CN103500228A - Similarity measuring method improved through collaborative filtering recommendation algorithm - Google Patents

Similarity measuring method improved through collaborative filtering recommendation algorithm Download PDF

Info

Publication number
CN103500228A
CN103500228A CN201310505323.1A CN201310505323A CN103500228A CN 103500228 A CN103500228 A CN 103500228A CN 201310505323 A CN201310505323 A CN 201310505323A CN 103500228 A CN103500228 A CN 103500228A
Authority
CN
China
Prior art keywords
mrow
msub
mover
sim
munder
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN201310505323.1A
Other languages
Chinese (zh)
Inventor
赵朋朋
吴健
冒九妹
鲜学丰
崔志明
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Suzhou University
Original Assignee
Suzhou University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Suzhou University filed Critical Suzhou University
Priority to CN201310505323.1A priority Critical patent/CN103500228A/en
Publication of CN103500228A publication Critical patent/CN103500228A/en
Pending legal-status Critical Current

Links

Images

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

Landscapes

  • Engineering & Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

A similarity measuring method improved through a collaborative filtering recommendation algorithm includes the following steps of (S1) building a rating matrix R(n*m) of n users in a user set U={U1, U2,..., Un} to m items in an item set I={I1, I2,..., Im}, taking Ra,i as representation of rating of an item Ii, wherein Ua belongs to U and Ii belongs to I, (S2) calculating the similarity sim(Ua, Ub) between a user Ua and a user Ub and the similarity sim(Ii, Ij) between an item Ii and an item Ij, defining a similarity influence divisor epsilon, so that sim'(Ua, Ub) equals to epsilon* sim(Ua, Ub) and sim'(Ii, Ij) equals to epsilon* sim'(Ii, Ij), (S3) taking a parameter lambada in an interval between 0 and 1, and predicting rating of the users to the items according to lambada, epsilon, an average rating value of the users to the items, similarity between the users and similarity between the items.

