommendation from the preference information sequence
Information sequence includes being sorted in the preference information sequence in the recommendation information of top n, wherein M is greater than N；And
Sequence sending module 400, for sending the recommendation information sequence to the request end, with the request end into
Row refreshes display.
In one possible implementation, referring to Fig. 8, the initiation sequence generation module 100 may include:
Recommendation information searching unit 110, for being looked into from recommendation information library according to the user information in the refresh requests
Corresponding each recommendation information is looked for, recommendation information set is obtained；
Information weight calculation unit 120, for calculating each in the recommendation information set according to preset proposed algorithm
The recommendation weighted value of recommendation information；And
Recommendation information sequencing unit 130, for by each recommendation information in the recommendation information set according to recommend weight
Value sequence；And
Recommendation information selection unit 140 is combined into initial information sequence for choosing sequence in preceding M recommendation informations.
In one possible implementation, referring to Fig. 9, the preferred sequence generation module 200 may include:
Candidate sequence generation unit 210, for adjusting putting in order for recommendation information in the initial information sequence at random,
Obtain candidate information sequence；
Probability calculation unit 220 is clicked, for calculating each recommendation of the candidate information sequence by sequence estimation model
Probability is clicked in the prediction of information；
Recommender score computing unit 230, the prediction for each recommendation information according to the candidate information sequence are clicked generally
Rate calculates the recommender score of the candidate information sequence；And
Sequence selection unit 240, for choosing the highest candidate information of recommender score from the candidate information sequence obtained
Sequence is as preference information sequence.
In one possible implementation, the candidate sequence generation unit 210 is specifically used for: to the initial information
The arrangement position of recommendation information carries out mutation operation in sequence, obtains candidate information sequence；Wherein, the mutation operation include from
Two recommendation informations are randomly selected in the initial information sequence carries out place-exchange.
In one possible implementation, referring to Figure 10, the refreshing apparatus can also include:
Training data module 500, for obtaining the training data of the sequence estimation model；Wherein, the training data
The click probability of each recommendation information in sample recommendation information sequence and the sample recommendation information sequence including the request end；
And
Model training module 600, it is for pressing the generation time of the sample recommendation information sequence, the training data is defeated
Enter the sequence estimation model, to be trained update to the sequence estimation model.
The function of described device can also execute corresponding software realization by hardware realization by hardware.It is described
Hardware or software include one or more modules corresponding with above-mentioned function.
The function of described device can also execute corresponding software realization by hardware realization by hardware.It is described
Hardware or software include one or more modules corresponding with above-mentioned function.
It include processor and memory, the memory in the refreshing structure of recommendation information in a possible design
Refreshing apparatus for recommendation information executes the refurbishing procedure of recommendation information in above-mentioned first aspect, the processor is configured to
For executing the program stored in the memory.The refreshing apparatus of the recommendation information can also include communication interface, be used for
The refreshing apparatus and other equipment or communication of recommendation information.
The embodiment of the present invention also provides a kind of refreshing terminal device of recommendation information, and as shown in figure 11, which includes: to deposit
Reservoir 21 and processor 22, being stored in memory 21 can be in the computer program on processor 22.Processor 22 executes calculating
The method for refreshing of the recommendation information in above-described embodiment is realized when machine program.The quantity of memory 21 and processor 22 can be one
It is a or multiple.
The equipment further include:
Communication interface 23, for the communication between processor 22 and external equipment.
Memory 21 may include high speed RAM memory, it is also possible to further include nonvolatile memory (non-volatile
Memory), a for example, at least magnetic disk storage.
If memory 21, processor 22 and the independent realization of communication interface 23, memory 21, processor 22 and communication are connect
Mouth 23 can be connected with each other by bus and complete mutual communication.Bus can be industry standard architecture (ISA,
Industry Standard Architecture) bus, external equipment interconnection (PCI, Peripheral Component) be total
Line or extended industry-standard architecture (EISA, Extended Industry Standard Component) bus etc..Always
Line can be divided into address bus, data/address bus, control bus etc..Only to be indicated with a thick line in Figure 11, but simultaneously convenient for indicating
Only a bus or a type of bus are not indicated.
Optionally, in specific implementation, if memory 21, processor 22 and communication interface 23 are integrated in chip piece
On, then memory 21, processor 22 and communication interface 23 can complete mutual communication by internal interface.
