CN109118379A - Recommended method and device based on social networks - Google Patents
Recommended method and device based on social networks Download PDFInfo
- Publication number
- CN109118379A CN109118379A CN201710483555.XA CN201710483555A CN109118379A CN 109118379 A CN109118379 A CN 109118379A CN 201710483555 A CN201710483555 A CN 201710483555A CN 109118379 A CN109118379 A CN 109118379A
- Authority
- CN
- China
- Prior art keywords
- node
- user
- target object
- recommended
- social networks
- 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 60
- 238000005295 random walk Methods 0.000 claims description 28
- 230000006399 behavior Effects 0.000 claims description 19
- 238000004422 calculation algorithm Methods 0.000 claims description 9
- 238000011156 evaluation Methods 0.000 claims description 6
- 238000004590 computer program Methods 0.000 claims 2
- 230000000694 effects Effects 0.000 abstract description 16
- 238000010586 diagram Methods 0.000 description 18
- 230000006870 function Effects 0.000 description 9
- 230000008569 process Effects 0.000 description 8
- 238000004364 calculation method Methods 0.000 description 6
- 230000009191 jumping Effects 0.000 description 6
- 230000008859 change Effects 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 5
- 230000005012 migration Effects 0.000 description 5
- 238000013508 migration Methods 0.000 description 5
- 230000008901 benefit Effects 0.000 description 3
- 235000013399 edible fruits Nutrition 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 230000006855 networking Effects 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 241000406668 Loxodonta cyclotis Species 0.000 description 1
- 230000001174 ascending effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000000151 deposition Methods 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 210000003733 optic disk Anatomy 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000012545 processing 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
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/01—Social networking
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Tourism & Hospitality (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention proposes a kind of recommended method and device based on social networks, wherein, method includes: by according to the social networks between user each in social networks, and the similitude obtained between each target object establishes figure recommended models, and then it calculates in figure recommended models, the centrality of each node, so that according to the centrality of each node, the corresponding target object of the node is recommended the corresponding user of the node.The figure recommended models are the graph models based on Node connectedness relationship, by the instruction of Node connectedness relationship, there are the similitudes corresponded between target object to be recommended in the user of social networks, the recommendation of target object is combined with the social networks between user to realize, solves the bad technical problem of the recommendation effect of target object in the prior art.
Description
Technical field
The present invention relates to Internet technical field more particularly to a kind of recommended methods and device based on social networks.
Background technique
When carrying out the recommendation based on social networks, has existed in the prior art and the emerging of user is indicated according to user tag
Matched product such as application program, commodity etc. is recommended user by the way of tag match by interest.
But this mode, due to needing to identify the interest of user, and the accuracy of recognition result is not high, so as to cause
Recommendation effect is bad.
Summary of the invention
The present invention is directed to solve at least some of the technical problems in related technologies.
For this purpose, present invention aims at a kind of recommended method and device based on social networks is proposed, by by target pair
The recommendation of elephant is combined with the social networks between user, to realize for some user, according to there are social networks
Other users similar purpose object to be recommended, determine the target object that need to recommend to the user, improve user and push away
The matching degree for the target object recommended solves the bad technical problem of recommendation effect in the prior art.
In order to achieve the above object, first aspect present invention embodiment proposes a kind of recommended method based on social networks, packet
Include following steps:
The social networks in social networks between each user are obtained, and obtain the similitude between each target object;
According to the social networks and the similitude, figure recommended models are established;Wherein, each in the figure recommended models
The connection relationship of node is used to indicate in the user there are social networks, the similitude between corresponding target object to be recommended;
It calculates in the figure recommended models, the centrality of each node;Wherein, the centrality is according to corresponding node in institute
State the quantity determination of institute's link node in figure recommended models;
According to the centrality of each node, the corresponding target object of the node is recommended into the corresponding user of the node.
Optionally, in one embodiment of this invention, the social networks include that user is paid close attention to involved in concern behavior
Be concerned user;It is described to establish figure recommended models, comprising:
Generate each node in the figure recommended models;Wherein, each node corresponds to the user in the social networks, and
User target object to be recommended;
For in the figure recommended models, there is similitude, and deposit between corresponding target object between corresponding user
In two nodes of concern behavior, described two nodes are connected using side;The direction on the side is paid close attention to from described two nodes
The corresponding node of user, which is directed toward in described two nodes, is concerned the corresponding node of user.
