CN108960958A - item recommendation method and device - Google Patents
item recommendation method and device Download PDFInfo
- Publication number
- CN108960958A CN108960958A CN201810395304.0A CN201810395304A CN108960958A CN 108960958 A CN108960958 A CN 108960958A CN 201810395304 A CN201810395304 A CN 201810395304A CN 108960958 A CN108960958 A CN 108960958A
- Authority
- CN
- China
- Prior art keywords
- recommendation
- article
- recommending
- sequence
- target item
- 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
- 238000000034 method Methods 0.000 title claims abstract description 44
- 230000003542 behavioural effect Effects 0.000 claims abstract description 11
- 238000011156 evaluation Methods 0.000 description 46
- 230000000694 effects Effects 0.000 description 6
- 239000012141 concentrate Substances 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 244000097202 Rathbunia alamosensis Species 0.000 description 1
- 235000009776 Rathbunia alamosensis Nutrition 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0631—Item recommendations
Abstract
This application discloses a kind of item recommendation method and devices.The method includes the recommendation factor of N number of target item and corresponding each target item is obtained according to user's history behavioral data, wherein N is the integer greater than 1;Multiple recommendation articles and corresponding first recommendation of each recommendation article are obtained according to N number of target item, and obtain each the first recommendation sequence for recommending article, wherein each recommendation article is related at least one target item;Data processing is carried out according to the sequence of the first recommendation and corresponding first recommendation of each recommendation article of the corresponding recommendation factor pair of each target item, corresponding second recommendation of each recommendation article is obtained, to obtain each the second recommendation sequence for recommending article.Present application addresses the very high articles of correlation to recommend user with largely being flocked together, and causes the technical problem of the reading fatigue of user.
Description
Technical field
This application involves data application technical fields, in particular to a kind of item recommendation method and device.
Background technique
With the high speed development of internet, the application scenarios of item recommendation method are more and more, are integrated into extensively
In many commercial systems for applications, it is more famous have Netflix Online Video recommender system, Amazon shopping at network store,
Today's tops etc..In fact, most of information platform all applies item recommendation method to some extent, as Sina News,
Today's tops etc..The associated recommendation that item recommendation method provides can help user preferably to find the relevant object with current item
Product, in electric business platform, relevant commodity can be with bundle sale, in information platform, and the correlation that item recommendation method is recommended provides
News can be used as depth is read or interest is found etc., be the very important applied field for embodying recommender system value
Scape.
In the prior art, item recommendation method is recommended based on user behavior, is normally based on the history row of user
For record, determines the similarity between article, recommended further according to article similarity.
During implementing the embodiment of the present application, inventor has found that the prior art at least has following technical problem:
In the prior art, the article about a topic is possible to largely occur a moment, between these articles
Correlation can be very high, in this way, the very high article of these correlations can recommend user with largely being flocked together, cause readding for user
Fatigue is read, the experience effect of user is influenced.
Summary of the invention
The main purpose of the application is to provide a kind of item recommendation method and device, to solve the very high article of correlation
User can be recommended with largely being flocked together, cause the problem of the reading fatigue of user.
To achieve the goals above, in a first aspect, the embodiment of the present application provides a kind of item recommendation method.
Include: according to the item recommendation method of the application
The recommendation factor of N number of target item and corresponding each target item is obtained according to user's history behavioral data,
In, N is the integer greater than 1;
Multiple recommendation articles and corresponding first recommendation of each recommendation article are obtained according to N number of target item, and are obtained
To each the first recommendation sequence for recommending article, wherein each recommendation article is related at least one target item;
It is corresponding according to the sequence of the first recommendation and each recommendation article of the corresponding recommendation factor pair of each target item
First recommendation carries out data processing, corresponding second recommendation of each recommendation article is obtained, to obtain each recommendation article
The second recommendation sequence.
Optionally, according to the sequence of the first recommendation and each recommendation article of the corresponding recommendation factor pair of each target item
Corresponding first recommendation carries out data processing, obtains corresponding second recommendation of each recommendation article, comprising:
The corresponding recommendation factor of each recommendation article is determined to relevant target item according to each recommendation article;
According to the first recommendation sequence and the corresponding recommendation factor of each recommendation article, obtain each recommendation relative to
Corresponding recommendation factor sequence determines serial number of each recommendation article in corresponding recommendation factor sequence;
According to serial number of each recommendation article in corresponding recommendation factor sequence to each recommendation article corresponding first
Recommendation carries out data processing, obtains corresponding second recommendation of each recommendation article.
Optionally, when recommending article, there are when at least two corresponding second recommendations, choose at least the two of recommendation article
Maximum second recommendation is as corresponding second recommendation of recommendation article in a second recommendation.
Optionally, each second recommendation for recommending article is ordered as each second recommendation for recommending article
Descending sort, this item recommendation method further include:
In each the second recommendation sequence for recommending article, obtains each recommendation condition and article is recommended to previous position
Similarity value, wherein the similarity value that the recommendation article to make number one includes is the first preset value;
Data processing is carried out to each the second recommendation for recommending article according to the similarity value that each recommendation article includes,
Each third recommendation for recommending article is obtained, to obtain each third recommendation sequence for recommending article.
Optionally, this item recommendation method further include:
Each generation time for recommending article is obtained, and according to current time and each generation time pair for recommending article
Each third recommendation for recommending article carries out data processing, obtains each the 4th recommendation for recommending article, to obtain every
A the 4th recommendation sequence for recommending article.
Second aspect, a kind of article recommendation apparatus provided by the embodiments of the present application, comprising:
First obtains module, for obtaining N number of target item and corresponding each target according to user's history behavioral data
The recommendation factor of article, wherein N is the integer greater than 1;
Second obtains module, for corresponding according to N number of target item multiple recommendation articles of acquisition and each recommendation article
The first recommendation, and obtain each the first recommendation sequence for recommending article, wherein each recommendations article and at least one mesh
It is related to mark article;
Third obtains module, is each pushed away according to the sequence of the first recommendation and the corresponding recommendation factor pair of each target item
It recommends corresponding first recommendation of article and carries out data processing, corresponding second recommendation of each recommendation article is obtained, to obtain
Each the second recommendation sequence for recommending article.
Optionally, third obtains module, is used for:
The corresponding recommendation factor of each recommendation article is determined to relevant target item according to each recommendation article;
According to the first recommendation sequence and the corresponding recommendation factor of each recommendation article, obtain each recommendation relative to
Corresponding recommendation factor sequence determines serial number of each recommendation article in corresponding recommendation factor sequence;
According to serial number of each recommendation article in corresponding recommendation factor sequence to each recommendation article corresponding first
Recommendation carries out data processing, obtains corresponding second recommendation of each recommendation article.
