CN110297987A - A kind of model recommended method, device, equipment and storage medium - Google Patents
A kind of model recommended method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN110297987A CN110297987A CN201910591329.2A CN201910591329A CN110297987A CN 110297987 A CN110297987 A CN 110297987A CN 201910591329 A CN201910591329 A CN 201910591329A CN 110297987 A CN110297987 A CN 110297987A
- Authority
- CN
- China
- Prior art keywords
- user
- model
- grade
- text
- quality degree
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9536—Search customisation based on social or collaborative filtering
Abstract
The embodiment of the invention discloses a kind of model recommended method, device, equipment and storage mediums, this method comprises: obtaining the comment text information of model text information and the user comment model that user is posted by；According to the model text information and the comment text information, the corresponding publication quality degree of the user is determined；The activity of the user is determined according to the publication quality degree, and the target model for recommending the user is determined according to the liveness.Technical solution through the embodiment of the present invention can be improved the accuracy of recommendation, and improve model recommendation effect, promote user experience.
Description
Technical field
The present embodiments relate to the information processing technology more particularly to a kind of model recommended method, device, equipment and storages
Medium.
Background technique
With the continuous development of Internet technology, so that various network models are all the fashion.Network model can refer to net
The article or opinion that the people deliver in forum, and communication and discussion can be carried out by way of being posted by and commenting on model.
For example, the user in live streaming platform can discuss live content and expression mood by discussion bar.
Currently, the high model of current hot topic degree is often recommended each user, so that user can when recommending model
With the model for clicking directly on browsing or downloading is recommended.However, the way of recommendation in the prior art does not consider each use
The demand at family and liveness situation in discussion bar, and be easy to be interfered by user behaviors such as some brush models, so that recommending
Model be not required for user, to reduce recommendation effect, and by repeatedly clicking or downloading unwanted note
Son.
Summary of the invention
The embodiment of the invention provides a kind of model recommended method, device, equipment and storage mediums, to improve the standard recommended
True property, to improve recommendation effect.
In a first aspect, the embodiment of the invention provides a kind of model recommended methods, comprising:
Obtain the comment text information of model text information and the user comment model that user is posted by；
According to the model text information and the comment text information, the corresponding publication quality degree of the user is determined；
Determine the activity of the user according to the publication quality degree, and determined according to the liveness recommend it is described
The target model of user.
Second aspect, the embodiment of the invention also provides a kind of model recommendation apparatus, comprising:
Post information obtains module, for obtaining the model text information and the user comment model that user is posted by
Comment text information；
Quality degree determining module is issued, for determining institute according to the model text information and the comment text information
State the corresponding publication quality degree of user；
Target model determining module, for determining the activity of the user according to the publication quality degree, and according to institute
It states liveness and determines the target model for recommending the user.
The third aspect, the embodiment of the invention also provides a kind of equipment, the equipment includes:
One or more processors；
Memory, 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 processing
Device realizes such as model recommended method provided by any embodiment of the invention.
Fourth aspect, the embodiment of the invention also provides a kind of computer readable storage mediums, are stored thereon with computer
Program realizes such as model recommended method provided by any embodiment of the invention when the program is executed by processor.
The embodiment of the present invention passes through the model text information and user comment model of each model issued according to user
Each comment text information, determine that user is posted by and comments on the degree of model, i.e., publication quality degree, and according to
Publication quality degree can more objectively and accurately determine the activity of the user, so as to avoid by user behaviors such as brush models
It is interfered, and recommends to user the target model to match with its liveness, so that the model recommended is required for user,
The accuracy recommended is improved, to improve recommendation effect.
Detailed description of the invention
Fig. 1 is a kind of flow chart for model recommended method that the embodiment of the present invention one provides；
Fig. 2 is a kind of flow chart of model recommended method provided by Embodiment 2 of the present invention；
Fig. 3 is a kind of structural schematic diagram for model recommendation apparatus that the embodiment of the present invention three provides；
Fig. 4 is a kind of structural schematic diagram for equipment that the embodiment of the present invention four provides.
Specific embodiment
The present invention is described in further detail with reference to the accompanying drawings and examples.It is understood that this place is retouched
The specific embodiment stated is used only for explaining the present invention rather than limiting the invention.It also should be noted that in order to just
Only the parts related to the present invention are shown in description, attached drawing rather than entire infrastructure.
Embodiment one
Fig. 1 is the flow chart of a kind of model recommended method that the embodiment of the present invention one provides, the present embodiment be applicable to
User carries out the case where personalized recommendation model.This method can be executed by model recommendation apparatus, which can be by software
And/or the mode of hardware is realized, is integrated in the equipment with the information processing function, such as in client or server.
This method specifically includes the following steps:
S110, the comment text information for obtaining model text information and user comment model that user is posted by.
Wherein, model can refer to the article that user issues in discussion bar.Model text information can refer to that user issues
The model text and temperature information of model.Model text can refer to the title and content of model.Temperature information can refer to pair
The comment information of model thumbs up information and forwarding information.Comment text information can refer to that user replys the comment text of model
Information and return information are thumbed up with to each comment.
Specifically, for each user, each of user publication can be obtained according to the user behaviors log of the user
The model text information of model and each comment text information of the user comment model.
S120, according to model text information and comment text information, determine the corresponding publication quality degree of user.
Wherein, publication quality degree can refer to that user is posted by and comments on the degree of model.
