CN109241451A - A kind of content combined recommendation method, apparatus and readable storage medium storing program for executing - Google Patents
A kind of content combined recommendation method, apparatus and readable storage medium storing program for executing Download PDFInfo
- Publication number
- CN109241451A CN109241451A CN201811324196.4A CN201811324196A CN109241451A CN 109241451 A CN109241451 A CN 109241451A CN 201811324196 A CN201811324196 A CN 201811324196A CN 109241451 A CN109241451 A CN 109241451A
- Authority
- CN
- China
- Prior art keywords
- content
- recommendation
- combination
- collection
- feature
- 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)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention provides a kind of content combined recommendation method, apparatus and readable storage medium storing program for executing, are related to Internet technical field, and the recommended method includes: to obtain preparation recommendation collection and preset recommendation number；Multiple content combinations are calculated according to the prepared recommendation collection and recommendation number, wherein include the content of the recommendation number in each content combination；Determine that recommendation combines in the combination of the multiple content；Push the recommendation combination.It is combined to obtain multiple combined arrangements by the content according to acquisition, then determines recommendation combination from multiple combined arrangements, the content of displaying is become more comprehensively, to improve user experience.
Description
Technical field
The present invention relates to Internet technical field, in particular to a kind of content combined recommendation method, apparatus and readable
Storage medium.
Background technique
Currently, browser main interface only can be with exposition content, and the content that can be shown is only in browser recommender system
There are several, in order to attract more customer consumption contents in main interface, it is necessary to select most to close in many contents
The content for being applicable in family, capable of most attracting the particular user, and the prior art pass through merely single content proposed algorithm sequence after
It is combined to recommendation, lacks the diversity etc. of feature between content, can not comprehensively attract user.
Summary of the invention
In order to overcome the deficiencies in the prior art described above, the present invention provides a kind of content recommendation method and device, for changing
The kind above problem.
To achieve the goals above, technical solution provided by the embodiment of the present invention is as follows:
In a first aspect, the embodiment of the present invention provides a kind of content recommendation method and device, comprising: obtain preparation recommendation
Collection and preset recommendation number；Multiple content groups are calculated according to the prepared recommendation collection and recommendation number
It closes；It wherein, include the content of the recommendation number in each content combination；It is determined in the combination of the multiple content
Recommendation combination；Push the recommendation combination.
With reference to first aspect, in some possible implementations, the acquisition preparation recommendation collection, comprising: use
At least two content acquisition modes acquire the content in presetting database, obtain content corresponding with every kind of content acquisition mode
Collection；Content is extracted in the corresponding content set of every kind of content acquisition mode by preset rules, the preparation is obtained and recommends
Content set.
With reference to first aspect, described to be acquired using at least two content acquisition modes in some possible implementations
Content in presetting database obtains content set corresponding with every kind of content acquisition mode, comprising: uses and described at least two
The corresponding collection rule of content acquisition mode carries out relevancy ranking to the content in the database, obtains and every kind
The corresponding content set according to relevancy ranking of content acquisition mode；It is corresponding, it is described to be adopted by preset rules in every kind of content
Content is extracted in the corresponding content set of mode set, obtains the prepared recommendation collection, comprising: each content of acquisition
The content for concentrating the highest default selection number of the degree of correlation, obtains the prepared recommendation collection.
With reference to first aspect, described according to the prepared recommendation collection and recommendation in some possible implementations
Multiple content combinations are calculated in content number, comprising: take the recommendation number as the number of each content combination, at random
It combines each content that the prepared recommendation is concentrated and obtains multiple content combinations.
With reference to first aspect, in some possible implementations, the preparation recommendation collection described in the random combine
In each content obtain the combination of multiple contents before, the method also includes: the removal prepared recommendation concentrates phase
Same content.
With reference to first aspect, described to determine to recommend in the combination of the multiple content in some possible implementations
Content combination, comprising: obtain the preset content feature set of each content in each content combination；Merge each described interior
The preset content feature set for holding each content in combination obtains the assemblage characteristic collection of each content combination；It obtains
Take the judgement feature set at family, the feature that the type for the feature that the assemblage characteristic collection includes and the judgement feature set include
Type includes: article correlated characteristic, user's habit at least one of feature and user's scene characteristic feature or various features
Combination；The judgement feature set and each assemblage characteristic collection are compared, determine in the combination of the multiple content described pushes away
Recommend content combination.
With reference to first aspect, in some possible implementations, the judgement feature set for obtaining user, comprising: obtain
Take the historical content combination shown to user；It is combined according to the historical content, determines the historical content combination of user's selection
For positive example determine collect, determine the non-selected historical content of user be negative example determine collection；The judgement feature set includes the positive example
Determine that collection and the negative example determine collection.
With reference to first aspect, in some possible implementations, it is described compare the judgements feature set and it is each described in
Assemblage characteristic collection determines the recommendation combination in the multiple content combination, comprising: collects the positive example and determines to concentrate
The feature set of the content combination collects the feature set that the negative example determines the combination of content described in collection as feature set is positively correlated
As negatively correlated feature set；According to the positive correlation feature set and the negatively correlated feature set to preset initial association relationship mould
Type is trained, and generates association relation model；Corresponding combination is combined according to the association relation model and the multiple content
Feature set determines the recommendation combination in the multiple content combination.