Optionally, third obtain module, be also used to when recommend article exist when at least two corresponding second recommendations, select
Maximum second recommendation at least two second recommendations for recommending article is taken to be used as corresponding second recommendation of recommendation article.
Optionally, this article recommendation apparatus further includes that the 4th acquisition module and the 5th obtain module;Each recommendation
Second recommendation of product is ordered as each second recommendation descending sort for recommending article;
4th obtains module, for obtaining each recommendation condition in each the second recommendation sequence for recommending article
Recommend previous position the similarity value of article, wherein the similarity value that the recommendation article to make number one includes is first default
Value;
5th obtains module, and the similarity value for including according to each recommendation article recommends the second of article to push away to each
It recommends value and carries out data processing, obtain each third recommendation for recommending article, recommend to obtain each third for recommending article
Value sequence.
Optionally, this article recommendation apparatus further includes the 6th acquisition module;
6th obtains module, for obtaining each generation time for recommending article, and according to current time and each pushes away
The time that generates for recommending article carries out data processing to each third recommendation for recommending article, obtains the 4th of each recommendation article
Recommendation, to obtain each the 4th recommendation sequence for recommending article.
The item recommendation method provided in the embodiment of the present application, by obtaining N number of mesh according to user's history behavioral data
Mark the recommendation factor of article and corresponding each target item, wherein N is the integer greater than 1;It is obtained according to N number of target item
Multiple recommendation articles and corresponding first recommendation of each recommendation article, and obtain each the first recommendation row for recommending article
Sequence, wherein each recommendation article is related at least one target item;According to the sequence of the first recommendation and each target item
Corresponding first recommendation of the corresponding each recommendation article of recommendation factor pair carries out data processing, obtains each recommendations article correspondence
The second recommendation, thus obtain it is each recommend article the second recommendation sequence.In this way, being pushed away according to the recommendation factor and first
Value sequence is recommended, recommendation article very high to correlation is handled, and has reached multiple recommendation dimensions to the recommendation for recommending article
Operation is carried out, the multifarious purpose of topic of recommendation results is improved, to realize the diversity for recommending article, improves use
The technical effect of the reading experience at family, and then solve the very high article of correlation and can recommend user with largely being flocked together, it makes
At the technical problem of the reading fatigue of user.
Detailed description of the invention
The attached drawing constituted part of this application is used to provide further understanding of the present application, so that the application's is other
Feature, objects and advantages become more apparent upon.The illustrative examples attached drawing and its explanation of the application is for explaining the application, not
Constitute the improper restriction to the application.In the accompanying drawings:
Fig. 1 is the flow chart according to a kind of item recommendation method of the embodiment of the present application;
Fig. 2 is the flow chart according to a kind of step S300 of the embodiment of the present application;
Fig. 3 is the flow chart according to another item recommendation method of the embodiment of the present application;
Fig. 4 is the flow chart according to a kind of step S500 of the embodiment of the present application;
Fig. 5 is the flow chart according to another item recommendation method of the embodiment of the present application;
Fig. 6 is the flow chart according to a kind of step S600 of the embodiment of the present application;
Fig. 7 is the flow chart according to a kind of step S620 of the embodiment of the present application;
Fig. 8 is the flow chart according to a kind of step S621 of the embodiment of the present application;
Fig. 9 is the flow chart according to a kind of step S200 of the embodiment of the present application;
Figure 10 is the structural schematic diagram according to a kind of article recommendation apparatus of the embodiment of the present application;
Figure 11 is the structural schematic diagram according to another article recommendation apparatus of the embodiment of the present application;
Figure 12 is the structural schematic diagram according to another article recommendation apparatus of the embodiment of the present application.
Specific embodiment
In order to make those skilled in the art more fully understand application scheme, below in conjunction in the embodiment of the present application
Attached drawing, the technical scheme in the embodiment of the application is clearly and completely described, it is clear that described embodiment is only
The embodiment of the application a part, instead of all the embodiments.Based on the embodiment in the application, ordinary skill people
Member's every other embodiment obtained without making creative work, all should belong to the model of the application protection
It encloses.
It should be noted that in the absence of conflict, the features in the embodiments and the embodiments of the present application can phase
Mutually combination.The application is described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.
The embodiment of the present application provides a kind of item recommendation method, as shown in Figure 1, this method includes the following steps, namely S100
To step S300:
S100, according to user's history behavioral data obtain the recommendation of N number of target item and corresponding each target item because
Son, wherein N is the integer greater than 1.
In the present embodiment, target item may be the information in information platform, such as the target item can be one
A sports news or entertainment news etc.;Target item may be the video in video platform, for example, the target item can be
One movie or television play etc.;The target item can be the commodity or service item in electric business platform, for example, the target item
It can be electronic product or the reference service project etc. on certain electric business platform.Correspondingly, recommendation relevant to the target item
Article can be the information in information platform;Or video platform in video, can also in electric business platform commodity or
Service item.
In an implementation, N number of target item is obtained according to user's history behavioral data, specifically: it is browsed using user's history
In nearest operation note (the operation article on the same day), according to category information, correlation and the operating time of these operation articles,
Target item of the M operation article therefrom extracted as recent interest;(3~5 are generally used using user's history browsing data
It operation article), according to these operation article category informations and correlation, the Q operation article therefrom extracted is as history
The target item of interest, wherein M and Q is the integer more than or equal to 0, and M+Q=N.And it is generated for each target item
Recommend the factor, each recommendation factor can be a kind of topic (such as the topics such as related to some star in amusement circle), be also possible to industry
Information of being engaged in or posterior infromation.Wherein, business information can be demand setting etc. in business, for example, need to certain topic phases
Article is closed to promote etc.;Posterior infromation can be historical experience, such as when the title of an article includes that some is specific
It include " Mei Xi " in content when word such as the target item is sports news, then the recommendation factor is and " Mei Xi " phase
It closes.
Wherein, the value of N can be 2,3,4,5,6 ...
S200 obtains multiple recommendation articles and corresponding first recommendation of each recommendation article according to N number of target item,
And obtain each the first recommendation sequence for recommending article, wherein each recommendation article is related at least one target item.