Optionally, in one embodiment of this invention, the similitude obtained between each target object, comprising:
According to the degree of correlation between the evaluation of the degree of correlation and each target object between the content of each target object, meter
Calculation obtains the similitude between each target object.
Optionally, in one embodiment of this invention, the centrality according to each node, by the corresponding mesh of the node
Object recommendation is marked to the corresponding user of the node, comprising:
According to the centrality of each node, it is ranked up for the node of the same user of correspondence;
According to the sequence, the corresponding target object of the node is recommended into the corresponding user of the node.
Optionally, in one embodiment of this invention, described according to the social networks and the similitude, establish figure
Before recommended models, further includes:
Determine the target object set comprising each target object;
According to the target object that each user browses, the subset of the target object set is generated respectively for each user;
Using the target object that the target object in the subset is to be recommended as corresponding user.
Optionally, in one embodiment of this invention, the centrality includes PageRank value;The calculating figure pushes away
It recommends in model, the centrality of each node, comprising:
The figure recommended models are calculated using Random Walk Algorithm, obtain the PageRank value of each node.
Optionally, in one embodiment of this invention, described that the figure recommended models are carried out using Random Walk Algorithm
It calculates, obtains the PageRank value of each node, comprising:
The random walk since each of figure recommended models node, node n is passed through in random walk path every timei
When, node n is jumped to probability c random selectioniClose on node njOn, it is pushed away alternatively, jumping to the figure at random with probability 1-c
It recommends any node in model and terminates random walk;
Node n in the figure recommended modelsiPageRank value PR (ni) be
Wherein, c is to jump probability, outjFor node njOut-degree, Vi TIt is to jump to node n when jumping generation at randomi's
Probability.
Optionally, in one embodiment of this invention, the Vi TIt is according to node njPriori knowledge determine;The elder generation
Testing knowledge includes node njNode n described in corresponding user preferencejThe determination degree of corresponding target object.
The recommended method based on social networks of the embodiment of the present invention, by according to the society between user each in social networks
Friendship relationship, and the similitude obtained between each target object establish figure recommended models, and then calculate in figure recommended models, each to tie
The centrality of point, so that it is corresponding that the corresponding target object of the node is recommended the node according to the centrality of each node
User.The figure recommended models are the graph models based on Node connectedness relationship, indicate there is social close by Node connectedness relationship
The similitude between target object to be recommended is corresponded in the user of system, to realize between recommendation and user by target object
Social networks combine, solve the bad technical problem of the recommendation effect of target object in the prior art.
In order to achieve the above object, second aspect of the present invention embodiment proposes a kind of recommendation apparatus based on social networks, packet
It includes:
Obtain module, for obtaining the social networks in social networks between each user, and obtain each target object it
Between similitude;
Module is established, for establishing figure recommended models according to the social networks and the similitude;Wherein, described
The connection relationship of each node is used to indicate in the user there are social networks in figure recommended models, corresponding target object to be recommended it
Between similitude;
Computing module, for calculating in the figure recommended models, the centrality of each node;Wherein, the centrality is root
It is determined according to the quantity of corresponding node institute's link node in the figure recommended models;
The corresponding target object of the node is recommended the knot for the centrality according to each node by recommending module
The corresponding user of point.
Optionally, in one embodiment of this invention, the social networks include that user is paid close attention to involved in concern behavior
Be concerned user;It is described to establish module, comprising:
Generation unit, for generating each node in the figure recommended models;Wherein, each node corresponds to the social networks
In user and the user target object to be recommended;
Connection unit, for similitude between corresponding target object, and corresponding in the figure recommended models
User between there are two nodes of concern behavior, described two nodes are connected using side;The direction on the side, from described two
It is paid close attention in a node in the described two nodes of the corresponding node direction of user and is concerned the corresponding node of user.
Optionally, in one embodiment of this invention, the acquisition module, is specifically used for:
According to the degree of correlation between the evaluation of the degree of correlation and each target object between the content of each target object, meter
Calculation obtains the similitude between each target object.
Optionally, in one embodiment of this invention, the recommending module, comprising:
Determination unit is ranked up for the centrality according to each node for the node of the same user of correspondence;
Recommendation unit, for it is corresponding that the corresponding target object of the node to be recommended the node according to the sequence
User.
Optionally, in one embodiment of this invention, described device, further includes:
Object determining module, for determining the target object set comprising each target object;The mesh browsed according to each user
Object is marked, generates the subset of the target object set respectively for each user;Using the target object in the subset as pair
Using the target object that family is to be recommended.