Second aspect, the embodiment of the present invention also provide a kind of content combined recommendation, and device includes: acquiring unit, place
Manage unit and transmission unit.Acquiring unit is for obtaining prepared recommendation collection and preset recommendation number；Processing unit
It is combined for multiple contents to be calculated according to the prepared recommendation collection and recommendation number, and in the multiple content
Combination determines that recommendation combines；The transmission unit is used to push the recommendation combination to user.
The third aspect, the embodiment of the present invention also provide a kind of readable storage medium storing program for executing, calculating are stored in readable storage medium storing program for executing
Machine program, when computer program is run on computers, so that any in computer execution first aspect or first aspect
Content combined recommendation method described in embodiment.
The beneficial effect comprise that
It is combined to obtain multiple content groups by obtaining prepared recommendation collection, then by the content that preliminary matter is concentrated
Close, then obtain recommendation combination from the combination of multiple contents, compared to single content comparative sorting now, can select
Better content combined strategy, so that the content combination recommended more attracts user.
To make above-mentioned purpose of the invention, feature and can a little be clearer and more comprehensible, the embodiment of the present invention is hereafter enumerated, and match
Appended attached drawing is closed, is elaborated.
Detailed description of the invention
In order to illustrate the technical solution of the embodiments of the present invention more clearly, below will be to needed in the embodiment attached
Figure is briefly described, it should be understood that the following drawings illustrates only certain embodiments of the present invention, therefore is not construed as pair
The restriction of range for those of ordinary skill in the art without creative efforts, can also be according to this
A little attached drawings obtain other relevant attached drawings.
Fig. 1 is a kind of flow diagram for content combined recommendation method that the embodiment of the present invention 1 provides；
Fig. 2 is a kind of flow diagram realized for the acquisition preparation recommendation collection that the embodiment of the present invention 1 provides；
Fig. 3 is the flow diagram for the determination recommendation combination that the embodiment of the present invention 1 provides；
Fig. 4 is a kind of functional block diagram for content combined recommendation that the embodiment of the present invention 2 provides；
Fig. 5 is the structural block diagram for a kind of electronic equipment that the embodiment of the present invention 3 provides.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is
A part of the embodiment of the present invention, instead of all the embodiments.The present invention being usually described and illustrated herein in the accompanying drawings is implemented
The component of example can be arranged and be designed with a variety of different configurations.
Therefore, the detailed description of the embodiment of the present invention provided in the accompanying drawings is not intended to limit below claimed
The scope of the present invention, but be merely representative of selected embodiment of the invention.Based on the embodiments of the present invention, this field is common
Technical staff's every other embodiment obtained without creative efforts belongs to the model that the present invention protects
It encloses.
It should be noted that in the absence of conflict, the feature in embodiment and embodiment in the present invention can phase
Mutually combination.
It should also be noted that similar label and letter indicate similar terms in following attached drawing, therefore, once a certain Xiang Yi
It is defined in a attached drawing, does not then need that it is further defined and explained in subsequent attached drawing.
In the description of the embodiment of the present invention, it should be noted that indicating position or positional relationship is based on shown in attached drawings
The orientation or positional relationship invention product using when the orientation or positional relationship usually put or this field
Orientation or positional relationship that technical staff usually understands or the invention product using when the orientation usually put or position close
System, is merely for convenience of description of the present invention and simplification of the description, rather than the device or element of indication or suggestion meaning must have
Specific orientation is constructed and operated in a specific orientation, therefore is not considered as limiting the invention.
Embodiment 1
Referring to FIG. 1, Fig. 1 is the flow diagram for the combined arrangement recommended method that the embodiment of the present invention proposes.In described
Holding proposed algorithm includes:
Step S101: preparation recommendation collection and preset recommendation number are obtained；
Step S102: multiple contents are calculated according to the prepared recommendation collection and recommendation number and are combined；
Step S103: determine that recommendation combines in the combination of the multiple content；
Step S104: the recommendation combination is pushed.
It should be noted that at present terminal there are two kinds of forms of computer end and mobile phone terminal, due to the two size completely not
Together, when user is using different terminal browsing pages or rests on using main interface, the terminal that can be used according to user
Corresponding suitable recommendation number is preset, so that including suitable number of content in the combination of each content.
It is combined to obtain multiple content groups by obtaining prepared recommendation collection, then by the content that preliminary matter is concentrated
It closes, then obtains recommendation combination from certain amount combined strategy obtained, arranged compared to compared with now single content
Sequence can select better content combined strategy, can select so that the content combination recommended more attracts user.
It will be described in detail the realization process of each step of the combined arrangement recommended method of proposition of the embodiment of the present invention below.
Referring to FIG. 2, Fig. 2 is a kind of process realized for the acquisition preparation recommendation collection that the embodiment of the present invention 1 provides
Schematic diagram.
Optionally, step S101 includes:
Step S201: acquiring the content in presetting database using at least two content acquisition modes, obtain with every kind in
Hold the corresponding content set of acquisition mode；
Step S202: content is extracted in the corresponding content set of every kind of content acquisition mode by preset rules, is obtained
To the prepared recommendation collection.