In an implementation, multiple methods for recommending article relevant to target item are obtained, can be based on commending contents side
Multiple recommendation articles relevant to target item that method generates, can be according to user collaborative recommended method or article collaborative party
Multiple recommendation articles relevant to target item that method generates are also possible to according to the generation of business rule recommended method and target
The relevant multiple recommendation articles of article can also be the union that above-mentioned each method generates, and should and concentrate raw including above-mentioned each method
At multiple recommendation articles relevant to target item, pass through the above method and obtain relevant to target item multiple recommendation articles
While, it can also obtain the corresponding recommendation of each recommendation article.
S300, according to the sequence of the first recommendation and each recommendation article pair of the corresponding recommendation factor pair of each target item
The first recommendation answered carries out data processing, corresponding second recommendation of each recommendation article is obtained, to obtain each recommendation
Second recommendation of article sorts.
In the present embodiment, because each recommendation article corresponds at least one target item, and each target item is corresponding
One recommendation factor, therefore, each recommendation article correspond at least one and recommend the factor, further according to recommendation article in the first recommendation
Ordinal position and corresponding the first recommendation for recommending the factor pair recommendation article in sequence carry out operation, obtain the recommendation
First recommendation of article.And then obtain each the second recommendation sequence for recommending article.In this way, according to the factor and the is recommended
The sequence of one recommendation, recommendation article very high to correlation are handled, and have been reached multiple recommendation dimensions and have been pushed away to recommendation article
It recommends value and carries out operation, improve the multifarious purpose of topic of recommendation results, to realize the diversity for recommending article, promoted
The technical effect of the reading experience of user.
As shown in Fig. 2, optionally, S300, according to the sequence of the first recommendation and the corresponding recommendation of each target item because
Son carries out data processing to corresponding first recommendation of each recommendation article, obtains each recommendation article corresponding second and recommends
Value includes the following steps S310 to S330 to obtain each the second recommendation sequence for recommending article:
S310 determines the corresponding recommendation factor of each recommendation article to relevant target item according to each recommendation article.
In the present embodiment, each recommendation article corresponds at least one target item, and each target item is one corresponding
Recommend the factor, therefore, each recommendation article corresponds at least one and recommends the factor.
S320 is sorted according to the first recommendation and the corresponding recommendation factor of each recommendation article, obtains each recommendation phase
It sorts for the corresponding recommendation factor, determines serial number of each recommendation article in corresponding recommendation factor sequence.
In the present embodiment, it according to ordinal position of the article in the sequence of the first recommendation is recommended, determines and corresponds to identical push away
The sequence between multiple recommendation articles of the factor is recommended, so that it is determined that each sequence for recommending article in corresponding recommendation factor sequence
Number.
For example, the recommendation article of the corresponding recommendation factor shares 3, respectively first recommends article, the second recommendation
Product and third recommend article, wherein the first recommendation article, the second recommendation article and third recommend article to sort in the first recommendation
In ordinal position be respectively the 2nd, the 9th and the 4th, then 3 recommendation conditions are for the corresponding recommendation factor
The serial number of sequence is respectively as follows: the first recommendation article and corresponds to serial number 1, and the second recommendation article corresponds to serial number 3, and third recommends article corresponding
Serial number 2.
S330, it is corresponding to each recommendation article according to serial number of each recommendation article in corresponding recommendation factor sequence
First recommendation carries out data processing, corresponding second recommendation of each recommendation article is obtained, to obtain each recommendation article
The second recommendation.
In the present embodiment, corresponding to recommendation article according to serial number of the recommendation article in corresponding recommendation factor sequence
First recommendation carries out data processing, which can be realized by following formula:
In formula, rec_score on the left of equal signiIt is second recommendation of i-th of recommendation article, rec_ on the right side of equal sign
scoreiIt is first recommendation of i-th of recommendation article,It is that i-th of recommendation article sorts in the corresponding recommendation factor
In serial number (or when the first recommendation be ordered as descending arrangement when, recommend the frequency of occurrence of the factor), constrsnIt is each push away
The corresponding predetermined constant of the factor is recommended, optionally, which can be according to the different numerical value for recommending predictor selection different, should
Predetermined constant can also use a default fixed numbers, such as can be with value for 1.
Further optionally, when recommending article, there are when at least two corresponding second recommendations, choose recommendation article extremely
Maximum second recommendation is as corresponding second recommendation of recommendation article in few two the second recommendations.
As shown in figure 3, optionally, each second recommendation for recommending article is ordered as each recommendation article
Second recommendation descending sort, this item recommendation method further include following steps S400 and step 500:
S400 obtains each recommendation condition and recommends previous position in each the second recommendation sequence for recommending article
The similarity value of article, wherein the similarity value that the recommendation article to make number one includes is the first preset value.
In the present embodiment, it comes primary recommendation article in the second recommendation descending sort and is pushed away because its front has no
Article is recommended, therefore, the similarity value of the recommendation article to make number one can be set as first preset value, wherein can
Selection of land, the value of the first preset value can be 0.
In an implementation, when obtaining similarity value of each recommendation condition to the recommendation article of previous position, can pass through
The similarity value for recommending article that the titles (title) that two adjacent recommendation articles include calculate recommendation conditions to previous position,
Wherein, calculation can use existing similarity calculating method.
S500 carries out data to each the second recommendation for recommending article according to the similarity value that each recommendation article includes
Processing obtains each third recommendation for recommending article, to obtain each third recommendation sequence for recommending article.
In the present embodiment, each recommendation article corresponding second is pushed away by the similarity value that each recommendation article includes
It recommends value and carries out data processing, corresponding second recommendation of each recommendation article is adjusted according to content information to realize
It is whole, and then generate each the first recommendation for recommending article.In this way, the content phase of two adjacent recommendation articles can be prevented
Seemingly, and then the recommendation of influence user is experienced.
As shown in figure 4, optionally, S500, the similarity value for including according to each recommendation article is to each recommendation article
Second recommendation carries out data processing, obtains each third recommendation for recommending article, to obtain each recommending the of article
The sequence of three recommendations, includes the following steps S510 and step S520:
S510 obtains each recommendation condition to the recommendation of previous position according to the similarity value that each recommendation article includes
The dissimilar angle value of product.
S520, according to each recommendation article dissimilar angle value for including to corresponding second recommendation of each recommendation article into
Row data processing obtains each third recommendation for recommending article, to obtain each third recommendation sequence for recommending article.
In the present embodiment, the third recommendation for recommending article can be recommendation article of the recommendation condition to previous position
Dissimilar angle value the second recommendation corresponding with the recommendation article between product.