Specifically, the model text information for each model that the present embodiment can be issued based on user, determines that user sends out
Each model quality degree of cloth, for example, can by the character length of model text, the number of reviews of commenting on the model, thumb up number
The addition result of amount and forwarding quantity is determined as the corresponding model quality degree of the model.Some model text in the present embodiment
Character length, number of reviews thumb up quantity and forward quantity more, then show that time of the user effort on the model is longer,
To which the degree of the model is higher.It can determine that user comments based on each comment text information of user comment model
Each comment quality degree of opinion, for example, the character length for commenting on the comment text of model and the comment can be thumbed up number
It measures and is determined as the corresponding comment quality degree of the model with the addition result for replying quantity.The word of some comment text in the present embodiment
Symbol length to the reply quantity of the comment and thumbs up that quantity is more, then shows that time of the user effort in the comment is longer, from
And the degree of the comment is higher.By the way that each model quality degree be added with each comment quality degree, and by phase
Result is added to be determined as the corresponding publication quality degree of the user, so as to objectively determine that user is posted by and comments on model
Quality degree.
S130, the activity of the user is determined according to publication quality degree, and determines the target for recommending user according to liveness
Model.
Wherein, liveness can refer to active degree of the user in discussion bar.Target model can refer to the work with user
The model to be recommended that jerk matches.
Specifically, publication quality degree directly can be determined as the activity of the user by the present embodiment, i.e. publication quality degree is got over
Height, then the activity of the user is also higher, objective and accurate so as to the degree that is posted by using user and comment on model
Ground reflects the activity of the user.And user is recommended using the model to match with liveness as target model, so as to
It improves and recommends efficiency.Illustratively, if executing the operation of above-mentioned steps S110-S130 in the server, server can be incited somebody to action
The target model determined is sent in client, to show target model on the display interface of client, so that user
It can be intuitive to see recommended target model.If executing the operation of above-mentioned steps S110-S130, client in the client
End can directly show target model when determining target model in the display interface, without carrying out the hair of target model
Operation is sent, to further improve display efficiency and recommend efficiency.
Illustratively, " the target model for recommending user is determined according to liveness " in S130 may include: that detection is lived
Whether jerk is greater than default liveness threshold value；If so, the model interested of user is determined according to the historical viewings information of user,
And using model interested as the target model for recommending user；Recommend if it is not, being then used as the current highest model of hot topic degree
To the target model of user.Wherein, historical viewings information can refer to the post information browsed before user.Post information can be with
Including but not limited to model classification, model number of reviews, forward quantity and thumb up quantity.Popular degree can be for reflecting model
Pouplarity.For example, when the number of reviews of model, forward quantity and thumb up quantity it is higher when, may indicate that the model
Popular degree is higher.
Specifically, by determining different Generalization bounds according to the activity of the user, to obtain the mesh to match
Mark model.When the activity of the user is greater than default liveness threshold value, show that the user often issues the model of high quality and comments
By can determine that user browses the highest model classification of frequency according to the historical viewings information of user at this time, and by the model
Popular model under classification is determined as model interested, i.e. target model；It can also determine and going through in historical viewings information
The highest similar model of history model similarity, and the user is recommended using the similar model as model interested.When user's
When liveness is less than or equal to default liveness threshold value, show that the user may be model and the comment of new user or publication
Quality is lower, i.e. the historical viewings information of the user does not have reference value, at this time can be directly highest by current hot topic degree
Model, so that the model recommended is model needed for user, realizes user as the target model for recommending user
Personalized recommendation, improve the accuracy and recommendation effect of recommendation, and can be to avoid repeatedly clicking or download unwanted note
Son.
The technical solution of the present embodiment is commented by the model text information and user for each model issued according to user
By each comment text information of model, determine that user is posted by and comments on the degree of model, i.e. publication quality degree,
And the activity of the user more accurately objectively can be determined according to publication quality degree, so as to avoid being used by brush model etc.
Family behavior is interfered, and the target model to match to user's recommendation with its liveness, so that the model recommended is user institute
It needs, improves the accuracy of recommendation, to improve recommendation effect.
Based on the above technical solution, S120 may include: each model text to user's publication and each comment
Paper this progress character duplicate removal processing；According to corresponding turn of character length, model text in the model text after duplicate removal processing
It sends out quantity, number of reviews and thumbs up quantity, determine the corresponding model quality degree of each model text；After duplicate removal processing
The corresponding reply quantity of character length, comment text in comment text and quantity is thumbed up, determines that each comment text is corresponding
Comment on quality degree；The corresponding publication quality degree of user is determined according to each model quality degree and each comment quality degree.
Wherein, character length can refer to the character quantity in model text or comment text.Specifically, the present embodiment
First the repeat character (RPT) in each model text and each comment text is removed, and according to the model text after duplicate removal processing or is commented
Character length in paper sheet determines corresponding model quality degree and comment quality degree, and by each model quality degree and each comment
The addition result of quality degree is determined as the corresponding publication quality degree of user, so as to improve the accuracy of matter measure calculation.
Illustratively, the corresponding model quality degree of each model text can be determined according to the following formula:
Wherein, DPRefer to the corresponding model quality degree of model text P；LpIt is to the model text after model text P duplicate removal processing
Character length in this；ncIt is the corresponding number of reviews of model text P；nfIt is the corresponding forwarding quantity of model text P；nlIt is note
This P of Ziwen is corresponding to thumb up quantity.
In the present embodiment, if the character length in model text after duplicate removal processing is longer, show that model is included
Information content it is bigger, user's the time it takes is longer, so that model quality is higher.If number of reviews forwards quantity and thumbs up
Quantity is more, then shows that the pouplarity of the model is higher, while the quality that also can reflect out the model is also higher.Also
It is to say, the model quality degree in the present embodiment and character length, number of reviews forward quantity and thumb up the directly proportional pass of quantity
System.
In above-mentioned formula (1), utilizeCharacterize the proportional relation between character length and model quality degree,
And it utilizesCharacterizing number of reviews, forwarding quantity and thumbing up the direct ratio between quantity and model quality degree
Relationship.ForFor, due to for other parameters index, character length LpThe order of magnitude it is larger, for example,
User gradation is 3, character length LpIt is 10000, to need to character length LpIt carries out taking logarithm process, realizes compression quantity
The purpose of grade difference, so that data variation is more steady, it is cumulative convenient for subsequent progress numerical value.By right1 is added to may be implemented
The case where Laplce is smooth, effectively avoids zero probability, improves the accuracy of calculating.
