CN107025277B - A kind of Quantitative marking method of user concealed feedback - Google Patents
A kind of Quantitative marking method of user concealed feedback Download PDFInfo
- Publication number
- CN107025277B CN107025277B CN201710186671.5A CN201710186671A CN107025277B CN 107025277 B CN107025277 B CN 107025277B CN 201710186671 A CN201710186671 A CN 201710186671A CN 107025277 B CN107025277 B CN 107025277B
- Authority
- CN
- China
- Prior art keywords
- user
- article
- feedback
- behavior
- scoring
- 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.)
- Active
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
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- 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
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
Abstract
The invention discloses a kind of Quantitative marking methods of user concealed feedback, comprising steps of collecting user to the operation behavior information of article, obtain user to the implicit feedback of article;Counting user calculates the significant property coefficient of every class implicit feedback behavior and the conversion ratio of the implicit feedback behavior to the implicit feedback information of article, obtains the primary scoring of all kinds of implicit feedback behaviors;The implicit feedback set that user is obtained as unit of user scores according to primary, carries out descending sort to the article in the implicit feedback set of user, obtains collating sequence;The score value that article corresponding to each sequence location in collating sequence is calculated using scoring learning model, obtains user-article implicit feedback Quantitative marking.Various types of user concealed feedback informations are converted to user to the score information of article by the present invention, propose a kind of user concealed feedback quantization methods of marking of high quality for the various personalized recommendation scenes with user concealed feedback information.
Description
Technical field
The present invention relates to computer recommending technical field, in particular to a kind of Quantitative marking method of user concealed feedback.
Background technique
The preference information of user is obtained often through the explicit scoring of user in personalized recommendation system, and then is recommended
Meet the article of user preference.For example, collaborative filtering is most popular, most widely used skill in current recommender system field
Art, and huge success is obtained in this field, a kind of user-article rating matrix is generally constructed in the algorithmic procedure, wherein
Each single item indicates explicit scoring of the user to respective articles, the scoring by prediction user to article, can recommend for user with
The interested article of the similar other users of the user interest.
However in many application scenarios, explicit scoring is difficult to obtain, even explicitly the commenting to article there is no user
Point, this brings very big challenge to the application of the recommended technologies such as collaborative filtering.
Summary of the invention
The purpose of the present invention is to overcome the shortcomings of the existing technology with deficiency, and the quantization for providing a kind of user concealed feedback is commented
Divide method, various types of user concealed feedback informations can be converted to user to the score information of article.
The purpose of the present invention is realized by the following technical solution: a kind of Quantitative marking method of user concealed feedback, packet
Include following steps:
S1, user is collected to the operation behavior information of article, obtain user to the implicit feedback of article;The user is to object
The operation behavior information of product includes the temporal information that the operation behavior classification of article and the behavior occur for user, the operation row
It is classification as specified by specific implementation scene and field experience;
The user refers to the implicit feedback of article, set of the user to the non-explicit scoring behavioural information of article.
S2, counting user calculate the significant property coefficient of every class implicit feedback behavior and are somebody's turn to do to the implicit feedback information of article
The conversion ratio of implicit feedback behavior obtains the corresponding primary scoring of all kinds of implicit feedback behaviors;
The significant property coefficient of the implicit feedback behavior refers to that the implicit feedback behavior with respective classes frequency occurs in just
The coefficient of correlativity, the coefficient are used to indicate the confidence level of the conversion ratio of current implicit feedback behavior;
The conversion ratio of the implicit feedback behavior refers to, under the premise of user has done corresponding implicit feedback behavior to article,
User finally the article occurs the probability of conversion behavior;The conversion behavior refers to that it is desired that user completes a trade company
Behavior.
S3, the implicit feedback set that user is obtained as unit of user, according to the first of the implicit feedback behavior in step S2
Grade scoring carries out descending sort to the article in the implicit feedback set of the user, obtains collating sequence;
The implicit feedback set of the user refers to, counts operated all items crossed of each user and its corresponding hidden
The set of formula feedback behavior.
S4, according to the collating sequence of the step S3, utilize scoring learning model to calculate each sequence location in collating sequence
The score value of corresponding article, sequence location refer to sequence permutation serial number, obtain the quantization of user-article implicit feedback and comment
Point.