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 included at least one embodiment or example of the invention.Moreover, particular features, structures, materials, or characteristics described
It may be combined in any suitable manner in any one or more of the embodiments or examples.In addition, without conflicting with each other, this
The technical staff in field can be by the spy of different embodiments or examples described in this specification and different embodiments or examples
Sign is combined.
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." first " is defined as a result, the feature of " second " can be expressed or hidden
It include at least one this feature containing ground.In the description of the present invention, the meaning of " plurality " is two or more, unless otherwise
Clear specific restriction.
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 specific logical function or process the step of executable instruction code module, segment or portion
Point, and the range of the preferred embodiment of the present invention 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 of the invention
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 computer-readable medium of the embodiment of the present invention can be computer-readable signal media or computer-readable deposit
Storage media either the two any combination.The more specific example at least (non-exclusive of computer readable storage medium
List) include the following: there is the electrical connection section (electronic device) of one or more wirings, portable computer diskette box (magnetic dress
Set), random access memory (RAM), read-only memory (ROM), erasable edit read-only storage (deposit by EPROM or flash
Reservoir), fiber device and portable read-only memory (CDROM).In addition, computer readable storage medium can even is that
Can the paper of print routine or other suitable media on it because can for example be swept by carrying out optics to paper or other media
It retouches, is then edited, interprets or handled when necessary with other suitable methods electronically to obtain program, then will
It is stored in computer storage.
In embodiments of the present invention, computer-readable signal media may include in a base band or as carrier wave a part
The data-signal of propagation, wherein carrying computer-readable program code.The data-signal of this propagation can use a variety of
Form, including but not limited to electromagnetic signal, optical signal or above-mentioned any appropriate combination.Computer-readable signal media is also
It can be any computer-readable medium other than computer readable storage medium, which can send, pass
It broadcasts or transmits for instruction execution system, input method or device use or program in connection.Computer can
The program code for reading to include on medium can transmit with any suitable medium, including but not limited to: wirelessly, electric wire, optical cable, penetrate
Frequently (Radio Frequency, RF) etc. or above-mentioned any appropriate combination.
It should be appreciated that each section of the invention 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.It, and in another embodiment, can be under well known in the art for example, if realized with hardware
Any one of column technology or their combination are realized: having a logic gates for realizing logic function to data-signal
Discrete logic, with suitable combinational logic gate circuit specific integrated circuit, programmable gate array (PGA), scene
Programmable 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 the program that relevant hardware can be instructed to complete by program, which can store in a kind of computer-readable storage
In medium, which when being executed, includes the steps that one or a combination set of embodiment of the method.
It, can also be in addition, each functional unit in each embodiment of the present invention can integrate in a processing module
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.If integrated module with
The form of software function module is realized and when sold or used as an independent product, also can store computer-readable at one
In storage medium.Storage medium can be read-only memory, disk or CD etc..
More than, only a specific embodiment of the invention, but scope of protection of the present invention is not limited thereto, and it is any to be familiar with
Those skilled in the art in the technical scope disclosed by the present invention, can readily occur in its various change or replacement, these
It should be covered by the protection scope of the present invention.Therefore, protection scope of the present invention should be with scope of protection of the claims
It is quasi-.
Claims (12)
1. a kind of method for refreshing of recommendation information characterized by comprising
The refresh requests at respond request end extract recommendation information from recommendation information set and are combined into initial information sequence；Wherein,
The initial information sequence includes M recommendation information；
Probability is clicked according to the prediction of each recommendation information in the initial information sequence, adjusts the arrangement of the initial information sequence
Sequentially, preference information sequence is obtained；
Recommendation information sequence is extracted from the preference information sequence, the recommendation information sequence includes the preference information sequence
Recommendation information of the middle sequence in top n, wherein M is greater than N；And
The recommendation information sequence is sent to the request end, is shown with carrying out refreshing in the request end.