ForFor, due to number of reviews, to forward quantity and thumb up quantity be for characterizing model
The same type of parameter of pouplarity, to calculate, can first determine number of reviews, forwarding quantity and thumb up to simplify
Then summation between quantity again carries out summation taking logarithm process, so as to more rapidly realize to these three parameters
Carry out the purpose of the compression order of magnitude.By rightAdd 1 also for realizing that Laplce is smooth, effectively avoids zero general
The case where rate, improves the accuracy of calculating.
Due to model quality degree and character length, number of reviews, forward quantity and to thumb up quantity proportional, thus
In order to meet this logical relation, need byWithIt is added, and will add up result and be determined as
Model quality degree.Compared to by character length, number of reviews, forwarding quantity and thumbing up the sum of quantity and be determined directly as model quality
Degree, i.e. DP=Lp+nc+nf+nlFor, the variation using the calculated model quality degree of formula (1) is more steady, for example, utilizing
The corresponding model quality degree of original calculated three models of formula are as follows: 90,3000,500, and utilize formula (1) calculated
The quality degree of these three models are as follows: 9.6,16.5,12.3, to reduce the influence of change degree, improve model quality degree
Reference value, and then improve the subsequent accuracy based on the calculated liveness of model quality degree.
Illustratively, the corresponding comment quality degree of each comment text can be determined according to the following formula:
Wherein, DMRefer to the corresponding model quality degree of comment text M；LMIt is to the comment text after comment text M duplicate removal processing
Character length in this；ncIt is the corresponding number of reviews of comment text M；nlBeing that comment text M is corresponding thumbs up quantity.Due to
In discussion bar, user is only capable of comment text being commented on and thumbed up again operation, operation can not be forwarded, thus the present embodiment
According to the corresponding character length L of comment textM, number of reviews ncWith thumb up quantity nl, can be determined based on above-mentioned formula (2)
The corresponding comment quality degree of the comment text.The Computing Principle of formula (2) is similar with above-mentioned formula (1) in the present embodiment, this
Place is no longer repeated.
Embodiment two
Fig. 2 is a kind of flow chart of model recommended method provided by Embodiment 2 of the present invention, and the present embodiment is in above-mentioned implementation
On the basis of example, " determining the activity of the user according to publication quality degree " is advanced optimized.Wherein with above-mentioned implementation
Example is identical or the explanation of corresponding term details are not described herein.
Referring to fig. 2, model recommended method provided in this embodiment specifically includes the following steps:
S210, the comment text information for obtaining model text information and user comment model that user is posted by.
S220, according to model text information and comment text information, determine the corresponding publication quality degree of user.
S230, account grade and concern information according to user, determine the user gradation of user.
Wherein, account grade can refer to the account grade in user itself discussion bar.Account grade in the present embodiment can
To be a specific numerical value.Concern information can refer to user main broadcaster's user information of interest, for example the user is in institute
Pay close attention to the discussion bar grade in main broadcaster user's discussion bar.If model and number of reviews that the user issues in some main broadcaster user's note
More, then discussion bar higher grade of the user in main broadcaster user's discussion bar.User gradation can refer to the integrated level of user.
Specifically, for each main broadcaster user of user's concern, the account grade of the user can be existed divided by the user
The result of discussion bar grade in main broadcaster user's discussion bar is determined as the corresponding grade of the main broadcaster user of interest.It will be of interest
The average value of the corresponding grade of each main broadcaster user can be determined as the user gradation of the user.
Illustratively, S230 may include: the bean vermicelli of each main broadcaster user paid close attention to according to the account grade of user, user
Board grade and discussion bar grade in each main broadcaster user, determine user to the bean vermicelli grade of each main broadcaster user；According to each powder
Noble's grade of silk grade and user, determines the user gradation of user.
Wherein, bean vermicelli board grade can be used for reflecting that the user beats reward degree to main broadcaster user.Such as when user is to certain
A main broadcaster user beat reward it is more, then user is in the bean vermicelli board higher grade of the main broadcaster user.Any one use in the present embodiment
Model can be posted by or commented on per family in the discussion bar of main broadcaster user.Discussion bar grade can be used for reflecting user in the master
Active degree in the discussion bar of broadcasting user.Bean vermicelli grade can be used for characterizing user to the integrated level of each main broadcaster user.It is expensive
Race's grade can refer to the level value that user is obtained by payment way.
Specifically, user can be determined to the bean vermicelli grade of each main broadcaster user by following formula:
Wherein, EikIt is bean vermicelli grade of the user i to main broadcaster user k；EiIt is the account grade of user i；It is that user i is closed
The bean vermicelli board grade of the main broadcaster user k of note；It is the discussion bar grade in main broadcaster user k.
It should be noted that if user i, which is not beaten, appreciated main broadcaster user k, then the bean vermicelli for the main broadcaster user k that user i is paid close attention to
Board grade is 0；If user i did not issue model or comment in the discussion bar of main broadcaster user k, user i is in main broadcaster
The discussion bar grade of user k is 0.
All bean vermicelli grades determined can be averaged by the present embodiment, obtain average bean vermicelli grade, and will put down
Equal bean vermicelli grade and noble's grade are averaged, and the result of acquisition can be used as the user gradation of user.
S240, concern number and bean vermicelli number according to user determine the first social grade of user.
Wherein, the first social grade can refer in the social circle formed by way of mutually paying close attention to, the society of user
Hand over grade.
Specifically, the concern number of user can be added by the present embodiment with bean vermicelli number, be will add up result and determined
For the first social grade of user.Illustratively, it obtains in all concern users of the user, pays close attention to number and bean vermicelli number
The sum of highest target pay close attention to user, and the sum of the concern number of the user and bean vermicelli number are paid close attention to user's divided by the target
The sum of concern number and bean vermicelli number, the result of acquisition is determined as the first social grade of user, to realize the unification of data
And normalization, convenient for calculating.