Description

Improved similarity measurement method in collaborative filtering recommendation algorithm
Technical Field
The invention relates to a Collaborative filtering (Collaborative filtering) recommendation technology in recommendation system research, in particular to an improved similarity measurement method in a Collaborative filtering recommendation algorithm.
Background
With the rapid spread of the internet and the rapid development of electronic commerce, the information data on the internet is growing sharply, and how to make users quickly and efficiently obtain required information from vast data oceans becomes increasingly urgent. Providing active recommendation services for users is also increasingly being applied to various web portals and e-commerce systems. These systems provide recommendation services to users by collecting their historical information, learning their interests and behavior patterns, and analyzing their behavior characteristics.
Collaborative filtering recommendation technologies are widely applied in the field of recommendation systems, and are mainly classified into two types: based on User-based Collaborative Filtering and Item-based Collaborative Filtering, the basic idea is to generate recommendations to target users based on nearest neighbors, and the final recommendation form is score prediction and Top-N recommendation. Tapestry was the earliest proposed collaborative filtering recommendation system, which records the viewpoint of reading articles by each user, and the target user needs to explicitly point out other users whose behaviors are similar to the behavior of the target user. GroupLens, Ringo, and Video recommenders are also earlier collaborative filtering recommendation systems that provide users with recommendation services such as movies, news, and music, respectively, through opinions of other users.
With the continuous expansion of the scale of the electronic commerce system, the number of users and project data are increased sharply, so that the scoring data of user projects are extremely sparse. Under the condition that user scoring data is extremely sparse, the traditional similarity measurement method depends on the number of commonly scored items, so that certain contingency exists in the traditional similarity measurement, and the nearest neighbor of a target user and the item obtained through calculation is inaccurate, so that the recommendation quality of a recommendation system is reduced.
The collaborative filtering recommendation algorithm mainly predicts the scores of the items by the users through the similarity, the similarity can be respectively measured according to the relationship between the users or the items, and the accuracy of the similarity measurement is directly related to the recommendation quality of the whole recommendation system.
The similarity calculation may be based on similarity calculation between users or on similarity calculation between items. With sim (U)a,Ub) Representing a user UaAnd user UbThe similarity between the users is obtained firstlyaAnd user UbAll the scored projects are then used for calculating the user U through different similarity measurement methodsaAnd user UbSimilarity between sim (U)a,Ub). In the same way, item IiAnd item IjThe similarity between them is denoted sim (I)i,Ij) Then, the item I is acquirediAnd item IjAll existing users are graded, and a project I is obtained according to the existing grading valuesiAnd item IjSimilarity between them sim (I)i,Ij)。
Common similarity metrics include: cosine similarity, correlation similarity, and modified cosine similarity. In the cosine similarity measure method, a user item score matrix R (n × m) is constructed. If the user does not score an item, then the user is assumed to have a score of 0 for the item. The performance of similarity calculation can be effectively improved by setting the unknown score of the user to 0, but when the number of users and items is very large and the item evaluation data of the user is extremely sparse, the reliability of setting the unknown score to 0 is not high.
In practice, the user's preference for the unscored items may not be the same or different. When the user UaAnd user UbWhen no item is scored, the scoring of the item by the user is set to be 0, and the calculation of the U of the user is undoubtedly carried outaAnd user UbThe similarity between them is improved because they match the termsThe goal scores will not necessarily be exactly the same as 0. Therefore, when the user score data is extremely sparse, setting the unknown score to 0 has a high influence on calculating the similarity value. When the user UaAnd user UbWhen one user gives a score to an item and the other user gives no score, setting the unknown score to 0 will make the calculated value of similarity smaller than its actual value, but when the user score data is extremely sparse, the effect will be small.
Therefore, under the condition that the user scoring data is extremely sparse, the cosine similarity cannot effectively measure the similarity between users, the calculated value of the cosine similarity actually improves the similarity between users, and the modified cosine similarity measurement method has the same problem.
In the correlation similarity measurement method, let
Figure BDA0000400762250000031
Representing a user UaScored item set, in computing user UaAnd user UbThe similarity between the users is calculated firstlyaAnd user UbCommon scored item intersection
Figure BDA0000400762250000033
Then in the item collection
Figure BDA0000400762250000034
Calculating the user U by the measurement method of the correlation similarityaAnd user UbThe similarity between them. However, the confidence in similarity measured by the relative similarity depends on the intersection of the scoring items
Figure BDA0000400762250000035
The greater the number of commonly scored items, the greater the confidence in the similarity of their measures. In the case of extremely sparse user scoring dataIn case of a collection of items scored jointly by two usersEven smaller, even if the scores are very similar across such a small set of items, it cannot be determined that the similarity between users is relatively high. When the existing scoring items of the users are the same, namely
Figure BDA0000400762250000037
The inter-user similarity is measured by their intersection, and the confidence of the similarity measurement result is higher. When in useWhen the similarity between users is measured through the intersection part of the user rating items, the similarity between the users is undoubtedly improved, because the scoring deviations of the users in the non-intersection part of the user rating items are not necessarily completely the same, but the similarity of the users is calculated only through the intersection part, and the method is similar to the method that the scoring deviations of the users in the non-intersection part are set to be the same and are 0, and the calculated similarity is higher than an actual value. Therefore, under the condition that the user scoring data is extremely sparse, the measurement method of the correlation similarity has certain disadvantages.