Preferably, the calculating step of the primary scoring of the implicit feedback behavior includes:
User-article pair counting c under S2-1, all kinds of implicit feedback behaviors of statisticsk, wherein k is corresponding implicit feedback
The classification of behavior;
On the basis of S2-2, user-article pair obtained in S2-1, statistics obtains user-object there are conversion behavior
The counting t of product pairk;
S2-3, the conversion ratio for calculating implicit feedback k, calculation formula tk/ck;
S2-4, the conspicuousness factor alpha for calculating implicit feedback behaviork:
Wherein λ is level of signifiance index, for controlling feedback count ckWeight, to determine feedback count ckTo significant
Property coefficient αkThe speed of influence;
S2-5, the primary scoring for calculating implicit feedback k, calculation method are the conversion ratio and its conspicuousness of the implicit feedback
Factor alphakProduct.
Further, λ≤1 in step S2-4, value should be reasonably selected by specifically used scene.
Further, λ=0.5 in step S2-4.
Preferably, step S3 method particularly includes:
S3-1, each article in user concealed feedback set is successively retrieved and its primary of corresponding feedback behavior is commented
Point, if the corresponding feedback behavior of an article, the primary scoring of the feedback behavior is as the scoring of user-article;If one
A article corresponds to multiple feedback behaviors, then primary scoring descending sort is pressed in the corresponding multiple implicit feedback behaviors of same article,
The primary scoring of wherein primary highest behavior of scoring is chosen as the scoring of user-article;
S3-2, each user is subjected to descending sort to user-article scoring of different articles;Successively obtain multiple use
The user at family-article collating sequence.
If retaining institute specifically, there is multiple highest scorings arranged side by side in step S3-1 during primary scoring descending sort
The primary scoring for the behavior arranged side by side having, chooses all behaviors arranged side by side and user-article collating sequence is added.
If comparing its primary scoring specifically, the scoring of plurality of user-article is identical in step S3-2 and calculating
Conversion ratio in journey, if conversion ratio is also identical, the sequence having the same in collating sequence of multiple user-articles.
Preferably, scoring learning model described in step S4 refers to, is converted to use according to the article ranking results of user
Model of the family to the implicit feedback Quantitative marking of article;The model should defer to ranking results subscript to the implicit feedback scoring of article
It is bigger, the lower criterion of article scoring.
Specifically, user calculates the scoring of the article of serial number i in collating sequence in the following way:
Wherein yiIt is binary variable, is used to indicate whether corresponding implicit feedback operation has correlation, if having correlation,
Then yi=1, it is otherwise 0;
The correlation of the implicit feedback refers to that can the feedback reflect user to the preference of article, if can reflect use
Preference of the family to article, then it is assumed that be relevant;Otherwise it is assumed that uncorrelated;The correlation of different classes of implicit feedback behavior by
It is embodied specified by scene or field experience.
Preferably, in the implicit feedback Quantitative marking result, if user to same article there are multiple scorings, take it
In highest scoring as user to the scoring of the final quantization of the article.
Compared with the prior art, the invention has the following advantages and beneficial effects:
The present invention proposes a kind of general sequencing schemes based on the scoring of implicit feedback primary, and various types of users are hidden
Formula feedback information is converted to user to the score information of article, is the various personalized recommendation fields with user concealed feedback information
Scape proposes a kind of user concealed feedback quantization methods of marking of high quality.
Detailed description of the invention
Fig. 1 is the step flow chart of Quantitative marking method in embodiment.
Specific embodiment
Present invention will now be described in further detail with reference to the embodiments and the accompanying drawings, but embodiments of the present invention are unlimited
In this.
A kind of Quantitative marking method of user concealed feedback, such as Fig. 1, comprising the following steps:
S1, user is collected to the operation behavior information of article, obtain user to the implicit feedback of article;
The user includes that the operation behavior classification of article and the behavior occur for user to the operation behavior information of article
Temporal information, the operation behavior classification is as specified by specific implementation scene and field experience;
The user refers to the implicit feedback of article, set of the user to the non-explicit scoring behavioural information of article.