In the present embodiment, made using three kinds of collaborative filtering based on user, topic model and correlation rule modes of recalling
Gathering algorithm for content acquisition mode, three kinds of content acquisition modes is different, and such as: the collaborative filtering based on user is will be with specific
User is the theme, it is of interest that the social property of the user, that is, more emphasize with the user have similar hobby other
The commending contents that user likes give the user；And the algorithm of topic model pays close attention to the theme of each article content, for specific
The theme that user likes carries out categorizing selection；Since three kinds of modes are all contents well-known to those skilled in the art, so this
Embodiment repeats no more this.Three kinds of content acquisition modes are respectively adopted, according to every from the database of preset recommendation
The feature of kind acquisition mode concern carries out classifying content or extracts a certain amount of content as a content set, to obtain three
The not exactly the same content set by sequence.In other embodiments, it can also be calculated using the collaborative filtering such as based on content
Method etc., the embodiment of the present invention is without limitation.
And the present invention also propose step S201 acquire presetting database in content during, by content ordering so as to
Prepared recommendation collection is obtained in step S202, extracting different content sets.
Optionally, step S201 includes: to use collection rule corresponding at least two content acquisitions mode,
Relevancy ranking is carried out to the content in the database, is obtained corresponding according to relevancy ranking with every kind of content acquisition mode
Content set；
Corresponding, after obtaining the content set, step S202 includes: degree of correlation highest in each content set of acquisition
Default selection number content, obtain the prepared recommendation collection.
It should be noted that the process for acquiring content can be understood as primary correlation in multiple content acquisition modes
Degree calculates the process that compares, when the higher content of the degree of correlation is calculated in the way of such as keyword, after calculating
Relevancy ranking, it can obtain in every kind of acquisition mode, according to the degree of correlation sort after result extract content, need
It is bright, in the present embodiment, highest two articles of the degree of correlation obtained under every kind of acquisition mode are selected, in other embodiments
In, it also can choose the degree of correlation highest one or three or more article etc., the present embodiment is without limitation.
In the present embodiment, after obtaining the different content set of concern feature using plurality of kinds of contents acquisition mode, in conjunction with every
The high content of the kind collected degree of correlation of content acquisition mode forms preparation recommendation collection.It in other embodiments, can be with
Classifying content is first carried out, further according to the classification of user's selection, using the content under the multiple hobby classifications of user as content set, in turn
Obtain different content sets.
After getting prepared recommendation collection, step S102 is executed, i.e., is combined the content that preparation recommendation is concentrated
Get up to obtain multiple content combinations.
Optionally, step S102 includes: the number with the recommendation number for the combination of each content, random combine institute
It states each content that prepared recommendation is concentrated and obtains multiple content combinations.
Illustrate the random combine: assuming that by three kinds of content acquisition modes collect three content sets A, B and
C extracts 3 contents from each content set and obtains the prepared recommendation collection:
[A1, A2, A3, B1, B2, B3, C1, C2, C3]
Above-mentioned nine contents are done into C9 3Random combine to get to most possibilities content combine, so as to be user
Determine the recommendation combination being more suitable, user more likes.
In every kind of content acquisition mode, due to it is preset be same content data base, although every kind of content acquisition side
The feature of formula pays close attention to difference, but each content set may choose identical content, such as: content A1With content B1Content it is special
It levies identical or for same piece article.
Therefore, each content that preparation recommendation is concentrated described in random combine obtains multiple content combinations: removal
The prepared recommendation concentrates identical content, then each content that preparation recommendation is concentrated described in random combine obtains
Multiple content combinations.
Prepared recommendation collection obtained above is done into re-scheduling processing, it is identical or for same piece article etc. to remove content
Content removal prevents in the content combination of recommendation and two identical contents occurs.
Referring to FIG. 3, Fig. 3 is the flow diagram for the determination recommendation combination that the embodiment of the present invention 1 provides.Further
Ground needs to determine that recommendation is combined according to certain rule after obtaining multiple content combinations.
Optionally, step S103 includes:
Step S301: the preset content feature set of each content in each content combination is obtained；
Step S302: merge the preset content feature set of each content in each content combination, obtain
The assemblage characteristic collection of each content combination；
Step S303: the judgement feature set of user, the type for the feature that the preset content feature set includes and institute are obtained
The type for stating the feature for determining that feature set includes is identical；
Step S304: comparing the judgement feature set and each assemblage characteristic collection, and the recommendation is calculated in determination
Content combination.
In existing machine learning, the algorithm model of Ordering and marking has: (Logistic Regression, is patrolled LR model
Volume regression model), GBDT model (Gradient Boosting Decision Tree, gradient promote decision-tree model) etc.,
In the present embodiment, given a mark using GBDT model to the combination of each content, and in GBDT model, select the judgement of user special
The judgement feature of user in collection as each marking node, then the assemblage characteristic collection that combines each content import model into
Row marking sequence, to obtain the recommendation combination.
It should be noted that the type of the feature includes content aspect, customer-side, scene aspect and equipment aspect
Deng, such as: include article's style in terms of content: education, medical treatment, customer-side include: gender, hobby, occupation etc., scene aspect
It include: current time, working day or festivals or holidays etc.；Equipment aspect includes: Android or apple, computer end or mobile phone terminal
Deng various features can be used in combination, and the present embodiment is without limitation.
Further, in the present embodiment, user determines the acquisition modes of feature set, i.e. step S303 includes:
It obtains and the historical content that user showed is combined；It is combined according to the historical content, determines going through for user's selection
History content combination be positive example determine collection, determine the non-selected historical content of user be negative example determine collection；The judgement feature set packet
It includes the positive example and determines that collection and the negative example determine collection.