In summary, in order to make the similarity value affected by the sparsity as little as possible, the present invention provides a method for improving the similarity metric by using the similarity impact factor.
Disclosure of Invention
The invention provides an improved similarity measurement method in a collaborative filtering recommendation algorithm, which comprises the following steps:
s1, creating a user set U = { U = { (U)1,U2,…,UnN user pairs in the set of items I = { I = } {1,I2,…,ImScoring matrix R (n × m) of m items in (n) }, with Ra,iRepresenting a user UaFor item IiScore of (1), whichMiddle Ua∈U,Ii∈I；
S2, calculating the user U respectivelyaAnd UbSimilarity sim (U) betweena,Ub) Item IiAnd IjSimilarity between them sim (I)i,Ij) Defining a similarity influencing factor epsilon, let sim' (U)a,Ub)=ε×sim(Ua,Ub)，sim'(Ii,Ij)=ε×sim(Ii,Ij)；
And S3, taking a parameter lambda in the [0,1] interval, and predicting the grade of the user to the project according to the lambda, the epsilon, the average value of the grade of the user to the project, the similarity between the users and the similarity between the projects.
Preferably, in step S2, the method further comprises <math> <mrow> <mi>sim</mi> <mrow> <mo>(</mo> <msub> <mi>U</mi> <mi>a</mi> </msub> <mo>,</mo> <msub> <mi>U</mi> <mi>b</mi> </msub> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mrow> <munder> <mi>&Sigma;</mi> <mrow> <mi>k</mi> <mo>&Element;</mo> <msub> <mi>I</mi> <mrow> <msub> <mi>U</mi> <mi>a</mi> </msub> <msub> <mi>U</mi> <mi>b</mi> </msub> </mrow> </msub> </mrow> </munder> <mrow> <mo>(</mo> <msub> <mi>R</mi> <mrow> <mi>a</mi> <mo>,</mo> <mi>k</mi> </mrow> </msub> <mo>-</mo> <mover> <msub> <mi>R</mi> <mi>a</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mrow> <mo>(</mo> <msub> <mi>R</mi> <mrow> <mi>b</mi> <mo>,</mo> <mi>k</mi> </mrow> </msub> <mo>-</mo> <mover> <msub> <mi>R</mi> <mi>b</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> </mrow> <mrow> <msqrt> <munder> <mi>&Sigma;</mi> <mrow> <mi>k</mi> <mo>&Element;</mo> <msub> <mi>I</mi> <msub> <mi>U</mi> <mi>a</mi> </msub> </msub> </mrow> </munder> <msup> <mrow> <mo>(</mo> <msub> <mi>R</mi> <mrow> <mi>a</mi> <mo>,</mo> <mi>k</mi> </mrow> </msub> <mo>-</mo> <mover> <msub> <mi>R</mi> <mi>a</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mn>2</mn> </msup> </msqrt> <msqrt> <munder> <mi>&Sigma;</mi> <mrow> <mi>k</mi> <mo>&Element;</mo> <msub> <mi>I</mi> <msub> <mi>U</mi> <mi>b</mi> </msub> </msub> </mrow> </munder> <msup> <mrow> <mo>(</mo> <msub> <mi>R</mi> <mrow> <mi>b</mi> <mo>,</mo> <mi>k</mi> </mrow> </msub> <mo>-</mo> <mover> <msub> <mi>R</mi> <mi>b</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mn>2</mn> </msup> </msqrt> </mrow> </mfrac> <mo>,</mo> </mrow> </math> The above-mentioned <math> <mrow> <mi>sim</mi> <mrow> <mo>(</mo> <msub> <mi>I</mi> <mi>i</mi> </msub> <mo>,</mo> <msub> <mi>I</mi> <mi>j</mi> </msub> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mrow> <munder> <mi>&Sigma;</mi> <mrow> <mi>k</mi> <mo>&Element;</mo> <msub> <mi>U</mi> <mrow> <msub> <mi>I</mi> <mi>i</mi> </msub> <msub> <mi>I</mi> <mi>j</mi> </msub> </mrow> </msub> </mrow> </munder> <mrow> <mo>(</mo> <msub> <mi>R</mi> <mrow> <mi>k</mi> <mo>,</mo> <mi>i</mi> </mrow> </msub> <mo>-</mo> <mover> <msub> <mi>R</mi> <mi>i</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mrow> <mo>(</mo> <msub> <mi>R</mi> <mrow> <mi>k</mi> <mo>,</mo> <mi>j</mi> </mrow> </msub> <mo>-</mo> <mover> <msub> <mi>R</mi> <mi>j</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> </mrow> <mrow> <msqrt> <munder> <mi>&Sigma;</mi> <mrow> <mi>k</mi> <mo>&Element;</mo> <msub> <mi>U</mi> <msub> <mi>I</mi> <mi>i</mi> </msub> </msub> </mrow> </munder> <msup> <mrow> <mo>(</mo> <msub> <mi>R</mi> <mrow> <mi>k</mi> <mo>,</mo> <mi>i</mi> </mrow> </msub> <mo>-</mo> <mover> <msub> <mi>R</mi> <mi>i</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mn>2</mn> </msup> </msqrt> <msqrt> <munder> <mi>&Sigma;</mi> <mrow> <mi>k</mi> <mo>&Element;</mo> <msub> <mi>U</mi> <msub> <mi>I</mi> <mi>j</mi> </msub> </msub> </mrow> </munder> <msup> <mrow> <mo>(</mo> <msub> <mi>R</mi> <mrow> <mi>k</mi> <mo>,</mo> <mi>j</mi> </mrow> </msub> <mo>-</mo> <mover> <msub> <mi>R</mi> <mi>j</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mn>2</mn> </msup> </msqrt> </mrow> </mfrac> <mo>,</mo> </mrow> </math> WhereinAnd
Figure BDA0000400762250000044
respectively represent users UaAnd user UbAverage score value for the scored item.
Preferably, in step S2, when sim' (U)a,Ub)=ε×sim(Ua,Ub) When is in use, the
Figure BDA0000400762250000045
When sim' (I)i,Ij)=ε×sim(Ii,Ij) When is in use, the
Figure BDA0000400762250000046
Wherein
Figure BDA00004007622500000412
For user UaAnd UbA set of commonly scored items,
Figure BDA00004007622500000413
And
Figure BDA00004007622500000414
are respectively a user UaAnd UbA scored set of items.
Preferably, in step S2, 0 ≦ ε ≦ 1.
Preferably, in step S3, the user U is predictedaFor the unviewed item IiIs scored as <math> <mrow> <msub> <mi>R</mi> <mrow> <mi>a</mi> <mo>,</mo> <mi>i</mi> </mrow> </msub> <mo>=</mo> <mi>&lambda;</mi> <mo>&times;</mo> <mrow> <mo>(</mo> <mover> <msub> <mi>R</mi> <mi>a</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>+</mo> <mfrac> <mrow> <munder> <mi>&Sigma;</mi> <mrow> <msub> <mi>U</mi> <mi>x</mi> </msub> <mo>&Element;</mo> <mi>U</mi> </mrow> </munder> <mrow> <mo>(</mo> <msup> <mi>sim</mi> <mo>&prime;</mo> </msup> <mrow> <mo>(</mo> <msub> <mi>U</mi> <mi>a</mi> </msub> <mo>,</mo> <msub> <mi>U</mi> <mi>x</mi> </msub> <mo>)</mo> </mrow> <mo>&times;</mo> <mrow> <mo>(</mo> <msub> <mi>R</mi> <mrow> <mi>x</mi> <mo>,</mo> <mi>i</mi> </mrow> </msub> <mo>-</mo> <mover> <msub> <mi>R</mi> <mi>x</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mo>)</mo> </mrow> </mrow> <mrow> <munder> <mi>&Sigma;</mi> <mrow> <msub> <mi>U</mi> <mi>x</mi> </msub> <mo>&Element;</mo> <mi>U</mi> </mrow> </munder> <msup> <mi>sim</mi> <mo>&prime;</mo> </msup> <mrow> <mo>(</mo> <msub> <mi>U</mi> <mi>a</mi> </msub> <mo>,</mo> <msub> <mi>U</mi> <mi>x</mi> </msub> <mo>)</mo> </mrow> </mrow> </mfrac> <mo>)</mo> </mrow> <mo>+</mo> <mrow> <mo>(</mo> <mn>1</mn> <mo>-</mo> <mi>&lambda;</mi> <mo>)</mo> </mrow> <mo>&times;</mo> <mrow> <mo>(</mo> <mover> <msub> <mi>R</mi> <mi>i</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>+</mo> <mfrac> <mrow> <munder> <mi>&Sigma;</mi> <mrow> <msub> <mi>I</mi> <mi>y</mi> </msub> <mo>&Element;</mo> <mi>I</mi> </mrow> </munder> <mrow> <mo>(</mo> <msup> <mi>sim</mi> <mo>&prime;</mo> </msup> <mrow> <mo>(</mo> <msub> <mi>I</mi> <mi>i</mi> </msub> <mo>,</mo> <msub> <mi>I</mi> <mi>y</mi> </msub> <mo>)</mo> </mrow> <mo>&times;</mo> <mrow> <mo>(</mo> <msub> <mi>R</mi> <mrow> <mi>a</mi> <mo>,</mo> <mi>y</mi> </mrow> </msub> <mo>-</mo> <mover> <msub> <mi>R</mi> <mi>y</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mo>)</mo> </mrow> </mrow> <mrow> <munder> <mi>&Sigma;</mi> <mrow> <msub> <mi>I</mi> <mi>y</mi> </msub> <mo>&Element;</mo> <mi>I</mi> </mrow> </munder> <msup> <mi>sim</mi> <mo>&prime;</mo> </msup> <mrow> <mo>(</mo> <msub> <mi>I</mi> <mi>i</mi> </msub> <mo>,</mo> <msub> <mi>I</mi> <mi>y</mi> </msub> <mo>)</mo> </mrow> </mrow> </mfrac> <mo>)</mo> </mrow> <mo>,</mo> </mrow> </math> Wherein