S250, the total number of forwards amount being posted by according to user to other users and total quantity is thumbed up, determines the of user
Two social grades.
Wherein, the second social grade can refer to the social grade of stealth that user carries out by way of forwarding and thumbing up.
Specifically, the second social grade S of user can be determined by following formula2:
S2=ml/α+mf
Wherein, mlIt is that user thumbs up total quantity to what other users were posted by；mfIt is that user is posted by other users
Total number of forwards amount；α is predetermined coefficient.In the present embodiment to thumb up total quantity often more much larger than total number of forwards amount, so as to
In a manner of by the way that total quantity will be thumbed up divided by predetermined coefficient, the magnitude for thumbing up total quantity and total number of forwards amount is matched, with
Just accuracy in computation is improved.Illustratively, the value range of predetermined coefficient α can be with are as follows: 6≤α≤10.
It should be noted that the present embodiment does not limit the sequence that executes of step S230-S250, such as step S230-
S250 can sequentially be executed after step S220, can also be executed in the front sequence in step S210.
S260, according to user gradation, the first social grade, the second social grade and publication quality degree, determine the work of user
Jerk.
Specifically, in discussion bar platform, user is posted by every time or the comment period of the day from 11 p.m. to 1 a.m can obtain corresponding number
According to so as to classify to all data generated in discussion bar, and determined respectively according to classification results based on data category
Corresponding user gradation, the first social grade, the second social grade and publication quality degree out.It should be noted that this implementation
Example is to determine aforementioned four parameter index based on all data present in discussion bar, that is to say, that only includes in discussion bar platform
This four parameter indexes, to can more accurately determine user's enlivening in discussion bar using this four parameter indexes simultaneously
Spend situation.Illustratively, user gradation, the first social grade, the second social grade and publication quality degree can be weighted
It is added, and will add up result and be determined as the activity of the user, it is more quasi- so as to further objectively embody user property
True determines the activity of the user, and then further increases recommendation effect.
Illustratively, the present embodiment can also determine according to the following formula the activity of the user:
Wherein, H is the activity of the user；J is user gradation；D is publication quality degree；S1It is the first social grade；S2It is
Two social grades；λ1It is the weighted value of user gradation；λ2It is the weighted value of the first social grade；λ3It is the weight for issuing quality degree
Value.
The present embodiment can be based on business scenario, improve to multivariate normal distributions, obtain above-mentioned active for determining
The formula (3) of degree, and make the value of liveness between 0 to 1.Illustratively, since the present embodiment is based on four parameters
Liveness is determined, so as in the probability density function of quaternary normal distributionOn the basis of improve, obtain above-mentioned formula
(3).It is possible, firstly, to by σ1As corresponding weighted value λ1, x1It is corresponding parameter index J that-μ, which is used as,.By in this present embodiment only
There are four parametric variables, so that the sum of weighted value of this four parametric variables is 1.And existing probability-distribution function be with
Variable be first incremented by and successively decrease afterwards, thus need byIt is improved toSo that calculated result can be with monotonic increase, the technology solved with this patent
Problem matches.Further, since being that the form of four multiplied by weight can in order to better reflect user property feature in denominator
With by σ1σ2σ3σ4As J × S1×D×S2Form.By way of e is rewritten as ln/lg, taking for liveness can be made
Value between 0 to 1, improve liveness variation stationarity, and pass through byIn plus 1,
The case where denominator is zero can be avoided result in.The present embodiment is relative to by user gradation, the first social grade, the second social grade
For being determined directly as the mode of liveness with the addition result of publication quality degree, by based on the thought of multivariate normal distributions
The hierarchy of the liveness of calculating is more obvious, and it is more objective to be distributed, and better reflects the difference of the liveness between different user
It is different, consequently facilitating corresponding Generalization bounds are arranged in the subsequent specific value for liveness, further increase the accuracy of recommendation
And effect.
In the present embodiment, user gradation, the first social grade, the second social grade and publication this four variables of quality degree
The sum of weighted value be 1.The size of each weight can be preset based on business scenario.Illustratively, since discussion bar is to use
Family is posted by and comments on the platform of model, thus avoiding being posted by and comment on model there are in the case where malice brush note
Quantity and degree be that most can directly reflect the size of user activity so that the publication matter in the present embodiment
Measure the weighted value highest of this parameter index, secondly, the forwarding quantity of model and thumb up quantity be user by note into
Row forwards and thumbs up behavior and improve, that is to say, that is to be influenced based on user itself behavior, so forwarding quantity and point
The recent liveness situation of user can be reflected to a certain extent by praising quantity, so that the weighted value of the second social grade takes second place.
Furthermore since user gradation is to need the behavior of certain time to accumulate just to can be improved, for example quantity of posting reaches certain journey
Grade just can be improved in degree, and what it reflected is the active situation of user's history, situation active in the recent period is unable to characterize, so user
The weighted value of grade is lower than the second social weighted value.Finally, the bean vermicelli number of user is not directly by user itself row
For raising, for example, user can't be the bean vermicelli of oneself, to can only reflect user activity situation indirectly, so that
The weighted value of one social grade is minimum.That is, the size relation of the weighted value of four variables can be set in the present embodiment
Are as follows: the social grade of weighted value > first of the weighted value > user gradation of the social grade of the weighted value > second of publication quality degree
Weighted value.For example, the weighted value of user gradation, the first social grade, publication quality degree and the second social grade can be distinguished
It is set as 0.2,0.1,0.4 and 0.3, obviously to symbolize the significance level of this four parameter indexes.
Illustratively, if the user gradation of user, the first social grade, publication quality degree and the second social grade are respectively
3,2,1,5, and λ1=0.2, λ2=0.1, λ3=0.4,1- λ1-λ2-λ3=0.3, then the activity of the user are as follows:It should be noted that
The present embodiment can also be final calculated result multiplied by 100, more can significantly embody the otherness between liveness.