Scoring behavior is the subset of operation behavior, removes scoring behavior in operation behavior, can also have browsing behavior, click row
For etc. indirect scoring behavior, implicit feedback refer to the behavior of indirect scoring behavior part in operation behavior.
S2, counting user calculate the significant property coefficient of every class implicit feedback and this are implicit to the implicit feedback information of article
The conversion ratio of feedback obtains the corresponding primary scoring of all kinds of implicit feedback behaviors.
The significant property coefficient of the implicit feedback refers to frequency correlation occur with the implicit feedback of respective classes
Coefficient, the coefficient be used for indicate current implicit feedback conversion ratio confidence level;
The conversion ratio of the implicit feedback refers to, under the premise of user has done corresponding implicit feedback behavior to article, user
Finally the article occurs the probability of conversion behavior;The conversion behavior refers to that user completes a desired behavior of trade company,
In e-commerce platform, user is primary conversion to the single purchase behavior of article.
The calculation method of the primary scoring of the implicit feedback behavior specifically:
Under S2-1, all kinds of implicit feedback behaviors of statistics, user-article pair counting ck, wherein k is corresponding implicit feedback
The classification of behavior;
On the basis of S2-2, user-article pair obtained in S2-1, statistics obtains user-object there are conversion behavior
The counting t of product pairk;
S2-3, the conversion ratio for calculating implicit feedback k, calculation formula tk/ck;
S2-4, the conspicuousness factor alpha for calculating implicit feedbackk;It calculates in the following way:
Wherein λ is level of signifiance index, for controlling feedback count ckWeight, to determine feedback count ckTo significant
Property coefficient αkThe speed of influence;The value of λ≤1, λ should be reasonably selected by specifically used scene, such as λ=0.5.
S2-5, the primary scoring for calculating implicit feedback k, calculation method are the conversion ratio and its conspicuousness of the implicit feedback
Factor alphakProduct.
S3, the implicit feedback set that user is obtained as unit of user, it is primary according to the implicit feedback in the step S2
Scoring carries out descending sort to the article in the implicit feedback set of the user, obtains collating sequence.
The implicit feedback set of the user refers to, counts operated all items crossed of each user and its corresponding hidden
The set of formula feedback behavior.
Descending sort is carried out to the article in set to refer to, successively retrieve each article in user concealed feedback set and
The primary scoring of its corresponding feedback behavior, if the corresponding feedback behavior of an article, the primary scoring of the feedback behavior
As the scoring of user-article.If an article corresponds to multiple feedback behaviors, by the corresponding multiple implicit feedbacks of same article
By primary scoring descending sort, the primary scoring for choosing wherein primary highest behavior of scoring is commented as user-article for behavior
Point;Then each user is subjected to descending sort to user-article scoring of different articles, if plurality of user-article
Scoring it is identical, then compare its primary scoring calculating process in conversion ratio, if conversion ratio is also identical, multiple user-articles
Sequence having the same;Successively obtain user-article collating sequence of multiple users.
If there are multiple highest scorings arranged side by side during primary scoring descending sort, chooses all behaviors addition arranged side by side and use
Family-article collating sequence.
S4, according to the collating sequence of the step S3, utilize scoring learning model to calculate each sequence location in collating sequence
The score value of article corresponding to (position refers to sequence permutation number) obtains user-article implicit feedback Quantitative marking.
The scoring learning model refers to, is converted to user to the implicit anti-of article according to the article ranking results of user
Present the model of Quantitative marking;The model should defer to the scoring of the implicit feedback of article that ranking results subscript is bigger, and article scoring is got over
Low criterion.User U calculates the scoring for the article for being designated as i under in sequence in the following way:
Wherein yiIt is binary variable, is used to indicate whether corresponding implicit feedback operation has correlation, if having correlation,
Then yi=1, it is otherwise 0.
The correlation of the implicit feedback refers to that can the feedback reflect user to the preference of article, if can reflect use
Preference of the family to article, then it is assumed that be relevant;Otherwise it is assumed that uncorrelated.Particularly, different classes of implicit feedback behavior
Correlation is as specified by specific implementation scene or field experience.