Figure BDA0000400762250000048
Figure BDA0000400762250000049
Respectively represent users Ua，UxThe average of the scores of the existing items,
Figure BDA00004007622500000410
Figure BDA00004007622500000411
respectively represent items Ii，IyThere is a mean value of the user scores.
Preferably, when λ =0, said Ra,iIs based on project similarity prediction score, when lambda =1, the Ra,iIs based on user similarity prediction scores.
According to the improved similarity measurement method in the collaborative filtering recommendation algorithm, provided by the invention, the similarity between users and the similarity between items are respectively calculated, and the similarity influence factor epsilon is defined to respectively correct the similarity values of the users and the items, so that the measurement mode of the similarity is improved. Meanwhile, according to the result of the improved similarity measurement, the average value of the user to the project score and other factors, the user score to the project is calculated, errors can be measured under the condition that data are extremely sparse, and therefore recommendation quality is improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
FIG. 1 is a flow chart of an improved similarity measure method in a collaborative filtering recommendation algorithm according to a preferred embodiment of the present invention;
FIG. 2 is a diagram illustrating a user-to-item scoring matrix R (n × m) according to a preferred embodiment of the present invention;
FIG. 3 is a flowchart of the collaborative filtering recommendation algorithm score prediction provided by the preferred embodiment of the present invention;
FIG. 4 is a flow chart of constructing a user or project similarity matrix according to the preferred embodiment of the present invention.
Detailed Description
The invention will be described in detail hereinafter with reference to the accompanying drawings in conjunction with embodiments. It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict.
Fig. 1 is a flowchart of an improved similarity measurement method in a collaborative filtering recommendation algorithm according to a preferred embodiment of the present invention. As shown in FIG. 1, the improved similarity measure method in the collaborative filtering recommendation algorithm according to the preferred embodiment of the present invention includes steps S1-S3.
Step S1: creating a set of users U = { U = }1,U2,…,UnN user pairs in the set of items I = { I = } {1,I2,…,ImScoring matrix R (n × m) of m items in (n) }, with Ra,iRepresenting a user UaFor item IiScore of (1), wherein Ua∈U,Ii∈I。
Specifically, FIG. 2 isThe preferred embodiment of the present invention provides a schematic diagram of a user-to-item scoring matrix R (n × m). As shown in fig. 2, the user-to-item scoring matrix R (n × m) has n rows and m columns, where n rows represent n users and m columns represent m items. If the user set is U and the project set is I, a certain user U is setaFor item Ii(wherein U isa∈U,IiE.g. I) is scored as Ra,iThen score Ra,iEmbodies the user UaFor item IiInterests and preferences.
Step S2: respectively calculate user UaAnd UbSimilarity sim (U) betweena,Ub) Item IiAnd IjSimilarity between them sim (I)i,Ij) Defining a similarity influencing factor epsilon, let sim' (U)a,Ub)=ε×sim(Ua,Ub)，sim'(Ii,Ij)=ε×sim(Ii,Ij)。
Specifically, the modified cosine similarity may modify the problem of deviation of different scoring metrics between different users in the cosine similarity measure method. Therefore, the present embodiment calculates the similarity sim (U) between users according to the modified cosine similarity measure methoda,Ub) And similarity sim (I) between itemsi,Ij)。
For example, if the user UaAnd user UbThe collective scored set of items is represented as
Figure BDA0000400762250000069
Figure BDA00004007622500000610
And
Figure BDA00004007622500000611
respectively represent users UaAnd user UbScored item set, user UaAnd user UbSimilarity between sim (U)a,Ub) Is shown as <math> <mrow> <mi>sim</mi> <mrow> <mo>(</mo> <msub> <mi>U</mi> <mi>a</mi> </msub> <mo>,</mo> <msub> <mi>U</mi> <mi>b</mi> </msub> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mrow> <munder> <mi>&Sigma;</mi> <mrow> <mi>k</mi> <mo>&Element;</mo> <msub> <mi>I</mi> <mrow> <msub> <mi>U</mi> <mi>a</mi> </msub> <msub> <mi>U</mi> <mi>b</mi> </msub> </mrow> </msub> </mrow> </munder> <mrow> <mo>(</mo> <msub> <mi>R</mi> <mrow> <mi>a</mi> <mo>,</mo> <mi>k</mi> </mrow> </msub> <mo>-</mo> <mover> <msub> <mi>R</mi> <mi>a</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mrow> <mo>(</mo> <msub> <mi>R</mi> <mrow> <mi>b</mi> <mo>,</mo> <mi>k</mi> </mrow> </msub> <mo>-</mo> <mover> <msub> <mi>R</mi> <mi>b</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> </mrow> <mrow> <msqrt> <munder> <mi>&Sigma;</mi> <mrow> <mi>k</mi> <mo>&Element;</mo> <msub> <mi>I</mi> <msub> <mi>U</mi> <mi>a</mi> </msub> </msub> </mrow> </munder> <msup> <mrow> <mo>(</mo> <msub> <mi>R</mi> <mrow> <mi>a</mi> <mo>,</mo> <mi>k</mi> </mrow> </msub> <mo>-</mo> <mover> <msub> <mi>R</mi> <mi>a</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mn>2</mn> </msup> </msqrt> <msqrt> <munder> <mi>&Sigma;</mi> <mrow> <mi>k</mi> <mo>&Element;</mo> <msub> <mi>I</mi> <msub> <mi>U</mi> <mi>b</mi> </msub> </msub> </mrow> </munder> <msup> <mrow> <mo>(</mo> <msub> <mi>R</mi> <mrow> <mi>b</mi> <mo>,</mo> <mi>k</mi> </mrow> </msub> <mo>-</mo> <mover> <msub> <mi>R</mi> <mi>b</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mn>2</mn> </msup> </msqrt> </mrow> </mfrac> <mo>.</mo> </mrow> </math> Wherein,and
Figure BDA0000400762250000063
respectively represent users UaAnd user UbAverage rating value, R, for given rated itemsa,kRepresenting a user UaFor item IkThe value of (a).
If the items I are to be paired togetheriAnd item IjThe set of users giving a score is represented as
Figure BDA00004007622500000613
And
Figure BDA00004007622500000614
respectively represent the pair item IiAnd item IjGiven a scored set of users, item IiAnd item IjSimilarity between them sim (I)i,Ij) Expressed as: <math> <mrow> <mi>sim</mi> <mrow> <mo>(</mo> <msub> <mi>I</mi> <mi>i</mi> </msub> <mo>,</mo> <msub> <mi>I</mi> <mi>j</mi> </msub> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mrow> <munder> <mi>&Sigma;</mi> <mrow> <mi>k</mi> <mo>&Element;</mo> <msub> <mi>U</mi> <mrow> <msub> <mi>I</mi> <mi>i</mi> </msub> <msub> <mi>I</mi> <mi>j</mi> </msub> </mrow> </msub> </mrow> </munder> <mrow> <mo>(</mo> <msub> <mi>R</mi> <mrow> <mi>k</mi> <mo>,</mo> <mi>i</mi> </mrow> </msub> <mo>-</mo> <mover> <msub> <mi>R</mi> <mi>i</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mrow> <mo>(</mo> <msub> <mi>R</mi> <mrow> <mi>k</mi> <mo>,</mo> <mi>j</mi> </mrow> </msub> <mo>-</mo> <mover> <msub> <mi>R</mi> <mi>j</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> </mrow> <mrow> <msqrt> <munder> <mi>&Sigma;</mi> <mrow> <mi>k</mi> <mo>&Element;</mo> <msub> <mi>U</mi> <msub> <mi>I</mi> <mi>i</mi> </msub> </msub> </mrow> </munder> <msup> <mrow> <mo>(</mo> <msub> <mi>R</mi> <mrow> <mi>k</mi> <mo>,</mo> <mi>i</mi> </mrow> </msub> <mo>-</mo> <mover> <msub> <mi>R</mi> <mi>i</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mn>2</mn> </msup> </msqrt> <msqrt> <munder> <mi>&Sigma;</mi> <mrow> <mi>k</mi> <mo>&Element;</mo> <msub> <mi>U</mi> <msub> <mi>I</mi> <mi>j</mi> </msub> </msub> </mrow> </munder> <msup> <mrow> <mo>(</mo> <msub> <mi>R</mi> <mrow> <mi>k</mi> <mo>,</mo> <mi>j</mi> </mrow> </msub> <mo>-</mo> <mover> <msub> <mi>R</mi> <mi>j</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mn>2</mn> </msup> </msqrt> </mrow> </mfrac> <mo>.</mo> </mrow> </math> wherein R isk,iRepresenting item IiBy user UkThe value of the score given is given by,