S270, the target model for recommending user is determined according to liveness.
The technical solution of the present embodiment, by according to user gradation, the first social grade, the second social grade and publication matter
Measurement is common to determine that the activity of the user further increases liveness so as to further objectively embody user property
Determining accuracy, and then further increase recommendation effect.
It is the embodiment of model recommendation apparatus provided in an embodiment of the present invention, the note of the device and the various embodiments described above below
Sub- recommended method belongs to the same inventive concept, the detail content of not detailed description in the embodiment of model recommendation apparatus, can
With the embodiment with reference to above-mentioned model recommended method.
Embodiment three
Fig. 3 is a kind of structural schematic diagram for model recommendation apparatus that the embodiment of the present invention three provides, and the present embodiment is applicable
In to user carry out personalized recommendation model the case where, which specifically includes: post information obtain module 310, publication quality
Spend determining module 320 and target model determining module 330.
Wherein, post information obtains module 310, for obtaining the model text information and user comment that user is posted by
The comment text information of model；Quality degree determining module 320 is issued, is used for according to model text information and comment text information,
Determine the corresponding publication quality degree of user；Target model determining module 330, for determining enlivening for user according to publication quality degree
Degree, and the target model for recommending user is determined according to liveness.
Optionally, quality degree determining module 320 is issued, is specifically used for:
Character duplicate removal processing is carried out to each model text and each comment text of user's publication；
According to the corresponding forwarding quantity of character length, model text, the number of reviews in the model text after duplicate removal processing
And quantity is thumbed up, determine the corresponding model quality degree of each model text；
According to the corresponding reply quantity of character length, comment text in the comment text after duplicate removal processing and thumb up number
Amount, determines the corresponding comment quality degree of each comment text；
The corresponding publication quality degree of user is determined according to each model quality degree and each comment quality degree.
Optionally, the corresponding model quality degree of each model text is determined according to the following formula:
Wherein, DPRefer to the corresponding model quality degree of model text P；LpIt is to the model text after model text P duplicate removal processing
Character length in this；ncIt is the corresponding number of reviews of model text P；nfIt is the corresponding forwarding quantity of model text P；nlIt is note
This P of Ziwen is corresponding to thumb up quantity.
Optionally, target model determining module 330 includes:
User gradation determination unit determines the user gradation of user for the account grade and concern information according to user；
First social level de-termination unit determines the first of user for the concern number and bean vermicelli number according to user
Social grade；
Second social level de-termination unit, total number of forwards amount for being posted by according to user to other users and thumbs up
Total quantity determines the second social grade of user；
Liveness determination unit, for according to user gradation, the first social grade, the second social grade and publication quality
Degree, determines the activity of the user.
Optionally, user gradation determination unit is specifically used for:
According to the account grade of user, the bean vermicelli board grade of each main broadcaster user of user's concern and in each main broadcaster
The discussion bar grade of user determines user to the bean vermicelli grade of each main broadcaster user；
According to noble's grade of each bean vermicelli grade and user, the user gradation of user is determined.
Optionally, the activity of the user is determined according to the following formula:
Wherein, H is the activity of the user；J is user gradation；D is publication quality degree；S1It is the first social grade；S2It is
Two social grades；λ1It is the weighted value of user gradation；λ2It is the weighted value of the first social grade；λ3It is the weight for issuing quality degree
Value.
Optionally, target model determining module 330 further include: target model determination unit is used for:
Whether detection liveness is greater than default liveness threshold value；
If so, determine the model interested of user according to the historical viewings information of user, and using model interested as
Recommend the target model of user；
If it is not, then using the current highest model of hot topic degree as the target model for recommending user.
Model recommendation apparatus provided by the embodiment of the present invention can be performed model provided by any embodiment of the invention and push away
Method is recommended, has and executes the corresponding functional module of model recommended method and beneficial effect.
Example IV
Fig. 4 is a kind of structural schematic diagram for equipment that the embodiment of the present invention four provides.Referring to fig. 4, which includes:
One or more processors 410；
Memory 420, for storing one or more programs；
When one or more programs are executed by one or more processors 410, so that one or more processors 410 are realized
The model recommended method as provided by above-mentioned any embodiment, this method comprises:
Obtain the comment text information of model text information and user comment model that user is posted by；
According to model text information and comment text information, the corresponding publication quality degree of user is determined；
The activity of the user is determined according to publication quality degree, and the target model for recommending user is determined according to liveness.
In Fig. 4 by taking a processor 410 as an example；Processor 410 and memory 420 in equipment can by bus or its
He connects mode, in Fig. 4 for being connected by bus.
Memory 420 is used as a kind of computer readable storage medium, can be used for storing software program, journey can be performed in computer
Sequence and module, if the corresponding program instruction/module of model recommended method in the embodiment of the present invention is (for example, model recommends dress
Post information in setting obtains module 310, publication quality degree determining module 320 and target model determining module 330).Processor
410 software program, instruction and the modules being stored in memory 420 by operation, are answered thereby executing the various functions of equipment
With and data processing, that is, realize above-mentioned model recommended method.
Memory 420 mainly includes storing program area and storage data area, wherein storing program area can store operation system
Application program needed for system, at least one function；Storage data area, which can be stored, uses created data etc. according to equipment.This
Outside, memory 420 may include high-speed random access memory, can also include nonvolatile memory, for example, at least one
Disk memory, flush memory device or other non-volatile solid state memory parts.In some instances, memory 420 can be into one
Step includes the memory remotely located relative to processor 410, these remote memories can pass through network connection to equipment.On
The example for stating network includes but is not limited to internet, intranet, local area network, mobile radio communication and combinations thereof.