In the Quantitative marking result, if user to same article there are multiple scorings, take wherein highest scoring to make
It scores for final quantization of the user to the article.
Specifically, there have certain platform to be collected into user concealed feedback to be as shown in the table:
Table 1: implicit feedback use-case
User | Article | Implicit feedback |
U1 | I1 | AE |
U1 | I1 | AC |
U1 | I1 | AZ |
U1 | I2 | AE |
U2 | I1 | AC |
U2 | I2 | AME |
U2 | I2 | AZ |
AE, AC, AZ, AME are a certain implicit feedback behavior respectively in table 1, and wherein AZ behavior is conversion behavior.Time letter
Breath has incorporated in operation behavior, for example " AE " can be " user repeatedly clicks in 3 seconds ".To the amount of above-mentioned implicit feedback
Changing scoring, steps are as follows:
1, the primary scoring of each implicit feedback type is calculated:
Under 1-1, all kinds of implicit feedback behaviors of statistics, user-article pair counting ck, wherein k is corresponding implicit feedback
The classification of behavior;
Calculate counting c [AE, AC, AME, AZ]=[2,2,1,2] of implicit feedback;
On the basis of 1-2, user-article pair obtained in 1-1, statistics obtains user-article there are conversion behavior
Pair counting tk;
The corresponding implicit feedback vector of all user-articles is obtained, if it exists conversion behavior, then all behaviors in the vector
Conversion count+1.For example, (U1, I1)=[AE, AC, AZ], because there is conversion behavior AZ, so the conversion of AE, AC, AZ count+
1.All user-articles pair are similarly traversed, the corresponding conversion of all feedback behaviors can be acquired and counted.
Calculate counting t [AE, AC, AME]=[1,1,1] of implicit feedback conversion
1-3, the conversion ratio for calculating implicit feedback k, calculation formula tk/ck;
Calculate conversion ratio [AE, AC, AME]=[0.5,0.5,1] of implicit feedback
AZ is conversion behavior, conversion ratio necessarily 1, therefore does not list calculating.
1-4, the conspicuousness factor alpha for calculating implicit feedbackk;It calculates in the following way:
Assuming that λ=0.5, then significant property coefficient a [AE, AC, AME, AZ]=[0.73,0.73,0.62,0.73]
1-5, the primary scoring for calculating implicit feedback k, calculation method are the conversion ratio and its conspicuousness of the implicit feedback
Factor alphakProduct;
Calculate primary scoring: [AE, AC, AME, AZ]=[0.365,0.365,0.62,0.73]
2, the implicit feedback set that user is calculated as unit of user, is commented according to the implicit feedback primary in the step 1
Point, descending sort is carried out to the article in set.
Obtain user-article implicit feedback set:
(U1, I1)=[AE, AC, AZ]
(U1, I2)=[AE]
(U2, I1)=[AC]
(U2, I2)=[AME, AZ]
To user-article (U1, I1), the primary scoring highest of feedback behavior AZ in set [AE, AC, AZ], therefore AZ
Primary scoring is used as the scoring of user-article (U1, I1);Successively obtain the scoring of different user-article:
(U1, I1)-AZ=0.73
(U1, I2)-AE=0.365
(U2, I1)-AC=0.365
(U2, I2)-AZ=0.73
Then descending sort is carried out to user-article scoring of each user, obtains collating sequence:
U1=[I1-AZ, I2-AE]
U2=[I2-AZ, I1-AC]
For user U2, since the corresponding behavior of I1 is AC, user-article scoring is that the corresponding behavior of 0.365, I2 is
AZ, user-article scoring is 0.73, therefore is [I2, I1], the i.e. corresponding row of I2 ranking ratio I1 high, I2 to result after I1, I2 sequence
Sequence serial number is that the corresponding sequence serial number of 1, I1 is 2.
If occurring U1=[(I1-AC, I2-AE)] at this time, the scoring of I1, I2 are identical, then further compare its primary at this time
Conversion ratio in the calculating process that scores, the big sequence of conversion ratio are forward;If conversion ratio is also identical, then I1, I2 are having the same
Sequence, then for user U1, I1, I2 collating sequence position is all 1.