In the past it is that the historical content that user showed combines by obtaining, as the training set of GBDT model, takes
The mode that positive and negative example judges jointly screens the combination of multiple contents in terms of two, also avoids selecting the content for enabling user dislike,
Improve the experience of user.In other embodiments, individually it can also be judged that the present embodiment is to this using positive example or negative example
With no restrictions.
It should be noted that often being trained if the historical information for only acquiring specific a certain user obtains determining feature set
Set content is very few, in other embodiments, user concern or the feature set of the user liked can also be acquired, by this feature collection
It is included in the training set of GBDT model, expands training set, further, in other embodiments, can also be expanded using other way
Big training set, the present embodiment are not particularly limited this.
After specific judgement feature set has been determined, the assemblage characteristic collection for combining each content is needed to substitute into GBDT model
Marking sequence is carried out, recommendation combination is obtained.
Optionally, step S304 includes:
It collects the positive example and determines that the feature set of the combination of content described in collection is used as positive correlation feature set, collect the negative example
Determine the feature set of the combination of content described in collection as negatively correlated feature set；According to the positive correlation feature set and the negative correlation
Feature set is trained preset initial association relational model, generates association relation model；According to the association relation model
And the multiple content combines corresponding assemblage characteristic collection, determines the recommendation combination in the multiple content combination.
Collection training is trained to initial GBDT model using positive and negative correlated characteristic collection, wherein positive and negative be with described
The output quantity of GBDT model optimizes GBDT mould using the feature that the positive and negative correlated characteristic is concentrated as the input quantity of GBDT model
Selection of the type to the assemblage characteristic collection in the combination of multiple contents, to obtain optimal content combined recommendation.
Optionally, the first implementation of step S304 are as follows: collect the positive example and determine the combination of content described in collection
Feature is as positive correlation feature, described in each single item feature and each single item for determining the assemblage characteristic concentration of each content combination
It is positively correlated whether feature is identical, and when to be, the content combines to obtain the first score.The honest score collects the negative example
Determine that the feature of the combination of content described in collection as negatively correlated feature, determines what the assemblage characteristic of each content combination was concentrated
Whether each single item feature and negative correlation feature described in each single item are identical, and when to be, the content combines to obtain the second score.Root
The final score value of corresponding content combination is calculated according to first score and second score of the combination of content described in each；Really
Determine the corresponding content group of final score value and is combined into recommendation combination.
Optionally, second of implementation of step S304 are as follows: collect the positive example and determine the combination of content described in collection
Feature set compares the assemblage characteristic collection and the positive correlation feature set of each content combination as positive correlation feature set,
Obtain the assemblage characteristic collection and first similarity for being positively correlated feature set；According to the be positively correlated feature set first default use
It clicks probability and obtains the first user click probability that the content combines with first similarity in family；
It collects the negative example and determines that the feature set of the combination of content described in collection is used as negatively correlated feature set, and compare each institute
The assemblage characteristic collection and the negatively correlated feature set for stating content combination, obtain the assemblage characteristic collection and the negatively correlated feature set
The first similarity；Probability is clicked according to the second pre-set user of negatively correlated feature set, it is to be understood that the second pre-set user
It clicks probability and is lower than first pre-set user click probability, according to second similarity and the second default use of negatively correlated collection
It clicks probability and obtains the second user click probability of content combination in family；
Probability is clicked in conjunction with the first user and second user clicks probability, clicks probability and second for example, by using the first user
The user that the mode that user's point probability is added obtains content combination clicks probability, and then compares user's point of each content combination
Probability is hit, recommendation combination is obtained.
In other embodiments, step S304 can also be used according to first similarity and second similarity
Other modes obtain recommendation combination, these are all that the feature combined strategy of content combination is compared to be pushed away with sample
Content combination is recommended, the present embodiment is without limitation.
In the present embodiment, whether each single item feature that more positive and negative correlated characteristic and assemblage characteristic are concentrated is identical, in turn
According to comparison result, each node is content combination marking, finally by the marking result algebraic addition of each node, is obtained most
Whole score value.It should be noted that in other embodiments, the mode of same Semantic judgement also can be used, by similar feature
It is classified as one kind to make a decision, the present embodiment is without limitation.
Further, in the present embodiment, identical as feature is positively correlated to get to one point, it is identical as negatively correlated feature, then
Obtain negative one point；In other embodiments, relevant score value, this implementation can also be provided according to the degree of correlation between two features
Example is without limitation.
It lifts a kind of possible situation to be illustrated the working principle of the embodiment of the present invention 1: assuming that using collaborative filtering, master
Topic three kinds of contents of model and correlation rule recall strategy as heretofore described content acquisition mode, obtain content set A, interior
Hold collection B and content set C, is obtained in user according to the marking sort algorithm such as euclidean metric of the collaborative filtering based on user
The similarity weight of appearance, and then the sequence of the recommendation in content set A is obtained, similarly according to the marking in topic model and correlation rule
Sort algorithm obtains the sequence of the recommendation in content set B and content set C.
And then assume to take first three content for sorting forward in content set to take in preparation recommendation collection in the present embodiment,
In this way, which it includes 9 contents that preparation recommendation, which is concentrated,.Then the content concentrated to preparation recommendation carries out re-scheduling, goes
After identical content in 9 contents, 6 contents are obtained, it is assumed that recommendation combination includes 3 contents；Therefore obtain it is pre-
After standby recommendation collection, after the preparation recommendation that preparation recommendation is concentrated is carried out random combine, i.e., according to C6 3Obtain 20
A content combination.