For example, corresponding second recommendation of a certain recommendation article is 0.9, recommendation article of the recommendation condition to previous position
Dissimilar angle value be 0.3, then, the third recommendation of the recommendation article is 0.27 (0.9 and 0.3 product).
Optionally, recommend the dissimilar angle value of article less than the second preset value previous position when there are recommendation conditions
When, the second preset value is chosen as the dissimilar degree for recommending article.
In an implementation, the second preset value of setting can be to prevent recommendation condition to the dissmilarity of the recommendation article of previous position
Angle value is too small, causes the third recommendation of the recommendation article too small, and then influences other subsequent steps to recommendation article processing
Operation.
In the present embodiment, each recommendation article corresponding second is pushed away according to the similarity value that each recommendation article includes
It recommends value and carries out data processing, obtain each third recommendation for recommending article, wherein the data processing can pass through following formula
It realizes:
rec_scorei=rec_scorei*max((1-sim(itemi,itemi-1)),a)
Wherein, rec_score on the left of equal signiIt is the third recommendation of i-th of recommendation article, rec_score on the right side of equal signi
It is second recommendation of i-th of recommendation article, sim (itemi,itemi-1) it is recommendation of i-th of recommendation condition to previous position
The similarity value of article, (1-sim (itemi,itemi-1)) it is not phase of i-th of recommendation condition to the recommendation article of previous position
Like angle value, a is the second preset value, wherein the value of the second preset value can be 0.08,0.1,0.2 etc., the technology of this field
Personnel can be specifically arranged according to actual needs.
As shown in figure 5, optionally, this item recommendation method further includes following steps S600:
S600 obtains each generation time for recommending article, and according to current time and each generation for recommending article
Time carries out data processing to each third recommendation for recommending article, obtains each the 4th recommendation for recommending article, thus
Obtain each the 4th recommendation sequence for recommending article.
In the present embodiment, the generation time for recommending article can be the uniform resource locator of the recommendation article
(Uniform Resource Locator, URL) generates the time;Current time is the time for generating and recommending.
In an implementation, each third for recommending article is pushed away according to current time and each generation time for recommending article
It recommends value and carries out data processing, therefore, during carrying out data processing to each third recommendation for recommending article, be added to
Influence of the time factor to third recommendation, it is contemplated that the timeliness for recommending article has reached the recommendation that timeliness is good
Product recommend the purpose of user, recommend to realize timeliness is good, high with the interest correlation of user recommendation article
User improves technical effect of the user to the interest for recommending article.
As shown in fig. 6, optionally, S600 is pushed away according to current time and each generation time for recommending article to each
The third recommendation for recommending article carries out data processing, each the 4th recommendation for recommending article is obtained, to obtain each recommendation
4th recommendation of article sorts, and includes the following steps S610 to step S630:
S610, when obtaining each generation for recommending article according to current time and each generation time for recommending article
It is long.
In an implementation, the generation time for recommending article is subtracted according to current time, it can obtain the life of the recommendation article
At duration, i.e. as being generated to time span existing for current time, the chronomere of the generation duration can be the recommendation article
Second, minute, hour, day etc..
For example, it is a sports news about Mei Xi that certain, which recommends article, the time should be generated about the sports news of Mei Xi
35 divide 18 seconds when for 17 days 18 April in 2018, and 45 divide 30 seconds when current time is 19 days 19 April in 2018, then should be about Mei Xi
Sports news generation duration be specially 1 hour 2 days 10 points 12 seconds, should body about Mei Xi when chronomere is hour
It educates when the generation of news a length of 49.17 hours, when chronomere is the second, is somebody's turn to do a length of when the generation about the sports news of Mei Xi
177012 seconds.
S620 obtains each recommendation according to the first preset duration evaluation of estimate and each generation duration for recommending article
The timeliness coefficient of product.
In the present embodiment, which is a preset value, can be configured according to actual needs
Or adjustment, the first preset duration evaluation of estimate can be a constant, be also possible to the value with chronomere.
In an implementation, the timeliness coefficient of article is recommended to evaluate due to the production duration of the recommendation article and the first preset duration
Value carries out operation and obtains.
For example, it is the social news about ecological environment that certain, which recommends article, it is somebody's turn to do the social news about ecological environment
Generation when it is 30 minutes a length of, the first preset duration evaluation of estimate be 1440 minutes, the operation mode be first preset duration
The ratio between with the generation duration, then, should be 48 about the timeliness coefficient of the social news of ecological environment (1440 minutes with 30 minutes
The ratio between).
S630 carries out operation to each third recommendation for recommending article according to each timeliness coefficient for recommending article, obtains
Each the 4th recommendation for recommending article is taken, to obtain each the 4th recommendation sequence for recommending article.
In the present embodiment, operation is carried out according to third recommendation of the timeliness coefficient of recommendation article to the recommendation article,
To obtain the 4th recommendation of the recommendation article.
For example, the operation mode is to recommend the product of the third recommendation of timeliness coefficient and the recommendation article of article, it should
The timeliness coefficient for recommending article is 2, and the third recommendation of the recommendation article is 0.9001, then, the 4th of the recommendation article pushes away
Recommending value is 1.8002 (2 multiplied by 0.9001).
As shown in fig. 7, optionally, step S620, according to the first preset duration evaluation of estimate and each life for recommending article
At duration, each timeliness coefficient for recommending article is obtained, includes the following steps S621 to step S622:
S621 obtains each the first duration evaluation of estimate for recommending article according to each generation duration for recommending article.
In the present embodiment, according to the evaluation of estimate for recommending the generation duration available correspondence of the article generation duration, i.e.,
For the first duration evaluation of estimate, which can be by obtaining the generation duration progress data processing for recommending article
It takes, for example, the generation duration of article will be recommended divided by a unit time, to obtain corresponding first duration of the generation duration
Evaluation of estimate.
For example, 4 days a length of when a certain generation for recommending article, the unit time is day, then the first duration evaluation of estimate can
Think 4.
S622 is each pushed away according to the first preset duration evaluation of estimate and each the ratio between the first duration evaluation of estimate for recommending article
Recommend the timeliness coefficient of article.
In the present embodiment, recommend the timeliness coefficient of article can the according to first preset duration evaluation of estimate and recommendation article
The ratio that obtains of the ratio between the first duration evaluation of estimate.
For example, the first preset duration evaluation of estimate is 1000, a certain the first duration evaluation of estimate for recommending article is 5, then, it should
The timeliness coefficient for recommending article is 200 (1000 divided by 5).