3, calculating user-article Quantitative marking is, user U is to the scoring for the article that collating sequence position is i using as follows
Mode calculates:
Wherein yiIt is binary variable, is used to indicate whether corresponding implicit feedback operation has correlation, if having correlation,
Then yi=1, it is otherwise 0.Further, it is assumed that (refer to AC, AE, AZ, AME) in aforesaid operations, other rows in addition to AE behavior
To be corelation behaviour (corelation behaviour is AC, AZ, AME), then:
The Quantitative marking that the Quantitative marking of U1-I1 is 1, U1-I2 is that the Quantitative marking of 0, U2-I1 is the amount of 0.63, U2-I2
Changing scoring is 1.
(there is highest primary arranged side by side if there is the highest implicit feedback primary identical situation that scores in step 2
Scoring), it is assumed for example that (U1, I1)=[AE, AC], then obtaining collating sequence is U1 for the scoring of AE with AC primary, conversion ratio
=[I1- (AE, AC), I2-AE], (U1, I1) has more than one behavior (AE, AC are arranged side by side) to enter the Quantitative marking stage herein.
In the Quantitative marking result, if user to same article there are multiple scorings, take wherein highest scoring to make
It scores for final quantization of the user to the article.
The above embodiment is a preferred embodiment of the present invention, but embodiments of the present invention are not by above-described embodiment
Limitation, other any changes, modifications, substitutions, combinations, simplifications made without departing from the spirit and principles of the present invention,
It should be equivalent substitute mode, be included within the scope of the present invention.
Claims (10)
1. a kind of Quantitative marking method of user concealed feedback, which comprises the following steps:
S1, user is collected to the operation behavior information of article, obtain user to the implicit feedback of article;The user is to article
Operation behavior information includes the temporal information that the operation behavior classification of article and the behavior occur for user, the operation behavior class
Not as specified by specific implementation scene and field experience;
The user refers to the implicit feedback of article, set of the user to the non-explicit scoring behavioural information of article;
S2, counting user calculate the significant property coefficient of every class implicit feedback behavior and this are implicit to the implicit feedback information of article
The conversion ratio of feedback behavior obtains the corresponding primary scoring of all kinds of implicit feedback behaviors;
The significant property coefficient of the implicit feedback behavior refers to that the implicit feedback behavior with respective classes frequency occurs and is positively correlated
The coefficient of relationship, the coefficient are used to indicate the confidence level of the conversion ratio of current implicit feedback behavior;
The conversion ratio of the implicit feedback behavior refers to, under the premise of user has done corresponding implicit feedback behavior to article, user
Finally the article occurs the probability of conversion behavior;The conversion behavior refers to that user completes a desired behavior of trade company;
S3, the implicit feedback set that user is obtained as unit of user, are commented according to the primary of the implicit feedback behavior in step S2
Point, descending sort is carried out to the article in the implicit feedback set of the user, obtains collating sequence;
The implicit feedback set of the user refers to, counts operated all items crossed of each user and its corresponding implicit anti-
The set of feedback behavior;
S4, according to the collating sequence of the step S3, it is right to calculate each sequence location institute in collating sequence using scoring learning model
The score value for the article answered obtains user-article implicit feedback Quantitative marking;Sequence location refers to sequence permutation serial number.
2. the Quantitative marking method of user concealed feedback according to claim 1, which is characterized in that the implicit feedback row
For the calculating step of primary scoring include:
User-article pair counting c under S2-1, all kinds of implicit feedback behaviors of statisticsk, wherein k is corresponding implicit feedback behavior
Classification;
On the basis of S2-2, user-article pair obtained in S2-1, statistics obtains user-article pair there are conversion behavior
Counting tk;
S2-3, the conversion ratio for calculating implicit feedback k, calculation formula tk/ck;
S2-4, the conspicuousness factor alpha for calculating implicit feedback behaviork:
Wherein λ is level of signifiance index, for controlling feedback count ckWeight, to determine feedback count ckTo conspicuousness system
Number αkThe speed of influence;
S2-5, the primary scoring for calculating implicit feedback k, calculation method are the conversion ratio and its significant property coefficient of the implicit feedback
αkProduct.