2. the method for refreshing of recommendation information as described in claim 1, which is characterized in that the refreshing at the respond request end is asked
It asks, recommendation information is extracted from recommendation information set and is combined into initial information sequence, comprising:
According to the user information in the refresh requests, corresponding each recommendation information is searched from recommendation information library, is recommended
Information aggregate；
According to preset proposed algorithm, the recommendation weighted value of each recommendation information in the recommendation information set is calculated；And
By each recommendation information in the recommendation information set according to recommendation weighted value sequence；And
Sequence is chosen in preceding M recommendation informations, is combined into initial information sequence.
3. the method for refreshing of recommendation information as described in claim 1, which is characterized in that described according to the initial information sequence
In the prediction of each recommendation information click probability, adjust putting in order for the initial information sequence, obtain preference information sequence, packet
It includes:
Putting in order for recommendation information in the initial information sequence is adjusted at random, obtains candidate information sequence；
Probability is clicked by the prediction that sequence estimation model calculates each recommendation information of the candidate information sequence；
Probability is clicked according to the prediction of each recommendation information of the candidate information sequence, calculates the recommendation of the candidate information sequence
Score；And
From the candidate information sequence obtained, the highest candidate information sequence of recommender score is chosen as preference information sequence.
4. the method for refreshing of recommendation information as claimed in claim 3, which is characterized in that described to adjust the initial information at random
Recommendation information puts in order in sequence, obtains candidate information sequence, comprising:
Mutation operation is carried out to the arrangement position of recommendation information in the initial information sequence, obtains candidate information sequence；Wherein,
The mutation operation includes randomly selecting two recommendation informations from the initial information sequence to carry out place-exchange.
5. the method for refreshing of recommendation information as claimed in claim 3, which is characterized in that the method for refreshing further include:
Obtain the training data of the sequence estimation model；Wherein, the training data includes that the sample of the request end is recommended
The click probability of each recommendation information in information sequence and the sample recommendation information sequence；And
By the generation time of the sample recommendation information sequence, the training data is inputted into the sequence estimation model, with right
The sequence estimation model is trained update.
6. a kind of refreshing apparatus of recommendation information characterized by comprising
Initiation sequence generation module extracts recommendation information group for the refresh requests at respond request end from recommendation information set
Synthesize initial information sequence；Wherein, the initial information sequence includes M recommendation information；
Preferred sequence generation module clicks probability, adjustment for the prediction according to each recommendation information in the initial information sequence
The initial information sequence puts in order, and obtains preference information sequence；
Recommend sequential extraction procedures module, for extracting recommendation information sequence, the recommendation information sequence from the preference information sequence
Column include being sorted in the preference information sequence in the recommendation information of top n, wherein M is greater than N；And
Sequence sending module, for sending the recommendation information sequence to the request end, to be refreshed in the request end
Display.
7. the refreshing apparatus of recommendation information as claimed in claim 6, which is characterized in that the initiation sequence generation module packet
It includes:
Recommendation information searching unit, for searching and corresponding to from recommendation information library according to the user information in the refresh requests
Each recommendation information, obtain recommendation information set；
Information weight calculation unit, for calculating each recommendation in the recommendation information set according to preset proposed algorithm
The recommendation weighted value of breath；And
Recommendation information sequencing unit, for each recommendation information in the recommendation information set to sort according to recommendation weighted value；
And
Recommendation information selection unit is combined into initial information sequence for choosing sequence in preceding M recommendation informations.
8. the refreshing apparatus of recommendation information as claimed in claim 6, which is characterized in that the preferred sequence generation module, packet
It includes:
Candidate sequence generation unit is waited for adjusting putting in order for recommendation information in the initial information sequence at random
Select information sequence；
Click probability calculation unit, for calculated by sequence estimation model the candidate information sequence each recommendation information it is pre-
Measuring point hits probability；
Recommender score computing unit, the prediction for each recommendation information according to the candidate information sequence are clicked probability, are calculated
The recommender score of the candidate information sequence；And
Sequence selection unit is made for from the candidate information sequence obtained, choosing the highest candidate information sequence of recommender score
For preference information sequence.
9. the refreshing apparatus of recommendation information as claimed in claim 8, which is characterized in that the candidate sequence generation unit is specific
For: mutation operation is carried out to the arrangement position of recommendation information in the initial information sequence, obtains candidate information sequence；Its
In, the mutation operation includes randomly selecting two recommendation informations from the initial information sequence to carry out place-exchange.