As shown in figure 8, optionally, step S621, each generation duration for recommending article of root obtains each recommendation article
First duration evaluation of estimate includes the following steps S6211 to step S6212:
S6211 obtains the second of each recommendation article according to the ratio between each generation duration for recommending article and unit time
Duration evaluation of estimate.
In the present embodiment, recommend article the second duration evaluation of estimate can according to the generation duration of the recommendation article with
The ratio that the ratio between unit time obtains.
For example, 30 hours a length of when a certain generation for recommending article, the unit time is hour, then the of the recommendation article
Two duration evaluations of estimate are 30, in another example, a length of 0 when another generation for recommending article, i.e. the recommendation article is raw in current time
At, then the second duration evaluation of estimate of the recommendation article is 0.
S6212 is obtained every according to the second preset duration evaluation of estimate and each the sum of the second duration evaluation of estimate for recommending article
A the first duration evaluation of estimate for recommending article.
It in the present embodiment, is 0 to prevent the first duration evaluation of estimate, the first duration evaluation of estimate is the evaluation of the second preset duration
Value and each the sum of the second duration evaluation of estimate for recommending article, wherein the second preset duration evaluation of estimate is the numerical value greater than 0.
For example, the second preset duration evaluation of estimate can be with value for 1, when a certain the second duration evaluation of estimate for recommending article is 0
When, the first duration evaluation of estimate of the recommendation article is 1.
In the present embodiment, S600 obtains each generation time for recommending article, and according to current time and each pushes away
The time that generates for recommending article carries out data processing to each third recommendation for recommending article, obtains the 4th of each recommendation article
Recommendation, specifically:
According to the third recommendation progress timeliness tune for generating time and current time to recommendation article for recommending article
Whole to obtain the 4th recommendation for recommending article, formula is as follows:
Wherein, rec_score on the left of equal signiIt is the 4th recommendation of i-th of recommendation article, rec_score_ on the right side of equal sign
I is the third recommendation of i-th of recommendation article, and cur_time is current time, and chronomere is second, create_timeiIt is
The generation time of i recommendation article, chronomere is second, consttime1It is the first preset duration evaluation of estimate, the first preset duration
Value those skilled in the art of evaluation of estimate can be specifically arranged according to actual needs, for example, the first preset duration evaluation of estimate
Value can be 1,5,10,30,100,300,1000 etc.;In formula, consttime2It is the second preset duration evaluation of estimate, the
Value those skilled in the art of two preset duration evaluations of estimate can be specifically arranged according to actual needs, for example, second is default
Duration evaluation of estimate can be with value for 1 etc.;In formula, (cur_time-create_timei) when being the generation of i-th of recommendation article
It is long, (cur_time-create_timei)/86400 are the second duration evaluation of estimate of i-th of recommendation article, (consttime2+
(cur_time-create_timei)/86400) be i-th of recommendation article the first duration evaluation of estimate.
As shown in figure 9, optionally, S200 obtains multiple recommendation articles and each recommendation article according to N number of target item
Corresponding first recommendation, and each the first recommendation sequence for recommending article is obtained, include the following steps S210 to step
S270:
S210 obtains at least one according to each target item and recommends to collect, wherein each recommendation, which is concentrated, includes and the target
The relevant multiple recommendation articles of article, and each recommendation article corresponds to the 5th recommendation.
In the present embodiment, each target item carries out recommending article sieve by each existing item recommendation method
Choosing can obtain a recommendation collection.
S220 is normalized corresponding 5th recommendation of each recommendation article for recommending collection to include, obtains each
Article is recommended to correspond to the 6th recommendation.
In the present embodiment, corresponding 5th recommendation of each recommendation article for recommending collection to include is normalized,
It obtains each recommendation article and corresponds to the 6th recommendation, can be realized by following formula:
Wherein, rec_score on the left of equal signiIt is a certain the 6th recommendation for recommending to concentrate i-th of recommendation article, equal sign is left
Side rec_scoreiIt is a certain the 5th recommendation for recommending to concentrate i-th of recommendation article, rec_score1It is that a certain recommendation of row is concentrated
Maximum 5th recommendation of numerical value.After each recommendation collection is normalized in this way, each 6th recommendation is in [0,1]
Section in.
Whole recommendation collection is permeated and recommends a union, wherein there are identical when concentrating in different recommendations by S230
Recommendation article when, recommending and concentrating the 6th recommendation of the recommendation article to be that different recommendations concentrates the recommendation article corresponding
The sum of the 6th recommendation.
S240 is recommending and is concentrating to determine each the 6th recommendation descending sort for recommending article.
S250 obtains each affiliated classification for recommending article.
S260 obtains each recommendation according to the 6th recommendation descending sort and each affiliated classification for recommending article
The affiliated classification of product sorts, and determines serial number of each recommendation article in corresponding affiliated classification sequence.
S270, it is corresponding to each recommendation article according to serial number of each recommendation article in corresponding affiliated classification sequence
6th recommendation carries out data processing, corresponding first recommendation of each recommendation article is obtained, to obtain each recommendation article
The first recommendation sequence.
In the present embodiment, the serial number according to each recommendation article in corresponding affiliated classification sequence is to each recommendation
Corresponding 6th recommendation of product carries out data processing, obtains corresponding first recommendation of each recommendation article, can be according to following
Formula is realized:
Wherein, rec_score on the left of equal signiIt is first recommendation of i-th of recommendation article, rec_score on the right side of equal signi
It is the 6th recommendation of i-th of recommendation article,Be i-th of recommendation article it is corresponding belonging to classification sequence in serial number (or
PersonIt is frequency of occurrence of i-th of recommendation article in the 6th recommendation descending sort), constcateIt is classification constant, it can
Be set as 1 or other be greater than 0 numerical value.
The item recommendation method provided in the embodiment of the present application, by obtaining N number of mesh according to user's history behavioral data
Mark the recommendation factor of article and corresponding each target item, wherein N is the integer greater than 1;It is obtained according to N number of target item
Multiple recommendation articles and corresponding first recommendation of each recommendation article, and obtain each the first recommendation row for recommending article
Sequence, wherein each recommendation article is related at least one target item;According to the sequence of the first recommendation and each target item
Corresponding first recommendation of the corresponding each recommendation article of recommendation factor pair carries out data processing, obtains each recommendations article correspondence
The second recommendation, thus obtain it is each recommend article the second recommendation sequence.In this way, being pushed away according to the recommendation factor and first
Value sequence is recommended, recommendation article very high to correlation is handled, and has reached multiple recommendation dimensions to the recommendation for recommending article
Operation is carried out, the multifarious purpose of topic of recommendation results is improved, to realize the diversity for recommending article, improves use
The technical effect of the reading experience at family, and then solve the very high article of correlation and can recommend user with largely being flocked together, it makes
At the technical problem of the reading fatigue of user.