Figure BDA0000400762250000065
and
Figure BDA0000400762250000066
respectively represent items IiAnd item IjThe average of the scores is available.
In the embodiment, a similarity influence factor epsilon is introduced to correct the conventional similarity measurement method. When sim' (U)a,Ub)=ε×sim(Ua,Ub) When is in use, the
Figure BDA0000400762250000067
When sim' (I)i,Ij)=ε×sim(Ii,Ij) When is in use, the
Figure BDA0000400762250000068
Here, ε is 0. ltoreq. ε.ltoreq.1.
The similarity measure between users is taken as an example in the following, according to
Figure BDA0000400762250000076
Figure BDA0000400762250000077
And
Figure BDA0000400762250000078
different corresponding relations and different values of epsilon are explained.
When in use
Figure BDA0000400762250000079
When, ε =1Represents the user UaAnd UbAll the scored items are the same, the similarity values obtained by the traditional similarity measurement method can fully reflect the similarity between users, and the corrected user similarity meets sim' (U)a,Ub)=sim(Ua,Ub)。
When in use
Figure BDA00004007622500000710
When, ε =0, represents user UaAnd UbIf all the scored items are completely different, the similarity values obtained by the traditional similarity measurement method cannot explain the similarity between users, and sim' (U) is used at the momenta,Ub)=0。
When in useWhen is 0<ε<1, represents a user UaAnd UbThe scoring items have non-intersection items, the influence factors correct the traditional similarity metric value according to the proportion of the user common scoring items in the user scored item set, and the corrected similarity is sim' (U)a,Ub)=ε×sim(Ua,Ub)<sim(Ua,Ub)。
Similarly, for the measure of similarity between projects, the measure may be based on the condition of the projects that are commonly scored in the projects.
Step S3: and taking a parameter lambda in the [0,1] interval, and predicting the grade of the user to the project according to the lambda, the epsilon, the average value of the grade of the user to the project, the similarity between the users and the similarity between the projects.
Specifically, this step will predict the user's rating of the item based on the improved similarity metric results, resulting in a corresponding recommendation.
For example for user UaUnviewed item IiPredicting user UaFor the unviewed item IiIs scored as <math> <mrow> <msub> <mi>R</mi> <mrow> <mi>a</mi> <mo>,</mo> <mi>i</mi> </mrow> </msub> <mo>=</mo> <mi>&lambda;</mi> <mo>&times;</mo> <mrow> <mo>(</mo> <mover> <msub> <mi>R</mi> <mi>a</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>+</mo> <mfrac> <mrow> <munder> <mi>&Sigma;</mi> <mrow> <msub> <mi>U</mi> <mi>x</mi> </msub> <mo>&Element;</mo> <mi>U</mi> </mrow> </munder> <mrow> <mo>(</mo> <msup> <mi>sim</mi> <mo>&prime;</mo> </msup> <mrow> <mo>(</mo> <msub> <mi>U</mi> <mi>a</mi> </msub> <mo>,</mo> <msub> <mi>U</mi> <mi>x</mi> </msub> <mo>)</mo> </mrow> <mo>&times;</mo> <mrow> <mo>(</mo> <msub> <mi>R</mi> <mrow> <mi>x</mi> <mo>,</mo> <mi>i</mi> </mrow> </msub> <mo>-</mo> <mover> <msub> <mi>R</mi> <mi>x</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mo>)</mo> </mrow> </mrow> <mrow> <munder> <mi>&Sigma;</mi> <mrow> <msub> <mi>U</mi> <mi>x</mi> </msub> <mo>&Element;</mo> <mi>U</mi> </mrow> </munder> <msup> <mi>sim</mi> <mo>&prime;</mo> </msup> <mrow> <mo>(</mo> <msub> <mi>U</mi> <mi>a</mi> </msub> <mo>,</mo> <msub> <mi>U</mi> <mi>x</mi> </msub> <mo>)</mo> </mrow> </mrow> </mfrac> <mo>)</mo> </mrow> <mo>+</mo> <mrow> <mo>(</mo> <mn>1</mn> <mo>-</mo> <mi>&lambda;</mi> <mo>)</mo> </mrow> <mo>&times;</mo> <mrow> <mo>(</mo> <mover> <msub> <mi>R</mi> <mi>i</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>+</mo> <mfrac> <mrow> <munder> <mi>&Sigma;</mi> <mrow> <msub> <mi>I</mi> <mi>y</mi> </msub> <mo>&Element;</mo> <mi>I</mi> </mrow> </munder> <mrow> <mo>(</mo> <msup> <mi>sim</mi> <mo>&prime;</mo> </msup> <mrow> <mo>(</mo> <msub> <mi>I</mi> <mi>i</mi> </msub> <mo>,</mo> <msub> <mi>I</mi> <mi>y</mi> </msub> <mo>)</mo> </mrow> <mo>&times;</mo> <mrow> <mo>(</mo> <msub> <mi>R</mi> <mrow> <mi>a</mi> <mo>,</mo> <mi>y</mi> </mrow> </msub> <mo>-</mo> <mover> <msub> <mi>R</mi> <mi>y</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mo>)</mo> </mrow> </mrow> <mrow> <munder> <mi>&Sigma;</mi> <mrow> <msub> <mi>I</mi> <mi>y</mi> </msub> <mo>&Element;</mo> <mi>I</mi> </mrow> </munder> <msup> <mi>sim</mi> <mo>&prime;</mo> </msup> <mrow> <mo>(</mo> <msub> <mi>I</mi> <mi>i</mi> </msub> <mo>,</mo> <msub> <mi>I</mi> <mi>y</mi> </msub> <mo>)</mo> </mrow> </mrow> </mfrac> <mo>)</mo> </mrow> <mo>,</mo> </mrow> </math> Wherein
Figure BDA0000400762250000073
Respectively represent users Ua，UxThe average of the scores of the existing items,
Figure BDA0000400762250000074
Figure BDA0000400762250000075
respectively represent items Ii，IyThere is a mean value of the user scores.
In this embodiment, R is λ =0a,iIs based on project similarity prediction score, when lambda =1, Ra,iIs based on user similarity prediction scores.
FIG. 3 is a flowchart of the collaborative filtering recommendation algorithm score prediction provided by the preferred embodiment of the present invention. FIG. 4 is a flow chart of constructing a user or project similarity matrix according to the preferred embodiment of the present invention. As shown in fig. 3 and 4, the technical solution of the present invention can be better understood by referring to fig. 1.
In summary, according to the improved similarity measurement method in the collaborative filtering recommendation algorithm provided by the preferred embodiment of the present invention, the similarity between users and the similarity between items are respectively calculated, and the similarity influence factor epsilon is defined to respectively correct the similarity between the users and the items, so as to improve the measurement manner of the similarity. Meanwhile, according to the result of the improved similarity measurement, the average value of the user to the project score and other factors, the user score to the project is calculated, errors can be measured under the condition that data are extremely sparse, and therefore recommendation quality is improved.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein. The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims (6)