The preset content feature set of each preparation recommendation itself is obtained again, thus in three of each combined arrangement
The assemblage characteristic collection that the preset content feature set merging of appearance forms the combined arrangement obtains particular user in the present embodiment
The scene characteristic that hobby feature, use habit feature and user use at this time is as judgement feature set, the hobby of particular user
Feature includes: favorite article category, the time etc. for browsing article, what user's use habit feature was often used when including: online
Network type, in nearly one week average stay time of the page etc., the scene characteristic that user uses at this time includes: on user terminal
Current time and date etc..And further, from the database of the preset user, according to going through for user displaying before
History content combination, judge determine feature set in each judgements feature be positive correlated characteristic or negative correlation feature.
In the present embodiment, it uses GBDT model to carry out marking sequence for each combination: will determine each in feature set
Item feature obtains the decision tree of GBDT model as each node in GBDT model, by the assemblage characteristic collection of each combined arrangement
And with aspect ratio pair in each node it is that each combined arrangement is given a mark, such as: when the feature of a certain node are as follows: beauty
Food, and the assemblage characteristic collection of a combined arrangement includes this feature of cuisines, which obtains 1 point, and another is combined
The assemblage characteristic collection of content does not include this feature of cuisines, then not score, after assemblage characteristic collection traversal all nodes,
Each combined arrangement is comprehensive in the scoring event of each node, the final score of each combined arrangement is obtained, and then sort
To the corresponding combined arrangement of highest score, using the combined arrangement as recommendation combined arrangement, and then user terminal interface is pushed to,
User is attracted to select to recommend combined arrangement.
By the content set for three concern features obtained according to three kinds of acquisition modes, selected from three content sets
Optimal two contents under the concern feature obtain multiple content combinations after doing random combine, and multiple combined combinations are special
Sign is ranked up marking according to user's history data and obtains most preferably recommending combination；By the way that assemblage characteristic is done marking sequence, compare
Existing single attribute of interest sequence, it is of interest that the cooperation of feature between content can guarantee the comprehensive of commending contents, reduce
The content isolation occurred under existing proposed algorithm.
Embodiment 2
Referring to FIG. 4, Fig. 4 is a kind of functional block diagram for content combined recommendation 40 that the embodiment of the present invention 2 provides.
The content combined recommendation 40 includes: acquiring unit 401, processing unit 402 and transmission unit 403.It obtains
Unit 401 is for obtaining prepared recommendation collection and preset recommendation number；Processing unit 402 is used for according to the preparation
Multiple content combinations are calculated in recommendation collection and recommendation number, and determine in recommendation in the combination of the multiple content
Hold combination；Transmission unit 403 is used to push the recommendation combination to user.
Optionally, acquiring unit 401 is also used to acquire using at least two content acquisition modes interior in presetting database
Hold, obtains content set corresponding with every kind of content acquisition mode；And it is corresponding in every kind of content acquisition mode by preset rules
Content is extracted in the content set, obtains the prepared recommendation collection.
Optionally, processing unit 402 is also used to using acquisition corresponding at least two content acquisitions mode
Rule carries out relevancy ranking to the content in the database, obtains corresponding according to related to every kind of content acquisition mode
Spend the content set of sequence；
Corresponding, acquiring unit 401 is also used to obtain default selection number, so processing unit 402 acquire it is each described
The content of the highest default selection number of the degree of correlation, obtains the prepared recommendation collection in content set.
Optionally, processing unit 402 is also used to the number with the recommendation number for the combination of each content, random groups
It closes each content that the prepared recommendation is concentrated and obtains multiple content combinations.
Optionally, processing unit 402 is also used to remove the prepared recommendation and concentrates identical content, then random combine
Each content that the prepared recommendation is concentrated obtains multiple content combinations.
Optionally, acquiring unit 401 is also used to obtain the preset content spy of each content in each content combination
Collection, and obtain the judgement feature set of user, it should be noted that preset content feature set and the judgement feature set include
Feature type is identical.
Corresponding, processing unit 402 is also used to merge the described default of each content in each content combination
Content characteristic collection obtains the assemblage characteristic collection of each content combination, so compare the judgements feature set and it is each described in
Assemblage characteristic collection determines and the recommendation combination is calculated.
Optionally, acquiring unit 401 is also used to obtain the historical content combination shown to user；Corresponding, processing is single
Member 402 is combined according to the historical content, and the historical content combination for determining that user selects determines to collect for positive example, determines that user is unselected
The historical content the selected example that is negative determines collection, it should be noted that positive example determines that collection determines that collection composition determines feature with the negative example
Collection.
Further, the feature that processing unit 402 collects that the positive example determines that content described in collection combines, which is used as, to be positively correlated
Whether feature, each single item feature for determining that the assemblage characteristic of each content combination is concentrated are positively correlated feature with described in each single item
Identical, when to be, the content combines to obtain the first score；
At the same time, each single item feature for determining that the assemblage characteristic of each content combination is concentrated is born with described in each single item
Whether correlated characteristic is identical, and when to be, the content combines to obtain the second score, and the first score is higher than the second score；And root
Each content, which is calculated, according to first score and second score of the combination of content described in each combines finally obtained point
Value；
It is final to determine that the corresponding content group of highest score is combined into recommendation combination, and then transmission unit 403 is pushed away to user
Recommendation is sent to combine.