10. the refreshing apparatus of recommendation information as claimed in claim 8, which is characterized in that the refreshing apparatus further include:
Training data module, for obtaining the training data of the sequence estimation model；Wherein, the training data includes described
The click probability of each recommendation information in the sample recommendation information sequence of request end and the sample recommendation information sequence；And
Model training module will be described in training data input for pressing the generation time of the sample recommendation information sequence
Sequence estimation model, to be trained update to the sequence estimation model.
11. a kind of refreshing terminal device for realizing recommendation information, which is characterized in that the terminal device includes:
One or more processors；
Storage device, for storing one or more programs；
When one or more of programs are executed by one or more of processors, so that one or more of processors
Realize the method for refreshing such as recommendation information as claimed in any one of claims 1 to 5.
12. a kind of computer readable storage medium, is stored with computer program, which is characterized in that the program is held by processor
The method for refreshing such as recommendation information as claimed in any one of claims 1 to 5 is realized when row.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810846642.1A CN109063104B (en) | 2018-07-27 | 2018-07-27 | Recommendation information refreshing method and device, storage medium and terminal equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810846642.1A CN109063104B (en) | 2018-07-27 | 2018-07-27 | Recommendation information refreshing method and device, storage medium and terminal equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109063104A true CN109063104A (en) | 2018-12-21 |
| CN109063104B CN109063104B (en) | 2020-11-10 |
Family
ID=64836690
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810846642.1A Active CN109063104B (en) | 2018-07-27 | 2018-07-27 | Recommendation information refreshing method and device, storage medium and terminal equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109063104B (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109670117A (en) * | 2018-12-28 | 2019-04-23 | 北京百度网讯科技有限公司 | The recommended method and its device of information list |
| CN111382346A (en) * | 2018-12-28 | 2020-07-07 | 第四范式（北京）技术有限公司 | Method and system for recommending content |
| CN111800446A (en) * | 2019-04-12 | 2020-10-20 | 北京沃东天骏信息技术有限公司 | Scheduling processing method, device, equipment and storage medium |
| CN112732927A (en) * | 2020-12-31 | 2021-04-30 | 平安资产管理有限责任公司 | Content similarity analysis method and device based on knowledge graph |
| CN112860985A (en) * | 2019-11-27 | 2021-05-28 | 云闪信(珠海市横琴)信息科技有限公司 | Application navigation method and device, electronic equipment and computer readable storage medium |
| CN113190758A (en) * | 2021-05-21 | 2021-07-30 | 聚好看科技股份有限公司 | Server and media asset recommendation method |
| CN113590947A (en) * | 2021-07-27 | 2021-11-02 | 润联软件系统(深圳)有限公司 | Information recommendation method and device, computer equipment and storage medium |
| CN114880562A (en) * | 2022-05-10 | 2022-08-09 | 北京百度网讯科技有限公司 | Method and device for recommending information |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140229416A1 (en) * | 2013-02-13 | 2014-08-14 | Samsung Electronics Co., Ltd. | Electronic apparatus and method of recommending contents to members of a social network |
| CN104123284A (en) * | 2013-04-24 | 2014-10-29 | 华为技术有限公司 | Recommendation method and server |
| CN104899195A (en) * | 2014-01-26 | 2015-09-09 | 武汉联影医疗科技有限公司 | Customized educational resource recommending method and apparatus |
| CN107463704A (en) * | 2017-08-16 | 2017-12-12 | 北京百度网讯科技有限公司 | Searching method and device based on artificial intelligence |
| CN107977411A (en) * | 2017-11-21 | 2018-05-01 | 腾讯科技(成都)有限公司 | Group recommending method, device, storage medium and server |
-
2018
- 2018-07-27 CN CN201810846642.1A patent/CN109063104B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140229416A1 (en) * | 2013-02-13 | 2014-08-14 | Samsung Electronics Co., Ltd. | Electronic apparatus and method of recommending contents to members of a social network |
| CN104123284A (en) * | 2013-04-24 | 2014-10-29 | 华为技术有限公司 | Recommendation method and server |
| CN104899195A (en) * | 2014-01-26 | 2015-09-09 | 武汉联影医疗科技有限公司 | Customized educational resource recommending method and apparatus |