The model recommended method that the equipment and above-described embodiment that the present embodiment proposes propose belongs to same inventive concept, does not exist
The technical detail of detailed description can be found in above-described embodiment in the present embodiment, and the present embodiment has execution model recommended method
Identical beneficial effect.
Embodiment five
The present embodiment five provides a kind of computer readable storage medium, is stored thereon with computer program, which is located
It manages and realizes such as model recommended method provided by any embodiment of the invention when device executes, this method comprises:
Obtain the comment text information of model text information and user comment model that user is posted by；
According to model text information and comment text information, the corresponding publication quality degree of user is determined；
The activity of the user is determined according to publication quality degree, and the target model for recommending user is determined according to liveness.
The computer storage medium of the embodiment of the present invention, can be using any of one or more computer-readable media
Combination.Computer-readable medium can be computer-readable signal media or computer readable storage medium.It is computer-readable
Storage medium can be for example but not limited to: electricity, magnetic, optical, electromagnetic, infrared ray or semiconductor system, device or device, or
Any above combination of person.The more specific example (non exhaustive list) of computer readable storage medium includes: with one
Or the electrical connections of multiple conducting wires, portable computer diskette, hard disk, random access memory (RAM), read-only memory (ROM),
Erasable programmable read only memory (EPROM or flash memory), optical fiber, portable compact disc read-only memory (CD-ROM), light
Memory device, magnetic memory device or above-mentioned any appropriate combination.In this document, computer readable storage medium can
With to be any include or the tangible medium of storage program, the program can be commanded execution system, device or device use or
Person is in connection.
Computer-readable signal media may include in a base band or as carrier wave a part propagate data-signal,
Wherein carry computer-readable program code.The data-signal of this propagation can take various forms, including but unlimited
In electromagnetic signal, optical signal or above-mentioned any appropriate combination.Computer-readable signal media can also be that computer can
Any computer-readable medium other than storage medium is read, which can send, propagates or transmit and be used for
By the use of instruction execution system, device or device or program in connection.
The program code for including on computer-readable medium can transmit with any suitable medium, including but not limited to:
Wirelessly, electric wire, optical cable, RF etc. or above-mentioned any appropriate combination.
The computer for executing operation of the present invention can be write with one or more programming languages or combinations thereof
Program code, programming language include object oriented program language, and such as Java, Smalltalk, C++ further include
Conventional procedural programming language-such as " C " language or similar programming language.Program code can be fully
Execute, partly execute on the user computer on the user computer, being executed as an independent software package, partially with
Part executes on the remote computer or executes on a remote computer or server completely on the computer of family.It is being related to far
In the situation of journey computer, remote computer can pass through the network of any kind, including local area network (LAN) or wide area network
(WAN), it is connected to subscriber computer, or, it may be connected to outer computer (such as led to using ISP
Cross internet connection).
Will be appreciated by those skilled in the art that each module of the above invention or each step can use general meter
Device is calculated to realize, they can be concentrated on single computing device, or be distributed in network constituted by multiple computing devices
On, optionally, they can be realized with the program code that computer installation can be performed, so as to be stored in storage
It is performed by computing device in device, perhaps they are fabricated to each integrated circuit modules or will be more in them
A module or step are fabricated to single integrated circuit module to realize.In this way, the present invention is not limited to any specific hardware and
The combination of software.
Note that the above is only a better embodiment of the present invention and the applied technical principle.It will be appreciated by those skilled in the art that
The present invention is not limited to specific embodiments here, be able to carry out for a person skilled in the art it is various it is apparent variation, again
Adjustment and substitution are without departing from protection scope of the present invention.Therefore, although by above embodiments to the present invention carried out compared with
For detailed description, but the present invention is not limited to the above embodiments only, without departing from the inventive concept, can be with
Including more other equivalent embodiments, and the scope of the invention is determined by the scope of the appended claims.
Claims (10)
1. a kind of model recommended method characterized by comprising
Obtain the comment text information of model text information and the user comment model that user is posted by；
According to the model text information and the comment text information, the corresponding publication quality degree of the user is determined；
The activity of the user is determined according to the publication quality degree, and the user is recommended according to liveness determination
Target model.
2. the method according to claim 1, wherein being believed according to the model text information and the comment text
Breath, determines the corresponding publication quality degree of the user, comprising:
Character duplicate removal processing is carried out to each model text and each comment text of user publication；
According to the character length in the model text after duplicate removal processing, the corresponding forwarding quantity of the model text, number of reviews
And quantity is thumbed up, determine the corresponding model quality degree of each model text；
According to the corresponding reply quantity of character length, the comment text in the comment text after duplicate removal processing and thumb up number
Amount, determines the corresponding comment quality degree of each comment text；
The corresponding publication quality degree of the user is determined according to each model quality degree and each comment quality degree.
3. according to the method described in claim 2, it is characterized in that, determining that each model text is corresponding according to the following formula
Model quality degree:
Wherein, DPRefer to the corresponding model quality degree of model text P；LpIt is in the model text after model text P duplicate removal processing
Character length；ncIt is the corresponding number of reviews of model text P；nfIt is the corresponding forwarding quantity of model text P；nlIt is model text
This P is corresponding to thumb up quantity.
4. the method according to claim 1, wherein determining enlivening for the user according to the publication quality degree
Degree, comprising:
According to the account grade of the user and concern information, the user gradation of the user is determined；
According to the concern number and bean vermicelli number of the user, the first social grade of the user is determined；
The total number of forwards amount that is posted by according to the user to other users and total quantity is thumbed up, determines the second of the user
Social grade；
According to the user gradation, the first social grade, the second social grade and the publication quality degree, institute is determined
State the activity of the user.
5. according to the method described in claim 4, it is characterized in that, according to the account grade of the user and concern information, really
The user gradation of the fixed user, comprising:
According to the account grade of the user, the bean vermicelli board grade of each main broadcaster user of user concern and each
The discussion bar grade of the main broadcaster user determines the user to the bean vermicelli grade of each main broadcaster user；
According to noble's grade of each bean vermicelli grade and the user, the user gradation of the user is determined.