3. the Quantitative marking method of user concealed feedback according to claim 2, which is characterized in that λ in step S2-4≤
1, value is reasonably selected according to specifically used scene.
4. the Quantitative marking method of user concealed feedback according to claim 2, which is characterized in that λ in step S2-4=
0.5。
5. the Quantitative marking method of user concealed feedback according to claim 1, which is characterized in that the specific side of step S3
Method are as follows:
S3-1, each article in user concealed feedback set and its primary scoring of corresponding feedback behavior are successively retrieved, if
The corresponding feedback behavior of one article, then the primary scoring of the feedback behavior is as the scoring of user-article;If an article
The corresponding multiple implicit feedback behaviors of same article are then chosen it by primary scoring descending sort by corresponding multiple feedback behaviors
The primary scoring of middle primary highest behavior of scoring is as the scoring of user-article;
S3-2, each user is subjected to descending sort to user-article scoring of different articles;Successively obtain multiple users'
User-article collating sequence.
6. the Quantitative marking method of user concealed feedback according to claim 5, which is characterized in that if in step S3-1 just
There are multiple highest scorings arranged side by side during grade scoring descending sort, then retains the primary scoring of all behaviors arranged side by side, choose
User-article collating sequence is added in all behaviors arranged side by side.
7. the Quantitative marking method of user concealed feedback according to claim 5, which is characterized in that if its in step S3-2
In multiple user-articles scoring it is identical, then compare its primary scoring calculating process in conversion ratio, if conversion ratio is also identical,
Then multiple user-articles sequence having the same in collating sequence, sequence serial number are identical.
8. the Quantitative marking method of user concealed feedback according to claim 1, which is characterized in that commented described in step S4
Divide learning model to refer to, user is converted to the model of the implicit feedback Quantitative marking of article according to collating sequence;The model
It should defer to the scoring of the implicit feedback of article that collating sequence serial number is bigger, article scores lower criterion.
9. the Quantitative marking method of user concealed feedback according to claim 8, which is characterized in that user is to collating sequence
The scoring of the article of middle serial number i calculates in the following way:
Wherein yiIt is binary variable, is used to indicate whether corresponding implicit feedback operation has correlation, if has correlation, yi
=1, it is otherwise 0;The correlation of the implicit feedback refers to that can the feedback reflect user to the preference of article, if can be anti-
User is reflected to the preference of article, then it is assumed that is relevant;Otherwise it is assumed that uncorrelated;The correlation of different classes of implicit feedback behavior
Property is as specified by specific implementation scene or field experience.
10. the Quantitative marking method of user concealed feedback according to claim 1, which is characterized in that the implicit feedback
In Quantitative marking result, if user to same article there are multiple scorings, take wherein it is highest scoring as user to the object
The final quantization of product scores.
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710186671.5A CN107025277B (en) | 2017-03-27 | 2017-03-27 | A kind of Quantitative marking method of user concealed feedback |
PCT/CN2017/118294 WO2018176937A1 (en) | 2017-03-27 | 2017-12-25 | Quantitative scoring method for implicit feedback of user |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710186671.5A CN107025277B (en) | 2017-03-27 | 2017-03-27 | A kind of Quantitative marking method of user concealed feedback |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107025277A CN107025277A (en) | 2017-08-08 |
CN107025277B true CN107025277B (en) | 2019-08-20 |
Family
ID=59526009
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710186671.5A Active CN107025277B (en) | 2017-03-27 | 2017-03-27 | A kind of Quantitative marking method of user concealed feedback |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN107025277B (en) |
WO (1) | WO2018176937A1 (en) |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107025277B (en) * | 2017-03-27 | 2019-08-20 | 华南理工大学 | A kind of Quantitative marking method of user concealed feedback |