Content combined recommendation 40 and aforementioned content combined recommendation method shown in FIG. 1 in the present embodiment are based on same
Invention under one design, passes through the aforementioned detailed description to content combined recommendation method and its various change form, this field skill
Art personnel can be apparent from the implementation process of the present embodiment content combined recommendation 40, so in order to illustrate the letter of book
Clean, details are not described herein.
Embodiment 3
Referring to FIG. 5, Fig. 5 is the structural block diagram for a kind of electronic equipment 50 that the embodiment of the present invention 3 provides.The electronics is set
Standby 50 include: memory 51 and processor 52.
The memory 51 and processor 52 are directly or indirectly electrically connected between each other, with realize data transmission or
Interaction.It is electrically connected for example, these elements can be realized between each other by one or more communication bus or signal wire.It is described pre-
If database includes that at least one can be stored in the memory 51 or solidify in the form of software or firmware (firmware)
Software function module in the operating system (operating system, OS) of the electronic equipment 50.The processor 52
For executing the executable module stored in memory 51.
Wherein, memory 51 may be, but not limited to, random access memory (Random Access Memory,
RAM), read-only memory (Read Only Memory, ROM), programmable read only memory (Programmable Read-Only
Memory, PROM), erasable read-only memory (Erasable Programmable Read-Only Memory, EPROM),
Electricallyerasable ROM (EEROM) (Electric Erasable Programmable Read-Only Memory, EEPROM) etc..
Wherein, memory 51 is for storing program, and the processor 52 executes described program after receiving and executing instruction, and aftermentioned
Method performed by the electronic equipment 50 for the flow definition that inventive embodiments any embodiment discloses can be applied to processor 52
In, or realized by processor 52.
Processor 52 may be a kind of IC chip, the processing capacity with signal.Above-mentioned processor can be
General processor, including central processing unit (Central Processing Unit, CPU), network processing unit (Network
Processor, NP) etc.；It can also be digital signal processor (DSP), specific integrated circuit (ASIC), ready-made programmable gate array
Arrange (FPGA) either other programmable logic device, discrete gate or transistor logic, discrete hardware components.It may be implemented
Or disclosed each method, step and logic diagram in the execution embodiment of the present invention.General processor can be microprocessor
Or the processor is also possible to any conventional processor etc..
Embodiment 4
The embodiment of the present invention 4 provides a kind of readable storage medium storing program for executing, is stored with computer program in the readable storage medium storing program for executing,
When the computer program is run on computers, so that the computer executes the combination of content described in embodiment 1 and pushes away
Recommend method.The present invention repeats no more this.
In several embodiments provided herein, it should be understood that disclosed device and method can also pass through
Other modes are realized.The apparatus embodiments described above are merely exemplary, for example, flow chart and block diagram in attached drawing
Show the device of multiple embodiments according to the present invention, the architectural framework in the cards of method and computer program product,
Function and operation.In this regard, each box in flowchart or block diagram can represent the one of a module, section or code
Part, a part of the module, section or code, which includes that one or more is for implementing the specified logical function, to be held
Row instruction.It should also be noted that function marked in the box can also be to be different from some implementations as replacement
The sequence marked in attached drawing occurs.For example, two continuous boxes can actually be basically executed in parallel, they are sometimes
It can execute in the opposite order, this depends on the function involved.It is also noted that every in block diagram and or flow chart
The combination of box in a box and block diagram and or flow chart can use the dedicated base for executing defined function or movement
It realizes, or can realize using a combination of dedicated hardware and computer instructions in the system of hardware.
In addition, each functional module in each embodiment of the present invention can integrate one independent portion of formation together
Point, it is also possible to modules individualism, an independent part can also be integrated to form with two or more modules.
It, can be with if the function is realized and when sold or used as an independent product in the form of software function module
It is stored in a computer readable storage medium.Based on this understanding, technical solution of the present invention is substantially in other words
The part of the part that contributes to existing technology or the technical solution can be embodied in the form of software products, the meter
Calculation machine software product is stored in a storage medium, including some instructions are used so that a computer equipment (can be a
People's computer, server or network equipment etc.) it performs all or part of the steps of the method described in the various embodiments of the present invention.
And storage medium above-mentioned includes: that USB flash disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), arbitrary access are deposited
The various media that can store program code such as reservoir (RAM, Random Access Memory), magnetic or disk.It needs
Illustrate, herein, relational terms such as first and second and the like be used merely to by an entity or operation with
Another entity or operation distinguish, and without necessarily requiring or implying between these entities or operation, there are any this realities
The relationship or sequence on border.Moreover, the terms "include", "comprise" or its any other variant are intended to the packet of nonexcludability
Contain, so that the process, method, article or equipment for including a series of elements not only includes those elements, but also including
Other elements that are not explicitly listed, or further include for elements inherent to such a process, method, article, or device.
In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that including the element
Process, method, article or equipment in there is also other identical elements.
These are only the preferred embodiment of the present invention, is not intended to restrict the invention, for those skilled in the art
For member, the invention may be variously modified and varied.All within the spirits and principles of the present invention, it is made it is any modification,
Equivalent replacement, improvement etc., should all be included in the protection scope of the present invention.