Based on technical concept identical with above-mentioned item recommendation method, the embodiment of the present application also provides a kind of recommendations of article
Device, as shown in Figure 10, comprising:
First obtains module 10, for obtaining N number of target item and corresponding each mesh according to user's history behavioral data
Mark the recommendation factor of article, wherein N is the integer greater than 1;
Second obtains module 20, for obtaining multiple recommendation articles and each recommendation article pair according to N number of target item
The first recommendation answered, and obtain it is each recommend article the first recommendation sequence, wherein each recommendation article and at least one
Target item is related;
Third obtains module 30, each according to the sequence of the first recommendation and the corresponding recommendation factor pair of each target item
Recommend corresponding first recommendation of article to carry out data processing, obtains corresponding second recommendation of each recommendation article, thus
To each the second recommendation sequence for recommending article.
Optionally, third obtains module 30, is used for:
The corresponding recommendation factor of each recommendation article is determined to relevant target item according to each recommendation article;
According to the first recommendation sequence and the corresponding recommendation factor of each recommendation article, obtain each recommendation relative to
Corresponding recommendation factor sequence determines serial number of each recommendation article in corresponding recommendation factor sequence;
According to serial number of each recommendation article in corresponding recommendation factor sequence to each recommendation article corresponding first
Recommendation carries out data processing, obtains corresponding second recommendation of each recommendation article.
Optionally, third obtains module 30, be also used to when recommend article there are when at least two corresponding second recommendations,
Maximum second recommendation at least two second recommendations for recommending article is chosen to recommend as recommendation article corresponding second
Value.
As shown in figure 11, optionally, this article recommendation apparatus further includes that the 4th acquisition module 40 and the 5th obtain module 50;
Each second recommendation for recommending article is ordered as each second recommendation descending sort for recommending article;
4th obtains module 40, for obtaining each recommendation article in each the second recommendation sequence for recommending article
Recommend the similarity value of article in relatively previous position, wherein the similarity value that the recommendation article to make number one includes is first pre-
If value;
5th obtains module 50, and the similarity value for including according to each recommendation article recommends the second of article to each
Recommendation carries out data processing, obtains each third recommendation for recommending article, pushes away to obtain each third for recommending article
Recommend value sequence.
Optionally, the 5th module 50 is obtained, is used for:
The recommendation article for obtaining each recommendation condition to previous position according to the similarity value that each recommendation article includes
Dissimilar angle value;
Corresponding second recommendation of each recommendation article is counted according to the dissimilar angle value that each recommendation article includes
According to processing, each third recommendation for recommending article is obtained, to obtain each third recommendation sequence for recommending article;
Optionally, recommend the dissimilar angle value of article less than the second preset value previous position when there are recommendation conditions
When, the second preset value is chosen as the dissimilar degree for recommending article.
As shown in figure 12, optionally, this article recommendation apparatus further includes the 6th acquisition module 60;
6th obtains module 60, for obtaining each generation time for recommending article, and according to current time and each
Recommend the time that generates of article to carry out data processing to each third recommendation for recommending article, obtains and each recommend the of article
Four recommendations, to obtain each the 4th recommendation sequence for recommending article.
Optionally, the 6th module 60 is obtained, is used for:
Each generation duration for recommending article is obtained according to current time and each generation time for recommending article;
According to the first preset duration evaluation of estimate and it is each recommend article generation duration, obtain it is each recommend article when
Imitate coefficient;
Operation is carried out to each third recommendation for recommending article according to each timeliness coefficient for recommending article, is obtained each
Recommend the 4th recommendation of article, to obtain each the 4th recommendation sequence for recommending article.
Optionally, the 6th module 60 is obtained, is used for:
Each the first duration evaluation of estimate for recommending article is obtained according to each generation duration for recommending article;
According to the first preset duration evaluation of estimate and each the ratio between the first duration evaluation of estimate for recommending article, each recommendation article
Timeliness coefficient.
Optionally, the 6th module 60 is obtained, is used for:
According to the ratio between each generation duration for recommending article and unit time, obtains each the second duration for recommending article and comment
Value;
According to the second preset duration evaluation of estimate and each the sum of the second duration evaluation of estimate for recommending article, each recommendation is obtained
First duration evaluation of estimate of article.
Optionally, second module 20 is obtained, is used for:
At least one is obtained according to each target item and recommends collection, wherein each recommendation, which is concentrated, includes and the target item
Relevant multiple recommendation articles, and each recommendation article corresponds to the 5th recommendation;
Corresponding 5th recommendation of each recommendation article for recommending collection to include is normalized, each recommendation is obtained
Product correspond to the 6th recommendation;
Whole recommendation collection is permeated and recommends a union, wherein is pushed away when being concentrated in different recommendations there are identical
When recommending article, recommending and concentrating the 6th recommendation of the recommendation article to be that the recommendation article corresponding the is concentrated in different recommendations
The sum of six recommendations;
Recommending and concentrating to determine each the 6th recommendation descending sort for recommending article;
Obtain each affiliated classification for recommending article;
According to the 6th recommendation descending sort and each affiliated classification for recommending article, each institute for recommending article is obtained
Belong to classification sequence, determines serial number of each recommendation article in corresponding affiliated classification sequence;
According to serial number of each recommendation article in corresponding affiliated classification sequence to each recommendation article the corresponding 6th
Recommendation carries out data processing, obtains corresponding first recommendation of each recommendations article, to obtain the of each recommendation article
The sequence of one recommendation.