Optionally, in one embodiment of this invention, the centrality includes PageRank value;The computing module, tool
Body is used for:
The figure recommended models are calculated using Random Walk Algorithm, obtain the PageRank value of each node.
Optionally, in one embodiment of this invention, the computing module, comprising:
Migration unit, for the random walk since each of figure recommended models node, random walk path
Pass through node n every timeiWhen, node n is jumped to probability c random selectioniClose on node njOn, alternatively, random with probability 1-c
It jumps to any node in the figure recommended models and terminates random walk;
Probability calculation unit, for the node n in the figure recommended modelsiPageRank value PR (ni) be
Wherein, c is to jump probability, outjFor node njOut-degree,It is to jump to node n when jumping generation at randomi's
Probability.
Optionally, in one embodiment of this invention, the Vi TIt is according to node njPriori knowledge determine;The elder generation
Testing knowledge includes node njNode n described in corresponding user preferencejThe determination degree of corresponding target object.
The recommendation apparatus based on social networks of the embodiment of the present invention is closed according to the social activity between user each in social networks
System, and the similitude obtained between each target object establish figure recommended models, and then calculate in figure recommended models, each node
Centrality, so that according to the centrality of each node, the corresponding target object of the node is recommended the corresponding use of the node
Family.The figure recommended models are the graph models based on Node connectedness relationship, and by the instruction of Node connectedness relationship, there are social networks
The similitude between target object to be recommended is corresponded in user, to realize the society between the recommendation and user by target object
Friendship relationship combines, and solves the bad technical problem of the recommendation effect of target object in the prior art.
In order to achieve the above object, third aspect present invention embodiment proposes a kind of computer equipment, comprising: processor;With
In the memory for storing the processor-executable instruction;Wherein, the processor is for executing method described in first aspect.
To achieve the goals above, fourth aspect present invention embodiment proposes a kind of computer readable storage medium, when
Instruction in the storage medium is performed by processor, enables a processor to execute method described in first aspect.
The additional aspect of the present invention and advantage will be set forth in part in the description, and will partially become from the following description
Obviously, or practice through the invention is recognized.
Detailed description of the invention
Above-mentioned and/or additional aspect and advantage of the invention will become from the following description of the accompanying drawings of embodiments
Obviously and it is readily appreciated that, in which:
Fig. 1 is a kind of flow diagram of the recommended method based on social networks provided by the embodiment of the present invention;
Fig. 2 is the flow diagram of recommended method of the another kind based on social networks provided by the embodiment of the present invention;
Fig. 3 is the schematic diagram of figure recommended models;
Fig. 4 is the flow diagram of another recommended method based on social networks provided by the embodiment of the present invention;
Fig. 5 is the schematic diagram of the PageRank value of each node in figure recommended models;
Fig. 6 is the schematic diagram for introducing the figure recommended models of priori knowledge;
Fig. 7 is a kind of structural schematic diagram of the recommendation apparatus based on social networks provided in an embodiment of the present invention;
Fig. 8 is the structural schematic diagram of another recommendation apparatus based on social networks provided in an embodiment of the present invention;And
Fig. 9 is the structural schematic diagram of the recommender system provided in an embodiment of the present invention based on social networks.
Specific embodiment
The embodiment of the present invention is described below in detail, examples of the embodiments are shown in the accompanying drawings, wherein from beginning to end
Same or similar label indicates same or similar element or element with the same or similar functions.Below with reference to attached
The embodiment of figure description is exemplary, it is intended to is used to explain the present invention, and is not considered as limiting the invention.
Product used by a user, referred to herein as target object, including application program, commodity etc. often exist larger
Otherness, different user according to respective preference selection needed for target object.In order to recommend that it is suitble to use to user
Target object by the way of a variety of recommendation target objects, selects two ways to be briefly situated between below in the prior art
It continues.
First way is based on matched recommended method.This method carrys out identity user first with a series of label
Interest and target object attribute, most matched target object is then recommended into user by way of tag match.Thing
In reality, as matching process, accuracy is highly dependent on the extraction of label and the accuracy of generation phase.
Second master slave mode excavates the point of interest of user in turn based on being the search log or download log by user
The method recommended.The advantages of such method, is that the interest of user is entirely to be dug by the search log of user or download log
Pick obtains, simple and easy without the manual participation of user.But discovery recommendation effect is bad in actual use.Because of user
The point of interest of the history point of interest current with user has differences, to be difficult to go out current interest according to historical interest point prediction
Point.