1. An improved similarity measurement method in a collaborative filtering recommendation algorithm is characterized by comprising the following steps:
s1, creating a user set U ═ U1,U2,…,UnN user pairs in the item set I ═ I1,I2,…,ImScoring matrix R (n × m) of m items in (n) }, with Ra,iRepresenting a user UaFor item IiScore of (1), wherein Ua∈U,Ii∈I；
S2, calculating the user U respectivelyaAnd UbBetweenSimilarity sim (U) ofa,Ub) Item IiAnd IjSimilarity between them sim (I)i,Ij) Defining a similarity influencing factor epsilon, let sim' (U)a,Ub)＝ε×sim(Ua,Ub)，sim'(Ii,Ij)＝ε×sim(Ii,Ij)；
And S3, taking a parameter lambda in the [0,1] interval, and predicting the grade of the user to the project according to the lambda, the epsilon, the average value of the grade of the user to the project, the similarity between the users and the similarity between the projects.
2. The method according to claim 1, wherein in step S2, the method comprises <math> <mrow> <mi>sim</mi> <mrow> <mo>(</mo> <msub> <mi>U</mi> <mi>a</mi> </msub> <mo>,</mo> <msub> <mi>U</mi> <mi>b</mi> </msub> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mrow> <munder> <mi>&Sigma;</mi> <mrow> <mi>k</mi> <mo>&Element;</mo> <msub> <mi>I</mi> <mrow> <msub> <mi>U</mi> <mi>a</mi> </msub> <msub> <mi>U</mi> <mi>b</mi> </msub> </mrow> </msub> </mrow> </munder> <mrow> <mo>(</mo> <msub> <mi>R</mi> <mrow> <mi>a</mi> <mo>,</mo> <mi>k</mi> </mrow> </msub> <mo>-</mo> <mover> <msub> <mi>R</mi> <mi>a</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mrow> <mo>(</mo> <msub> <mi>R</mi> <mrow> <mi>b</mi> <mo>,</mo> <mi>k</mi> </mrow> </msub> <mo>-</mo> <mover> <msub> <mi>R</mi> <mi>b</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> </mrow> <mrow> <msqrt> <munder> <mi>&Sigma;</mi> <mrow> <mi>k</mi> <mo>&Element;</mo> <msub> <mi>I</mi> <msub> <mi>U</mi> <mi>a</mi> </msub> </msub> </mrow> </munder> <msup> <mrow> <mo>(</mo> <msub> <mi>R</mi> <mrow> <mi>a</mi> <mo>,</mo> <mi>k</mi> </mrow> </msub> <mo>-</mo> <mover> <msub> <mi>R</mi> <mi>a</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mn>2</mn> </msup> </msqrt> <msqrt> <munder> <mi>&Sigma;</mi> <mrow> <mi>k</mi> <mo>&Element;</mo> <msub> <mi>I</mi> <msub> <mi>U</mi> <mi>b</mi> </msub> </msub> </mrow> </munder> <msup> <mrow> <mo>(</mo> <msub> <mi>R</mi> <mrow> <mi>b</mi> <mo>,</mo> <mi>k</mi> </mrow> </msub> <mo>-</mo> <mover> <msub> <mi>R</mi> <mi>b</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mn>2</mn> </msup> </msqrt> </mrow> </mfrac> <mo>,</mo> </mrow> </math> The above-mentioned <math> <mrow> <mi>sim</mi> <mrow> <mo>(</mo> <msub> <mi>I</mi> <mi>i</mi> </msub> <mo>,</mo> <msub> <mi>I</mi> <mi>j</mi> </msub> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mrow> <munder> <mi>&Sigma;</mi> <mrow> <mi>k</mi> <mo>&Element;</mo> <msub> <mi>U</mi> <mrow> <msub> <mi>I</mi> <mi>i</mi> </msub> <msub> <mi>I</mi> <mi>j</mi> </msub> </mrow> </msub> </mrow> </munder> <mrow> <mo>(</mo> <msub> <mi>R</mi> <mrow> <mi>k</mi> <mo>,</mo> <mi>i</mi> </mrow> </msub> <mo>-</mo> <mover> <msub> <mi>R</mi> <mi>i</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mrow> <mo>(</mo> <msub> <mi>R</mi> <mrow> <mi>k</mi> <mo>,</mo> <mi>j</mi> </mrow> </msub> <mo>-</mo> <mover> <msub> <mi>R</mi> <mi>j</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> </mrow> <mrow> <msqrt> <munder> <mi>&Sigma;</mi> <mrow> <mi>k</mi> <mo>&Element;</mo> <msub> <mi>U</mi> <msub> <mi>I</mi> <mi>i</mi> </msub> </msub> </mrow> </munder> <msup> <mrow> <mo>(</mo> <msub> <mi>R</mi> <mrow> <mi>k</mi> <mo>,</mo> <mi>i</mi> </mrow> </msub> <mo>-</mo> <mover> <msub> <mi>R</mi> <mi>i</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mn>2</mn> </msup> </msqrt> <msqrt> <munder> <mi>&Sigma;</mi> <mrow> <mi>k</mi> <mo>&Element;</mo> <msub> <mi>U</mi> <msub> <mi>I</mi> <mi>j</mi> </msub> </msub> </mrow> </munder> <msup> <mrow> <mo>(</mo> <msub> <mi>R</mi> <mrow> <mi>k</mi> <mo>,</mo> <mi>j</mi> </mrow> </msub> <mo>-</mo> <mover> <msub> <mi>R</mi> <mi>j</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mn>2</mn> </msup> </msqrt> </mrow> </mfrac> <mo>,</mo> </mrow> </math> Wherein
Figure FDA0000400762240000013
And
Figure FDA0000400762240000014
respectively represent users UaAnd user UbAverage score value for the scored item.
3. The method of claim 1, wherein in step S2, when sim' (U)a,Ub)＝ε×sim(Ua,Ub) When is in use, the
Figure FDA0000400762240000015
When sim' (I)i,Ij)＝ε×sim(Ii,Ij) When is in use, the
Figure FDA0000400762240000016
Wherein
Figure FDA0000400762240000017
For user UaAnd UbA set of commonly scored items,
Figure FDA0000400762240000018
And
Figure FDA0000400762240000019
are respectively a user UaAnd UbA scored set of items.
4. The method of claim 1, wherein 0 ≦ ε ≦ 1 in step S2.
5. The method of claim 1, wherein in step S3, a user U is predictedaFor the unviewed item IiIs scored as <math> <mrow> <msub> <mi>R</mi> <mrow> <mi>a</mi> <mo>,</mo> <mi>i</mi> </mrow> </msub> <mo>=</mo> <mi>&lambda;</mi> <mo>&times;</mo> <mrow> <mo>(</mo> <mover> <msub> <mi>R</mi> <mi>a</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>+</mo> <mfrac> <mrow> <munder> <mi>&Sigma;</mi> <mrow> <msub> <mi>U</mi> <mi>x</mi> </msub> <mo>&Element;</mo> <mi>U</mi> </mrow> </munder> <mrow> <mo>(</mo> <msup> <mi>sim</mi> <mo>&prime;</mo> </msup> <mrow> <mo>(</mo> <msub> <mi>U</mi> <mi>a</mi> </msub> <mo>,</mo> <msub> <mi>U</mi> <mi>x</mi> </msub> <mo>)</mo> </mrow> <mo>&times;</mo> <mrow> <mo>(</mo> <msub> <mi>R</mi> <mrow> <mi>x</mi> <mo>,</mo> <mi>i</mi> </mrow> </msub> <mo>-</mo> <mover> <msub> <mi>R</mi> <mi>x</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mo>)</mo> </mrow> </mrow> <mrow> <munder> <mi>&Sigma;</mi> <mrow> <msub> <mi>U</mi> <mi>x</mi> </msub> <mo>&Element;</mo> <mi>U</mi> </mrow> </munder> <msup> <mi>sim</mi> <mo>&prime;</mo> </msup> <mrow> <mo>(</mo> <msub> <mi>U</mi> <mi>a</mi> </msub> <mo>,</mo> <msub> <mi>U</mi> <mi>x</mi> </msub> <mo>)</mo> </mrow> </mrow> </mfrac> <mo>)</mo> </mrow> <mo>+</mo> <mrow> <mo>(</mo> <mn>1</mn> <mo>-</mo> <mi>&lambda;</mi> <mo>)</mo> </mrow> <mo>&times;</mo> <mrow> <mo>(</mo> <mover> <msub> <mi>R</mi> <mi>i</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>+</mo> <mfrac> <mrow> <munder> <mi>&Sigma;</mi> <mrow> <msub> <mi>I</mi> <mi>y</mi> </msub> <mo>&Element;</mo> <mi>I</mi> </mrow> </munder> <mrow> <mo>(</mo> <msup> <mi>sim</mi> <mo>&prime;</mo> </msup> <mrow> <mo>(</mo> <msub> <mi>I</mi> <mi>i</mi> </msub> <mo>,</mo> <msub> <mi>I</mi> <mi>y</mi> </msub> <mo>)</mo> </mrow> <mo>&times;</mo> <mrow> <mo>(</mo> <msub> <mi>R</mi> <mrow> <mi>a</mi> <mo>,</mo> <mi>y</mi> </mrow> </msub> <mo>-</mo> <mover> <msub> <mi>R</mi> <mi>y</mi> </msub> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mo>)</mo> </mrow> </mrow> <mrow> <munder> <mi>&Sigma;</mi> <mrow> <msub> <mi>I</mi> <mi>y</mi> </msub> <mo>&Element;</mo> <mi>I</mi> </mrow> </munder> <msup> <mi>sim</mi> <mo>&prime;</mo> </msup> <mrow> <mo>(</mo> <msub> <mi>I</mi> <mi>i</mi> </msub> <mo>,</mo> <msub> <mi>I</mi> <mi>y</mi> </msub> <mo>)</mo> </mrow> </mrow> </mfrac> <mo>)</mo> </mrow> <mo>,</mo> </mrow> </math> Wherein
Figure FDA0000400762240000022
Figure FDA0000400762240000023
Respectively represent users Ua，UxThe average of the scores of the existing items,
Figure FDA0000400762240000024
Figure FDA0000400762240000025
respectively represent items Ii，IyThere is a mean value of the user scores.
6. The method of claim 5, wherein R is when λ =0a,iIs based on project similarity prediction score, when lambda =1, the Ra,iIs based on user similarity prediction scores.
CN201310505323.1A 2013-10-23 2013-10-23 Similarity measuring method improved through collaborative filtering recommendation algorithm Pending CN103500228A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201310505323.1A CN103500228A (en) 2013-10-23 2013-10-23 Similarity measuring method improved through collaborative filtering recommendation algorithm