6. according to the method described in claim 4, it is characterized in that, determining the activity of the user according to the following formula:
Wherein, H is the activity of the user；J is the user gradation；D is the publication quality degree；S1It is first society
Hand over grade；S2It is the described second social grade；λ1It is the weighted value of the user gradation；λ2It is the power of the described first social grade
Weight values；λ3It is the weighted value of the publication quality degree.
7. -6 any method according to claim 1, which is characterized in that recommend the use according to liveness determination
The target model at family, comprising:
Detect whether the liveness is greater than default liveness threshold value；
If so, determining the model interested of the user according to the historical viewings information of the user, and will be described interested
Model is as the target model for recommending the user；
If it is not, then using the current highest model of hot topic degree as the target model for recommending the user.
8. a kind of model recommendation apparatus characterized by comprising
Post information obtains module, for obtaining commenting for model text information that user is posted by and the user comment model
By text information；
Quality degree determining module is issued, for determining the use according to the model text information and the comment text information
The corresponding publication quality degree in family；
Target model determining module, for determining the activity of the user according to the publication quality degree, and according to the work
Jerk determines the target model for recommending the user.
9. a kind of equipment, which is characterized in that the equipment includes:
One or more processors；
Memory, for storing one or more programs；
When one or more of programs are executed by one or more of processors, so that one or more of processors are real
The now model recommended method as described in any in claim 1-7.
10. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the program is by processor
The model recommended method as described in any in claim 1-7 is realized when execution.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910591329.2A CN110297987A (en) | 2019-07-02 | 2019-07-02 | A kind of model recommended method, device, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910591329.2A CN110297987A (en) | 2019-07-02 | 2019-07-02 | A kind of model recommended method, device, equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110297987A true CN110297987A (en) | 2019-10-01 |
Family
ID=68029999
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910591329.2A Pending CN110297987A (en) | 2019-07-02 | 2019-07-02 | A kind of model recommended method, device, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110297987A (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110990690A (en) * | 2019-11-12 | 2020-04-10 | 上海易点时空网络有限公司 | Post recommendation method and device |
| CN111770354A (en) * | 2020-07-02 | 2020-10-13 | 广州酷狗计算机科技有限公司 | Information interaction method, device, terminal and storage medium |
| CN111859196A (en) * | 2020-07-29 | 2020-10-30 | 北京达佳互联信息技术有限公司 | Information recommendation method and device, electronic equipment and storage medium |
| CN111949876A (en) * | 2020-08-14 | 2020-11-17 | 北京字节跳动网络技术有限公司 | Information processing method and device, electronic equipment and computer readable storage medium |
| CN112153426A (en) * | 2020-09-21 | 2020-12-29 | 腾讯科技（深圳）有限公司 | Content account management method and device, computer equipment and storage medium |
| CN112699659A (en) * | 2020-12-31 | 2021-04-23 | 五八有限公司 | Method and device for identifying false information of house source, electronic equipment and storage medium |
| CN112732780A (en) * | 2020-12-30 | 2021-04-30 | 深圳市网联安瑞网络科技有限公司 | Figure network activity degree calculation method, system, processing terminal and computer equipment |
| CN113836437A (en) * | 2021-09-14 | 2021-12-24 | 上海任意门科技有限公司 | Method, electronic device, and storage medium for post recommendation |
| CN114860969A (en) * | 2022-05-19 | 2022-08-05 | 北京字跳网络技术有限公司 | Information display method and device, electronic equipment and storage medium |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106294363A (en) * | 2015-05-15 | 2017-01-04 | 厦门美柚信息科技有限公司 | A kind of forum postings evaluation methodology, Apparatus and system |
| CN106570763A (en) * | 2016-11-09 | 2017-04-19 | 福建中金在线信息科技有限公司 | User influence evaluation method and system |
| CN106874314A (en) * | 2015-12-14 | 2017-06-20 | 腾讯科技（深圳）有限公司 | The method and apparatus of information recommendation |
| US20170308589A1 (en) * | 2016-04-26 | 2017-10-26 | Facebook, Inc. | Recommendations from Comments on Online Social Networks |
| CN107562836A (en) * | 2017-06-07 | 2018-01-09 | 北京航空航天大学 | Method is recommended based on the answerer of topic model and machine learning |
| CN109408726A (en) * | 2018-11-09 | 2019-03-01 | 大连海事大学 | Question answering person's recommended method in question and answer website |
| CN109446427A (en) * | 2018-11-02 | 2019-03-08 | 北京奇艺世纪科技有限公司 | A kind of information recommendation method and device |
| CN109711875A (en) * | 2018-12-19 | 2019-05-03 | 口碑(上海)信息技术有限公司 | Content recommendation method and device |
| CN109862013A (en) * | 2019-01-31 | 2019-06-07 | 武汉斗鱼网络科技有限公司 | A kind of direct broadcasting room recommended method, storage medium, electronic equipment and system |
| CN109871491A (en) * | 2019-03-20 | 2019-06-11 | 江苏满运软件科技有限公司 | Forum postings recommended method, system, equipment and storage medium |
-
2019
- 2019-07-02 CN CN201910591329.2A patent/CN110297987A/en active Pending
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106294363A (en) * | 2015-05-15 | 2017-01-04 | 厦门美柚信息科技有限公司 | A kind of forum postings evaluation methodology, Apparatus and system |
| CN106874314A (en) * | 2015-12-14 | 2017-06-20 | 腾讯科技（深圳）有限公司 | The method and apparatus of information recommendation |