Claims (10)
1. a kind of content combined recommendation method, which is characterized in that the recommended method includes:
Obtain preparation recommendation collection and preset recommendation number；
Multiple content combinations are calculated according to the prepared recommendation collection and recommendation number；Wherein, each described interior
Hold the content in combination comprising the recommendation number；
Determine that recommendation combines in the combination of the multiple content；
Push the recommendation combination.
2. recommended method according to claim 1, which is characterized in that the acquisition preparation recommendation collection, comprising:
The content in presetting database is acquired using at least two content acquisition modes, is obtained corresponding with every kind of content acquisition mode
Content set；
Content is extracted in the corresponding content set of every kind of content acquisition mode by preset rules, the preparation is obtained and recommends
Content set.
3. recommended method according to claim 2, which is characterized in that described to be acquired using at least two content acquisition modes
Content in presetting database obtains content set corresponding with every kind of content acquisition mode, comprising:
Using collection rule corresponding at least two content acquisitions mode, the content in the database is carried out
Relevancy ranking obtains the content set according to relevancy ranking corresponding with every kind of content acquisition mode；
It is corresponding, it is described to extract content in the corresponding content set of every kind of content acquisition mode by preset rules, it obtains
The prepared recommendation collection, comprising:
The content for acquiring the highest default selection number of the degree of correlation in each content set, obtains the prepared recommendation
Collection.
4. recommended method according to claim 1, which is characterized in that described according to the prepared recommendation collection and recommendation
Multiple content combinations are calculated in content number, comprising:
It take the recommendation number as the number of each content combination, preparation recommendation described in random combine is concentrated each
Content obtains multiple contents combinations.
5. recommended method according to claim 4, which is characterized in that the preparation recommendation collection described in the random combine
In each content obtain the combination of multiple contents before, the method also includes:
It removes the prepared recommendation and concentrates identical content.
6. recommended method according to claim 1, which is characterized in that described to determine to recommend in the combination of the multiple content
Content combination, comprising:
Obtain the preset content feature set of each content in each content combination；
The preset content feature set for merging each content in each content combination, obtains each content group
The assemblage characteristic collection of conjunction；
The judgement feature set of user is obtained, the type for the feature that the assemblage characteristic collection includes and the judgement feature set include
The type of feature includes: article correlated characteristic, user's habit at least one of feature and user's scene characteristic feature or more
The combination of kind feature；
The judgement feature set and each assemblage characteristic collection are compared, is determined in the recommendation in the multiple content combination
Hold combination.
7. recommended method according to claim 6, which is characterized in that the judgement feature set for obtaining user, comprising:
It obtains and the historical content that user showed is combined；
It is combined according to the historical content, the historical content combination for determining that user selects determines to collect for positive example, determines that user is unselected
The historical content selected be negative example determine collection；The judgement feature set includes that the positive example determines that collection and the negative example determine collection.
8. recommended method according to claim 7, which is characterized in that it is described compare the judgements feature set and it is each described in
Assemblage characteristic collection determines the recommendation combination in the multiple content combination, comprising:
It collects the positive example and determines that the feature set of the combination of content described in collection is used as positive correlation feature set, collect the negative example and determine
The feature set of the combination of content described in collection is as negatively correlated feature set；
Preset initial association relational model is trained according to the positive correlation feature set and the negatively correlated feature set, it is raw
At association relation model；
Corresponding assemblage characteristic collection is combined according to the association relation model and the multiple content, determines the multiple content group
Recommendation combination in conjunction.
9. a kind of content combined recommendation, which is characterized in that described device includes:
Acquiring unit, for obtaining prepared recommendation collection and preset recommendation number；
Processing unit is combined for multiple contents to be calculated according to the prepared recommendation collection and recommendation number, and
Determine that recommendation combines in the combination of the multiple content；
Transmission unit, for pushing the recommendation combination to user.