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201310505323.1A CN103500228A (en) 2013-10-23 2013-10-23 Similarity measuring method improved through collaborative filtering recommendation algorithm

Publications (1)

Publication Number Publication Date
CN103500228A true CN103500228A (en) 2014-01-08

Family

ID=49865439

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201310505323.1A Pending CN103500228A (en) 2013-10-23 2013-10-23 Similarity measuring method improved through collaborative filtering recommendation algorithm

Country Status (1)

Country Link
CN (1) CN103500228A (en)

Cited By (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103927347A (en) * 2014-04-01 2014-07-16 复旦大学 Collaborative filtering recommendation algorithm based on user behavior models and ant colony clustering
CN104021153A (en) * 2014-05-19 2014-09-03 江苏金智教育信息技术有限公司 Personalized recommendation method for campus books
CN104166732A (en) * 2014-08-29 2014-11-26 合肥工业大学 Project collaboration filtering recommendation method based on global scoring information
CN104182518A (en) * 2014-08-25 2014-12-03 苏州大学 Collaborative filtering recommendation method and device
CN104298772A (en) * 2014-10-29 2015-01-21 吴健 Collaborative filtering recommendation method and device optimizing neighbor selection
CN104715399A (en) * 2015-04-09 2015-06-17 苏州大学 Grading prediction method and grading prediction system
CN104731866A (en) * 2015-02-27 2015-06-24 湖南大学 Individual gourmet recommending method based on position
CN105843860A (en) * 2016-03-17 2016-08-10 山东大学 Microblog attention recommendation method based on parallel item-based collaborative filtering algorithm
CN109117442A (en) * 2017-06-23 2019-01-01 腾讯科技（深圳）有限公司 A kind of application recommended method and device
CN109389442A (en) * 2017-08-04 2019-02-26 北京京东尚科信息技术有限公司 Method of Commodity Recommendation and device, storage medium and electric terminal
CN111414533A (en) * 2019-01-04 2020-07-14 北京京东尚科信息技术有限公司 Recommendation information generation method and device, electronic equipment and storage medium