The article recommendation apparatus provided in the embodiment of the present application obtains module 10 by first, for going through according to user
History behavioral data obtains the recommendation factor of N number of target item and corresponding each target item, wherein N is the integer greater than 1;
Second obtains module 20, for obtaining multiple recommendation articles and each recommendation article corresponding first according to N number of target item
Recommendation, and obtain each the first recommendation sequence for recommending article, wherein each recommendation article and at least one target item
It is related;Third obtains module 30, for every according to the sequence of the first recommendation and the corresponding recommendation factor pair of each target item
Corresponding first recommendation of a recommendation article carries out data processing, obtains corresponding second recommendation of each recommendation article, thus
Obtain each the second recommendation sequence for recommending article.In this way, according to recommending the factor and the first recommendation to sort, to correlation
Very high recommendation article is handled, and has been reached multiple recommendation dimensions and has been carried out operation to the recommendation for recommending article, has improved and push away
The multifarious purpose of topic of result is recommended, to realize the diversity for recommending article, improves the skill of the reading experience of user
Art effect, and then solve the very high article of correlation and can recommend user with largely being flocked together, cause the reading fatigue of user
The technical issues of.
Obviously, those skilled in the art should be understood that each module of above-mentioned the application or each step can be with general
Computing device realize that they can be concentrated on a single computing device, or be distributed in multiple computing devices and formed
Network on, optionally, they can be realized with the program code that computing device can perform, it is thus possible to which they are stored
Be performed by computing device in the storage device, perhaps they are fabricated to each integrated circuit modules or by they
In multiple modules or step be fabricated to single integrated circuit module to realize.In this way, the application be not limited to it is any specific
Hardware and software combines.
The foregoing is merely preferred embodiment of the present application, are not intended to limit this application, for the skill of this field
For art personnel, various changes and changes are possible in this application.Within the spirit and principles of this application, made any to repair
Change, equivalent replacement, improvement etc., should be included within the scope of protection of this application.
Claims (10)
1. a kind of item recommendation method, which is characterized in that the described method includes:
The recommendation factor of N number of target item and corresponding each target item is obtained according to user's history behavioral data, wherein N
For the integer greater than 1;
Multiple recommendation articles and corresponding first recommendation of each recommendation article are obtained according to N number of target item,
And obtain each the first recommendation sequence for recommending article, wherein each recommendation article and at least one described mesh
It is related to mark article;
According to first recommendation sequence and each recommendation of the corresponding recommendation factor pair of each target item
Corresponding first recommendation of product carries out data processing, corresponding second recommendation of each recommendation article is obtained, to obtain
Each the second recommendation sequence for recommending article.
2. item recommendation method according to claim 1, which is characterized in that it is described according to first recommendation sort with
And corresponding first recommendation of each recommendation article of the corresponding recommendation factor pair of each target item carries out at data
Reason obtains corresponding second recommendation of each recommendation article, comprising:
According to each recommendation article and the relevant target item determine the corresponding recommendation of each recommendation article because
Son;
According to first recommendation sequence and the corresponding recommendation factor of each recommendation article, each recommendation is obtained
Object sorts relative to the corresponding recommendation factor, determines each serial number for recommending article in corresponding recommendation factor sequence;
It is corresponding to each recommendation article according to each serial number for recommending article in corresponding recommendation factor sequence
First recommendation carries out data processing, obtains corresponding second recommendation of each recommendation article.
3. item recommendation method according to claim 2, which is characterized in that when there are at least two pairs for the recommendation article
When the second recommendation answered, maximum second recommendation is as institute in selection at least two second recommendations for recommending article
It states and recommends corresponding second recommendation of article.
4. item recommendation method according to claim 1, which is characterized in that each second recommendation for recommending article
It is ordered as each second recommendation descending sort for recommending article, the method also includes:
In each the second recommendation sequence for recommending article, obtains each recommendation condition and pushed away described in previous position
Recommend the similarity value of article, wherein the similarity value that the recommendation article to make number one includes is the first preset value;
Data are carried out to each second recommendation for recommending article according to each similarity value for recommending article to include
Processing obtains each third recommendation for recommending article, to obtain each third recommendation row for recommending article
Sequence.
5. item recommendation method according to claim 4, which is characterized in that the method also includes:
Obtain it is each it is described recommend article the generation time, and according to current time and it is each it is described recommend article generation when
Between to it is each it is described recommend article third recommendation carry out data processing, obtain it is each it is described recommend article the 4th recommend
Value, to obtain each the 4th recommendation sequence for recommending article.
6. a kind of article recommendation apparatus, which is characterized in that described device includes:
First obtains module, for obtaining N number of target item and corresponding each target item according to user's history behavioral data
The recommendation factor, wherein N is integer greater than 1;
Second obtains module, for obtaining multiple recommendation articles and each recommendation article according to N number of target item
Corresponding first recommendation, and obtain each the first recommendation sequence for recommending article, wherein each recommendation article
It is related to target item described at least one;
Third obtains module, every according to first recommendation sequence and the corresponding recommendation factor pair of each target item
Corresponding first recommendation of a recommendation article carries out data processing, obtains corresponding second recommendation of each recommendations article
Value, to obtain each the second recommendation sequence for recommending article.
7. article recommendation apparatus according to claim 6, which is characterized in that the third obtains module, is used for:
According to each recommendation article and the relevant target item determine the corresponding recommendation of each recommendation article because
Son;
According to first recommendation sequence and the corresponding recommendation factor of each recommendation article, each recommendation is obtained
Object sorts relative to the corresponding recommendation factor, determines each serial number for recommending article in corresponding recommendation factor sequence;
It is corresponding to each recommendation article according to each serial number for recommending article in corresponding recommendation factor sequence
First recommendation carries out data processing, obtains corresponding second recommendation of each recommendation article.
8. article recommendation apparatus according to claim 7, which is characterized in that the third obtains module, is also used to work as institute
Stating recommendation article, there are when at least two corresponding second recommendations, choose at least two second recommendations for recommending article
In maximum second recommendation as corresponding second recommendation of the recommendation article.
9. article recommendation apparatus according to claim 6, which is characterized in that described device further include the 4th acquisition module and
5th obtains module;Each second recommendation for recommending article is ordered as each the second recommendation drop for recommending article
Sequence sequence;
Described 4th obtains module, for obtaining each described push away in each the second recommendation sequence for recommending article
Recommend the relatively previous position of the article similarity value for recommending article, wherein the phase that the recommendation article to make number one includes
It is the first preset value like angle value;
Described 5th obtains module, and the similarity value for including according to each recommendation article is to each recommendation article
The second recommendation carry out data processing, each third recommendation for recommending article is obtained, to obtain each described push away
Recommend the third recommendation sequence of article.