| CN107463704A (en) * | 2017-08-16 | 2017-12-12 | 北京百度网讯科技有限公司 | Searching method and device based on artificial intelligence |
| CN107977411A (en) * | 2017-11-21 | 2018-05-01 | 腾讯科技(成都)有限公司 | Group recommending method, device, storage medium and server |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109670117A (en) * | 2018-12-28 | 2019-04-23 | 北京百度网讯科技有限公司 | The recommended method and its device of information list |
| CN111382346A (en) * | 2018-12-28 | 2020-07-07 | 第四范式（北京）技术有限公司 | Method and system for recommending content |
| CN109670117B (en) * | 2018-12-28 | 2023-03-31 | 北京百度网讯科技有限公司 | Information list recommendation method and device |
| CN111382346B (en) * | 2018-12-28 | 2023-09-01 | 第四范式（北京）技术有限公司 | Method and system for recommending content |
| CN111800446A (en) * | 2019-04-12 | 2020-10-20 | 北京沃东天骏信息技术有限公司 | Scheduling processing method, device, equipment and storage medium |
| CN111800446B (en) * | 2019-04-12 | 2023-11-07 | 北京沃东天骏信息技术有限公司 | Scheduling processing method, device, equipment and storage medium |
| CN112860985A (en) * | 2019-11-27 | 2021-05-28 | 云闪信(珠海市横琴)信息科技有限公司 | Application navigation method and device, electronic equipment and computer readable storage medium |
| CN112732927A (en) * | 2020-12-31 | 2021-04-30 | 平安资产管理有限责任公司 | Content similarity analysis method and device based on knowledge graph |
| CN113190758A (en) * | 2021-05-21 | 2021-07-30 | 聚好看科技股份有限公司 | Server and media asset recommendation method |
| CN113190758B (en) * | 2021-05-21 | 2023-01-20 | 聚好看科技股份有限公司 | Server and media asset recommendation method |
| CN113590947A (en) * | 2021-07-27 | 2021-11-02 | 润联软件系统(深圳)有限公司 | Information recommendation method and device, computer equipment and storage medium |
| CN114880562A (en) * | 2022-05-10 | 2022-08-09 | 北京百度网讯科技有限公司 | Method and device for recommending information |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109063104B (en) | 2020-11-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109063104A (en) | Method for refreshing, device, storage medium and the terminal device of recommendation information | |
| CN108446374B (en) | User's Intention Anticipation method, apparatus, electronic equipment, storage medium | |
| CN110442790B (en) | Method, device, server and storage medium for recommending multimedia data | |
| CN108256093B (en) | Collaborative filtering recommendation algorithm based on multiple interests and interest changes of users | |
| CN110020094B (en) | Display method and related device for search results | |
| CN111046286B (en) | Object recommendation method and device and computer storage medium | |
| CN110543598B (en) | Information recommendation method and device and terminal | |
| CN109345302A (en) | Machine learning model training method, device, storage medium and computer equipment | |
| CN111382361B (en) | Information pushing method, device, storage medium and computer equipment | |
| CN109086439A (en) | Information recommendation method and device | |
| CN109460514A (en) | Method and apparatus for pushed information | |
| CN108122122A (en) | Advertisement placement method and system | |
| CN109598576A (en) | Service recommendation method, device and equipment | |
| CN111859149A (en) | Information recommendation method and device, electronic equipment and storage medium | |
| CN111767466A (en) | Recommendation information recommendation method and device based on artificial intelligence and electronic equipment | |
| CN111210258A (en) | Advertisement putting method and device, electronic equipment and readable storage medium | |
| CN108595493A (en) | Method for pushing and device, storage medium, the electronic device of media content | |
| CN108921624A (en) | advertisement fusion method, device, storage medium and terminal device | |
| CN112364184B (en) | Method, device, server and storage medium for ordering multimedia data | |
| KR20210062522A (en) | Control method, device and program of user participation keyword selection system | |
| CN112488781A (en) | Search recommendation method and device, electronic equipment and readable storage medium | |
| CN116108267A (en) | Recommendation method and related equipment | |
| CN112559777A (en) | Content item delivery method and device, computer equipment and storage medium | |
| CN108319612A (en) | Audient's Media Recommendation Method and system | |
| CN109146606A (en) | A kind of brand recommended method, electronic equipment, storage medium and system |
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 |