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101320461A (en) * 2008-07-01 2008-12-10 浙江大学 Cooperated filtering method based on resistor network and sparse data estimation
CN102609523A (en) * 2012-02-10 2012-07-25 上海视畅信息科技有限公司 Collaborative filtering recommendation algorithm based on article sorting and user sorting
CN102779131A (en) * 2011-05-12 2012-11-14 同济大学 Collaborative filtering recommending method based on multiple-similarity of users
CN103092911A (en) * 2012-11-20 2013-05-08 北京航空航天大学 K-neighbor-based collaborative filtering recommendation system for combining social label similarity

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101320461A (en) * 2008-07-01 2008-12-10 浙江大学 Cooperated filtering method based on resistor network and sparse data estimation
CN102779131A (en) * 2011-05-12 2012-11-14 同济大学 Collaborative filtering recommending method based on multiple-similarity of users
CN102609523A (en) * 2012-02-10 2012-07-25 上海视畅信息科技有限公司 Collaborative filtering recommendation algorithm based on article sorting and user sorting
CN103092911A (en) * 2012-11-20 2013-05-08 北京航空航天大学 K-neighbor-based collaborative filtering recommendation system for combining social label similarity

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
任磊: "推荐系统关键技术研究", 《中国博士学位论文全文数据库·信息科技辑》 *
吴月萍等: "改进相似性度量方法的协同过滤推荐算法", 《计算机应用与软件》 *
王晓堤等: "基于云模型的时间修正协同过滤推荐算法", 《计算机工程与科学》 *

Cited By (17)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103927347A (en) * 2014-04-01 2014-07-16 复旦大学 Collaborative filtering recommendation algorithm based on user behavior models and ant colony clustering
CN104021153A (en) * 2014-05-19 2014-09-03 江苏金智教育信息技术有限公司 Personalized recommendation method for campus books
CN104021153B (en) * 2014-05-19 2017-06-06 江苏金智教育信息技术有限公司 A kind of personalized recommendation method of campus books
CN104182518A (en) * 2014-08-25 2014-12-03 苏州大学 Collaborative filtering recommendation method and device
CN104182518B (en) * 2014-08-25 2017-12-26 苏州大学 A kind of collaborative filtering recommending method and device
CN104166732B (en) * 2014-08-29 2017-04-12 合肥工业大学 Project collaboration filtering recommendation method based on global scoring information
CN104166732A (en) * 2014-08-29 2014-11-26 合肥工业大学 Project collaboration filtering recommendation method based on global scoring information
CN104298772A (en) * 2014-10-29 2015-01-21 吴健 Collaborative filtering recommendation method and device optimizing neighbor selection
CN104731866A (en) * 2015-02-27 2015-06-24 湖南大学 Individual gourmet recommending method based on position
CN104731866B (en) * 2015-02-27 2020-05-19 湖南松桂坊电子商务有限公司 Personalized food recommendation method based on position
CN104715399A (en) * 2015-04-09 2015-06-17 苏州大学 Grading prediction method and grading prediction system
CN104715399B (en) * 2015-04-09 2018-03-02 苏州大学 A kind of score in predicting method and system
CN105843860A (en) * 2016-03-17 2016-08-10 山东大学 Microblog attention recommendation method based on parallel item-based collaborative filtering algorithm
CN105843860B (en) * 2016-03-17 2019-03-22 山东大学 A kind of microblogging concern recommended method based on parallel item-based collaborative filtering
CN109117442A (en) * 2017-06-23 2019-01-01 腾讯科技（深圳）有限公司 A kind of application recommended method and device
CN109389442A (en) * 2017-08-04 2019-02-26 北京京东尚科信息技术有限公司 Method of Commodity Recommendation and device, storage medium and electric terminal
CN111414533A (en) * 2019-01-04 2020-07-14 北京京东尚科信息技术有限公司 Recommendation information generation method and device, electronic equipment and storage medium

Similar Documents

Publication Publication Date Title
CN103500228A (en) Similarity measuring method improved through collaborative filtering recommendation algorithm
Park et al. Reversed CF: A fast collaborative filtering algorithm using a k-nearest neighbor graph
CN104935963B (en) A kind of video recommendation method based on timing driving
Ramezani et al. A pattern mining approach to enhance the accuracy of collaborative filtering in sparse data domains
CN109933721B (en) Interpretable recommendation method integrating user implicit article preference and implicit trust
CN102779182A (en) Collaborative filtering recommendation method for integrating preference relationship and trust relationship
CN102495864A (en) Collaborative filtering recommending method and system based on grading
Jiao et al. A novel learning rate function and its application on the SVD++ recommendation algorithm
Hu et al. Bayesian personalized ranking based on multiple-layer neighborhoods
Lu et al. Scalable news recommendation using multi-dimensional similarity and Jaccard–Kmeans clustering
CN106126549A (en) A kind of community&#39;s trust recommendation method decomposed based on probability matrix and system thereof
CN106294859A (en) A kind of item recommendation method decomposed based on attribute coupling matrix
CN107491557A (en) A kind of TopN collaborative filtering recommending methods based on difference privacy
CN102135999A (en) User credibility and item nearest neighbor combination Internet recommendation method
CN105430505A (en) IPTV program recommending method based on combined strategy
Zarei et al. A memory-based collaborative filtering recommender system using social ties
Chen et al. A fuzzy matrix factor recommendation method with forgetting function and user features
CN110059257B (en) Project recommendation method based on score correction
Ha et al. Social filtering using social relationship for movie recommendation
Ali et al. Inferring context with reliable collaborators: a novel similarity estimation method for recommender systems
Margaris et al. Enhancing rating prediction quality through improving the accuracy of detection of shifts in rating practices
Chen et al. Trust-based collaborative filtering algorithm in social network
Sreepada et al. Revisiting tendency based collaborative filtering for personalized recommendations
CN114912031A (en) Mixed recommendation method and system based on clustering and collaborative filtering
JP2008305229A (en) Demand forecast method and device

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C02 Deemed withdrawal of patent application after publication (patent law 2001)
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20140108