10. article recommendation apparatus according to claim 9, which is characterized in that described device further includes the 6th acquisition module;
Described 6th obtains module, for obtaining each generation time for recommending article, and according to current time and often
A time that generates for recommending article carries out data processing to each third recommendation for recommending article, obtains each institute
The 4th recommendation for recommending article is stated, to obtain each the 4th recommendation sequence for recommending article.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810395304.0A CN108960958B (en) | 2018-04-27 | 2018-04-27 | Item recommendation method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810395304.0A CN108960958B (en) | 2018-04-27 | 2018-04-27 | Item recommendation method and device |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108960958A true CN108960958A (en) | 2018-12-07 |
CN108960958B CN108960958B (en) | 2022-03-22 |
Family
ID=64498965
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810395304.0A Active CN108960958B (en) | 2018-04-27 | 2018-04-27 | Item recommendation method and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108960958B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108763318A (en) * | 2018-04-27 | 2018-11-06 | 达而观信息科技(上海)有限公司 | item recommendation method and device |
Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2010134733A (en) * | 2008-12-05 | 2010-06-17 | Dainippon Printing Co Ltd | Information recommendation device, information recommendation method, and program |
US20100268661A1 (en) * | 2009-04-20 | 2010-10-21 | 4-Tell, Inc | Recommendation Systems |
CN104598643A (en) * | 2015-02-13 | 2015-05-06 | 成都品果科技有限公司 | Article similarity contribution factor, similarity acquiring method, as well as article recommendation method and system thereof |
CN104615631A (en) * | 2014-10-29 | 2015-05-13 | 中国建设银行股份有限公司 | Information recommendation method and device |
CN105183731A (en) * | 2014-06-04 | 2015-12-23 | 腾讯科技(深圳)有限公司 | Method, device, and system for generating recommended information |
CN105809479A (en) * | 2016-03-07 | 2016-07-27 | 海信集团有限公司 | Item recommending method and device |
CN106687952A (en) * | 2014-09-26 | 2017-05-17 | 甲骨文国际公司 | Techniques for similarity analysis and data enrichment using knowledge sources |
CN107092616A (en) * | 2016-11-02 | 2017-08-25 | 北京小度信息科技有限公司 | A kind of object order method and device |
CN107948748A (en) * | 2017-11-30 | 2018-04-20 | 奇酷互联网络科技(深圳)有限公司 | Recommend method, equipment, mobile terminal and the computer-readable storage medium of video |
-
2018
- 2018-04-27 CN CN201810395304.0A patent/CN108960958B/en active Active
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2010134733A (en) * | 2008-12-05 | 2010-06-17 | Dainippon Printing Co Ltd | Information recommendation device, information recommendation method, and program |
US20100268661A1 (en) * | 2009-04-20 | 2010-10-21 | 4-Tell, Inc | Recommendation Systems |
CN105183731A (en) * | 2014-06-04 | 2015-12-23 | 腾讯科技(深圳)有限公司 | Method, device, and system for generating recommended information |
CN106687952A (en) * | 2014-09-26 | 2017-05-17 | 甲骨文国际公司 | Techniques for similarity analysis and data enrichment using knowledge sources |
CN104615631A (en) * | 2014-10-29 | 2015-05-13 | 中国建设银行股份有限公司 | Information recommendation method and device |
CN104598643A (en) * | 2015-02-13 | 2015-05-06 | 成都品果科技有限公司 | Article similarity contribution factor, similarity acquiring method, as well as article recommendation method and system thereof |
CN105809479A (en) * | 2016-03-07 | 2016-07-27 | 海信集团有限公司 | Item recommending method and device |
CN107092616A (en) * | 2016-11-02 | 2017-08-25 | 北京小度信息科技有限公司 | A kind of object order method and device |
CN107948748A (en) * | 2017-11-30 | 2018-04-20 | 奇酷互联网络科技(深圳)有限公司 | Recommend method, equipment, mobile terminal and the computer-readable storage medium of video |
Non-Patent Citations (1)
Title |
---|
刘迎春: "基于信任和推荐关系的可信服务发现", 《系统工程理论与实践》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108763318A (en) * | 2018-04-27 | 2018-11-06 | 达而观信息科技(上海)有限公司 | item recommendation method and device |
CN108763318B (en) * | 2018-04-27 | 2022-04-19 | 达而观信息科技(上海)有限公司 | Item recommendation method and device |
Also Published As
Publication number | Publication date |
---|---|
CN108960958B (en) | 2022-03-22 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103823908B (en) | Content recommendation method and server based on user preference | |
US9535992B2 (en) | Recommendation engine | |
CN111209476B (en) | Recommendation method, model generation method, device, medium and equipment | |
US8725673B2 (en) | Evaluating an item based on user reputation information | |
US6489968B1 (en) | System and method for exposing popular categories of browse tree | |
US7783617B2 (en) | Personals advertisement affinities in a networked computer system | |
CN111461841B (en) | Article recommendation method, device, server and storage medium | |
Xiang et al. | Time-dependent models in collaborative filtering based recommender system | |
Selke et al. | Pushing the boundaries of crowd-enabled databases with query-driven schema expansion | |
CN101124575A (en) | Method and system for generating recommendations | |
Zhao et al. | Service quality evaluation by exploring social users’ contextual information | |
CN104199896A (en) | Video similarity determining method and video recommendation method based on feature classification | |
CN103678518A (en) | Method and device for adjusting recommendation lists | |
Coelho et al. | A personalized travel recommendation system using social media analysis | |
CN108765051B (en) | Item recommendation method and device | |
CN108595493A (en) | Method for pushing and device, storage medium, the electronic device of media content | |
CN105809275A (en) | Item scoring prediction method and apparatus | |
EP2701111A1 (en) | Group composition based recommender system and method | |
CN108960958A (en) | item recommendation method and device | |
CN108763318A (en) | item recommendation method and device | |
Alrawhani et al. | Real estate recommender system using case-based reasoning approach | |
CN108734548B (en) | Item recommendation method and device | |
CN108734549B (en) | Item recommendation method and device | |
CN110727863A (en) | Project recommendation method based on complex network collaborative filtering | |
Zhao et al. | An improved item-based movie recommendation algorithm |
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 | ||
CP03 | Change of name, title or address | ||
CP03 | Change of name, title or address |
Address after: Room 501, 502, 503, No. 66 Boxia Road, China (Shanghai) Pilot Free Trade Zone, Pudong New Area, Shanghai, March 2012 Patentee after: Daguan Data Co.,Ltd. Address before: Room 515, building Y1, No. 112, liangxiu Road, Pudong New Area, Shanghai 201203 Patentee before: DATAGRAND INFORMATION TECHNOLOGY (SHANGHAI) Co.,Ltd. |