10. a kind of readable storage medium storing program for executing, which is characterized in that computer program is stored in the readable storage medium storing program for executing, when described
When computer program is run on computers, so that computer execution is interior as described in any one of claim 1-8
Hold combined recommendation method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811324196.4A CN109241451B (en) | 2018-11-08 | 2018-11-08 | Content combination recommendation method and device and readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811324196.4A CN109241451B (en) | 2018-11-08 | 2018-11-08 | Content combination recommendation method and device and readable storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109241451A true CN109241451A (en) | 2019-01-18 |
| CN109241451B CN109241451B (en) | 2021-07-16 |
Family
ID=65077533
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811324196.4A Active CN109241451B (en) | 2018-11-08 | 2018-11-08 | Content combination recommendation method and device and readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109241451B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110543598A (en) * | 2019-09-06 | 2019-12-06 | 腾讯科技（深圳）有限公司 | information recommendation method and device and terminal |
| WO2020168841A1 (en) * | 2019-02-21 | 2020-08-27 | 北京京东尚科信息技术有限公司 | Network resource pushing method, device, and storage medium |
| CN111626767A (en) * | 2020-04-29 | 2020-09-04 | 拉扎斯网络科技（上海）有限公司 | Resource data distribution method, device and equipment |
| CN113254843A (en) * | 2021-06-29 | 2021-08-13 | 腾讯科技（深圳）有限公司 | Information pushing method and device and storage medium |
| CN113763092A (en) * | 2020-11-10 | 2021-12-07 | 北京沃东天骏信息技术有限公司 | Method, device, equipment and computer readable medium for recommending commodities |
| CN113806634A (en) * | 2021-09-17 | 2021-12-17 | 中国联合网络通信集团有限公司 | Service package recommendation method and device and server |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105740468A (en) * | 2016-03-07 | 2016-07-06 | 达而观信息科技（上海）有限公司 | Individuation recommendation method and system combined with content publisher information |
| CN107743249A (en) * | 2017-11-27 | 2018-02-27 | 四川长虹电器股份有限公司 | A kind of CTR predictor methods based on Model Fusion |
| CN107944063A (en) * | 2018-01-16 | 2018-04-20 | 马上消费金融股份有限公司 | It is a kind of that method and system are recommended based on the news of topic model and groups of users |
| CN108304556A (en) * | 2018-02-06 | 2018-07-20 | 中国传媒大学 | The personalized recommendation method being combined with collaborative filtering based on content |
-
2018
- 2018-11-08 CN CN201811324196.4A patent/CN109241451B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105740468A (en) * | 2016-03-07 | 2016-07-06 | 达而观信息科技（上海）有限公司 | Individuation recommendation method and system combined with content publisher information |
| CN107743249A (en) * | 2017-11-27 | 2018-02-27 | 四川长虹电器股份有限公司 | A kind of CTR predictor methods based on Model Fusion |
| CN107944063A (en) * | 2018-01-16 | 2018-04-20 | 马上消费金融股份有限公司 | It is a kind of that method and system are recommended based on the news of topic model and groups of users |
| CN108304556A (en) * | 2018-02-06 | 2018-07-20 | 中国传媒大学 | The personalized recommendation method being combined with collaborative filtering based on content |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020168841A1 (en) * | 2019-02-21 | 2020-08-27 | 北京京东尚科信息技术有限公司 | Network resource pushing method, device, and storage medium |
| US11483253B2 (en) | 2019-02-21 | 2022-10-25 | Beijing Jingdong Shangke Information Technology Co., Ltd. | Network resource pushing method, device, and storage medium |
| CN110543598A (en) * | 2019-09-06 | 2019-12-06 | 腾讯科技（深圳）有限公司 | information recommendation method and device and terminal |
| CN111626767A (en) * | 2020-04-29 | 2020-09-04 | 拉扎斯网络科技（上海）有限公司 | Resource data distribution method, device and equipment |
| CN111626767B (en) * | 2020-04-29 | 2023-09-08 | 拉扎斯网络科技（上海）有限公司 | Resource data issuing method, device and equipment |
| CN113763092A (en) * | 2020-11-10 | 2021-12-07 | 北京沃东天骏信息技术有限公司 | Method, device, equipment and computer readable medium for recommending commodities |
| CN113254843A (en) * | 2021-06-29 | 2021-08-13 | 腾讯科技（深圳）有限公司 | Information pushing method and device and storage medium |
| CN113254843B (en) * | 2021-06-29 | 2021-10-01 | 腾讯科技（深圳）有限公司 | Information pushing method and device and storage medium |
| CN113806634A (en) * | 2021-09-17 | 2021-12-17 | 中国联合网络通信集团有限公司 | Service package recommendation method and device and server |
| CN113806634B (en) * | 2021-09-17 | 2023-05-30 | 中国联合网络通信集团有限公司 | Service package recommending method, device and server |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109241451B (en) | 2021-07-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109241451A (en) | A kind of content combined recommendation method, apparatus and readable storage medium storing program for executing | |
| CN105224699B (en) | News recommendation method and device | |
| US9703783B2 (en) | Customized news stream utilizing dwelltime-based machine learning | |
| CN110532479A (en) | A kind of information recommendation method, device and equipment | |
| EP2048605B1 (en) | System and method for performing discovery of digital information in a subject area | |
| CN109543111A (en) | Recommendation information screening technique, device, storage medium and server | |
| CN106294425B (en) | The automatic image-text method of abstracting and system of commodity network of relation article | |
| US9031944B2 (en) | System and method for providing multi-core and multi-level topical organization in social indexes | |
| CN110097412A (en) | Item recommendation method, device, equipment and storage medium | |
| CN106339502A (en) | Modeling recommendation method based on user behavior data fragmentation cluster | |
| CN105159932B (en) | A kind of data retrieval engine and ordering system and method | |
| WO2019056661A1 (en) | Search term pushing method and device, and terminal | |
| CN105723402A (en) | Systems and methods for determining influencers in a social data network | |
| CN108648058B (en) | Product sorting method and device, electronic equipment and storage medium | |
| WO2014056408A1 (en) | Information recommending method, device and server | |
| CN111191112A (en) | Electronic reading data processing method, device and storage medium | |
| KR101925950B1 (en) | Method and device for recommending contents based on inflow keyword and relevant keyword for contents | |
| CN104361109B (en) | The method and apparatus for determining picture the selection result | |
| Nakatsuji et al. | Collaborative filtering by analyzing dynamic user interests modeled by taxonomy | |
| CN108694211B (en) | Application distribution method and device | |
| KR101976056B1 (en) | System and method for recommendation | |
| Ashraf et al. | Personalized news recommendation based on multi-agent framework using social media preferences | |
| CN113761084B (en) | POI search ranking model training method, ranking device, method and medium | |
| US20150206220A1 (en) | Recommendation Strategy Portfolios | |
| Souali et al. | An automatic ethical-based recommender system for e-commerce |
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 |