| US20170308589A1 (en) * | 2016-04-26 | 2017-10-26 | Facebook, Inc. | Recommendations from Comments on Online Social Networks |
| CN106570763A (en) * | 2016-11-09 | 2017-04-19 | 福建中金在线信息科技有限公司 | User influence evaluation method and system |
| CN107562836A (en) * | 2017-06-07 | 2018-01-09 | 北京航空航天大学 | Method is recommended based on the answerer of topic model and machine learning |
| CN109446427A (en) * | 2018-11-02 | 2019-03-08 | 北京奇艺世纪科技有限公司 | A kind of information recommendation method and device |
| CN109408726A (en) * | 2018-11-09 | 2019-03-01 | 大连海事大学 | Question answering person's recommended method in question and answer website |
| CN109711875A (en) * | 2018-12-19 | 2019-05-03 | 口碑(上海)信息技术有限公司 | Content recommendation method and device |
| CN109862013A (en) * | 2019-01-31 | 2019-06-07 | 武汉斗鱼网络科技有限公司 | A kind of direct broadcasting room recommended method, storage medium, electronic equipment and system |
| CN109871491A (en) * | 2019-03-20 | 2019-06-11 | 江苏满运软件科技有限公司 | Forum postings recommended method, system, equipment and storage medium |
Non-Patent Citations (1)
| Title |
|---|
| LITTLEMAGIC: "设计一个属于自己的内容热度值算法", 《HTTP://EVENTS.JIANSHU.IO/P/FB454A4B383D》, 15 April 2019 (2019-04-15) * |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110990690A (en) * | 2019-11-12 | 2020-04-10 | 上海易点时空网络有限公司 | Post recommendation method and device |
| CN111770354B (en) * | 2020-07-02 | 2022-10-18 | 广州酷狗计算机科技有限公司 | Information interaction method, device, terminal and storage medium |
| CN111770354A (en) * | 2020-07-02 | 2020-10-13 | 广州酷狗计算机科技有限公司 | Information interaction method, device, terminal and storage medium |
| CN111859196A (en) * | 2020-07-29 | 2020-10-30 | 北京达佳互联信息技术有限公司 | Information recommendation method and device, electronic equipment and storage medium |
| CN111859196B (en) * | 2020-07-29 | 2023-11-28 | 北京达佳互联信息技术有限公司 | Information recommendation method and device, electronic equipment and storage medium |
| CN111949876A (en) * | 2020-08-14 | 2020-11-17 | 北京字节跳动网络技术有限公司 | Information processing method and device, electronic equipment and computer readable storage medium |
| CN112153426A (en) * | 2020-09-21 | 2020-12-29 | 腾讯科技（深圳）有限公司 | Content account management method and device, computer equipment and storage medium |
| CN112153426B (en) * | 2020-09-21 | 2023-08-29 | 腾讯科技（深圳）有限公司 | Content account management method and device, computer equipment and storage medium |
| CN112732780A (en) * | 2020-12-30 | 2021-04-30 | 深圳市网联安瑞网络科技有限公司 | Figure network activity degree calculation method, system, processing terminal and computer equipment |
| CN112732780B (en) * | 2020-12-30 | 2024-04-02 | 深圳市网联安瑞网络科技有限公司 | Character network liveness calculation method, system, processing terminal and computer equipment |
| CN112699659B (en) * | 2020-12-31 | 2022-09-30 | 五八有限公司 | Method and device for identifying false information of house source, electronic equipment and storage medium |
| CN112699659A (en) * | 2020-12-31 | 2021-04-23 | 五八有限公司 | Method and device for identifying false information of house source, electronic equipment and storage medium |
| CN113836437A (en) * | 2021-09-14 | 2021-12-24 | 上海任意门科技有限公司 | Method, electronic device, and storage medium for post recommendation |
| CN113836437B (en) * | 2021-09-14 | 2024-01-30 | 上海任意门科技有限公司 | Method for post recommendation, electronic device and storage medium |
| CN114860969A (en) * | 2022-05-19 | 2022-08-05 | 北京字跳网络技术有限公司 | Information display method and device, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110297987A (en) | A kind of model recommended method, device, equipment and storage medium | |
| CN107332902B (en) | The user of online customer service system asks distribution method, device and computing device | |
| CN108363602B (en) | Intelligent UI (user interface) layout method and device, terminal equipment and storage medium | |
| CN109190024A (en) | Information recommendation method, device, computer equipment and storage medium | |
| CN104123284B (en) | The method and server of a kind of recommendation | |
| CN108632311A (en) | Information-pushing method and device | |
| WO2014193700A1 (en) | Social media pricing engine | |
| CN107229754B (en) | Information sorting method and device, electronic equipment and storage medium | |
| CN110490625A (en) | User preference determines method and device, electronic equipment, storage medium | |
| CN110111152A (en) | A kind of content recommendation method, device and server | |
| CN109190028A (en) | Activity recommendation method, apparatus, electronic equipment and storage medium based on big data | |
| CN110134783A (en) | Method, apparatus, equipment and the medium of personalized recommendation | |
| CN108470023A (en) | The recommendation method and device of business function | |
| CN109087162A (en) | Data processing method, system, medium and calculating equipment | |
| CN107729473B (en) | Article recommendation method and device | |
| CN109389365A (en) | Processing method, device and the electronic equipment of multiple person cooperational formula document | |
| CN109783740A (en) | Pay close attention to the sort method and device of the page | |
| KR102097045B1 (en) | Method and apparatus to recommend products reflecting characteristics of users | |
| CN116541610A (en) | Training method and device for recommendation model | |
| CN112184302A (en) | Product recommendation method and device, rule engine and storage medium | |
| CN109064283A (en) | Method of Commodity Recommendation and device, computer readable storage medium | |
| CN110245684A (en) | Data processing method, electronic equipment and medium | |
| CN109740066A (en) | Information recommendation method, information recommending apparatus, storage medium and electronic equipment | |
| CN110020112A (en) | Object Push method and its system | |
| CN109993566A (en) | A kind of method and apparatus for predicting product objective data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20191001 |