CN109885644B (en) * | 2019-04-08 | 2021-04-06 | 浙江大学城市学院 | Importance evaluation method for searching and sorting of Internet of things item information |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1799052A (en) * | 2003-06-04 | 2006-07-05 | 索尼网上服务有限责任公司 | Content recommendation device with an arrangement engine |
CN1875639A (en) * | 2003-11-06 | 2006-12-06 | 诺基亚公司 | Automatic personal playlist generation with implicit user feedback |
CN1972299A (en) * | 2005-11-08 | 2007-05-30 | 索尼网上服务有限责任公司 | Content item provision method |
CN101320375A (en) * | 2008-07-04 | 2008-12-10 | 浙江大学 | Digital book search method based on user click action |
Family Cites Families (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102789499B (en) * | 2012-07-16 | 2015-08-12 | 浙江大学 | Based on the collaborative filtering method of implicit relationship situated between article |
CN103996143A (en) * | 2014-05-12 | 2014-08-20 | 华东师范大学 | Movie marking prediction method based on implicit bias and interest of friends |
US20150332169A1 (en) * | 2014-05-15 | 2015-11-19 | International Business Machines Corporation | Introducing user trustworthiness in implicit feedback based search result ranking |
CN106296305A (en) * | 2016-08-23 | 2017-01-04 | 上海海事大学 | Electric business website real-time recommendation System and method under big data environment |
CN107025277B (en) * | 2017-03-27 | 2019-08-20 | 华南理工大学 | A kind of Quantitative marking method of user concealed feedback |
-
2017
- 2017-03-27 CN CN201710186671.5A patent/CN107025277B/en active Active
- 2017-12-25 WO PCT/CN2017/118294 patent/WO2018176937A1/en active Application Filing
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1799052A (en) * | 2003-06-04 | 2006-07-05 | 索尼网上服务有限责任公司 | Content recommendation device with an arrangement engine |
CN1875639A (en) * | 2003-11-06 | 2006-12-06 | 诺基亚公司 | Automatic personal playlist generation with implicit user feedback |
CN1972299A (en) * | 2005-11-08 | 2007-05-30 | 索尼网上服务有限责任公司 | Content item provision method |
CN101320375A (en) * | 2008-07-04 | 2008-12-10 | 浙江大学 | Digital book search method based on user click action |
Also Published As
Publication number | Publication date |
---|---|
CN107025277A (en) | 2017-08-08 |
WO2018176937A1 (en) | 2018-10-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107491813B (en) | Long-tail group recommendation method based on multi-objective optimization | |
CN105678607B (en) | A kind of Order Batch method based on improved K-Means algorithm | |
CN108763314A (en) | A kind of interest recommends method, apparatus, server and storage medium | |
CN104462383B (en) | A kind of film based on a variety of behavior feedbacks of user recommends method | |
CN103886047B (en) | Towards the online recommendation method of distribution of stream data | |
CN111428147A (en) | Social recommendation method of heterogeneous graph volume network combining social and interest information | |
CN105095219B (en) | Micro-blog recommendation method and terminal | |
CN102609523A (en) | Collaborative filtering recommendation algorithm based on article sorting and user sorting | |
CN107526850A (en) | Social networks friend recommendation method based on multiple personality feature mixed architecture | |
CN107368540A (en) | The film that multi-model based on user's self-similarity is combined recommends method | |
CN104077412B (en) | A kind of microblog users interest Forecasting Methodology based on more Markov chains | |
CN106776701B (en) | Problem determination method and device for item recommendation | |
CN106897914A (en) | A kind of Method of Commodity Recommendation and system based on topic model | |
CN105069129B (en) | Adaptive multi-tag Forecasting Methodology | |
CN105183909B (en) | social network user interest predicting method based on Gaussian mixture model | |
Li et al. | Content-based filtering recommendation algorithm using HMM | |
CN109902235A (en) | User preference based on bat optimization clusters Collaborative Filtering Recommendation Algorithm | |
CN106202151A (en) | One is used for improving the multifarious method of personalized recommendation system | |
CN107025277B (en) | A kind of Quantitative marking method of user concealed feedback | |
Vall et al. | The Importance of Song Context in Music Playlists. | |
CN103336771A (en) | Data similarity detection method based on sliding window | |
CN105809275A (en) | Item scoring prediction method and apparatus | |
CN107247753A (en) | A kind of similar users choosing method and device | |
Piliponyte et al. | Sequential music recommendations for groups by balancing user satisfaction | |
CN109508407A (en) | The tv product recommended method of time of fusion and Interest Similarity |
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 |