For the bad problem of recommendation effect, the embodiment of the invention provides the recommended methods based on social networks.This hair
The method that bright embodiment provides, core concept are: if a certain user is generally used there are the other users of social networks
A certain similar target object, even when the user is not directed to the interesting measure of this target object, it also should be by this target
Object recommendation is to user.Based on the aforementioned description for the prior art it is found that existing recommended method do not have will be between user
The social networks such as mutual concern be introduced into figure recommended models, the reason is that the social pass such as mutual concern between user
System, and the log by user, by taking target object is application program (APP) as an example, such as from searching for log, installation log, individual
Correlation between the application program that label obtains is two category informations, is difficult the social networks and the phase of target object between user
Mutual relation is used in conjunction with.In the embodiment of the present invention, these two types of information are combined together to obtain by figure recommended models
Obtain more reasonable recommendation results.
Below with reference to the accompanying drawings the recommended method and device based on social networks of the embodiment of the present invention are described.
Fig. 1 is a kind of flow diagram of the recommended method based on social networks provided by the embodiment of the present invention, such as Fig. 1
It is shown, should recommended method based on social networks the following steps are included:
Step 101, the social networks in social networks between each user are obtained, and obtain the phase between each target object
Like property.
Specifically, social networks may include concern user involved in concern behavior and be concerned user, here
Concern behavior, usually in concern user to occurring under being concerned the interested situation of user.In addition, social networks can be with
User etc. is initiated including being added user and addition involved in plusing good friend behavior.
Target object may include various products, and product here includes virtual product and physical product, such as: it is virtual
Product it is relatively common be application program.Can according between the content of each target object the degree of correlation and each target object
Evaluation between the degree of correlation, the similitude between each target object is calculated.
Step 102, according to social networks and similitude, figure recommended models are established.
Wherein, the connection relationship of each node is used to indicate in the user there are social networks in figure recommended models, it is corresponding to
Recommend the similitude between target object.As a kind of possible implementation, figure recommended models include each node, and node is corresponding
User in social networks, and user's target object to be recommended is corresponded to, two nodes interconnected in figure recommended models
There are social networks between corresponding user, and there are similitudes between corresponding target object.
As concern user involved in social networks include concern behavior and when being concerned user, figure can be firstly generated and pushed away
Each node in model is recommended, and then in figure recommended models, there is similitude and corresponding use between corresponding target object
There are two nodes of concern behavior between family, connect two nodes using side.Wherein, the direction on side is closed from two nodes
The corresponding node of note user, which is directed toward in described two nodes, is concerned the corresponding node of user.
Step 103, it calculates in figure recommended models, the centrality of each node.
Wherein, centrality is determined according to the quantity of corresponding node institute's link node in figure recommended models.
Step 104, according to the centrality of each node, the corresponding target object of node is recommended into the corresponding user of node.
Specifically, it according to the centrality of each node, determines and the corresponding target object of node is recommended into the corresponding use of node
The probability at family.If probability is higher than preset threshold, the corresponding target object of node is recommended into the corresponding user of node.
In the present embodiment, by according to the social networks between user each in social networks, and each target object of acquisition
Between similitude establish figure recommended models, and then calculate in figure recommended models, the centrality of each node, so as to according to each node
Centrality, the corresponding target object of the node is recommended into the corresponding user of the node.The figure recommended models are to be based on
The graph model of Node connectedness relationship, by the instruction of Node connectedness relationship, there are correspond to target pair to be recommended in the user of social networks
Similitude as between combines the recommendation of target object with the social networks between user to realize, and solves existing
The technical problem for having the recommendation effect of target object in technology bad.
In order to establish figure recommended models in an embodiment in clear explanation, present embodiments provide another based on social network
The recommended method of network, it is main according to concern behavior in the present embodiment, determine the social relationships between user.This is because, concern
Compared to other social actions, it is related more to embody interest between user for behavior, in other words, user often with it is another
In the relevant situation of one user interest, concern behavior can just occur, be used as in the present embodiment using concern behavior and generate graph model
Foundation, the accuracy of recommendation can be effectively improved.Fig. 2 is another kind provided by the embodiment of the present invention based on social networks
The flow diagram of recommended method.
As shown in Fig. 2, this method may comprise steps of:
Step 201, user's set and the concern relation between target object set and user are determined.
Specifically, in the present embodiment, target object can be application program, thus be APP by target object aggregated label,
User's aggregated label is U.In this step, it is also necessary to obtain any user u in user's set UrBuddy list be denoted as F (ur)。
Step 202, it calculates in target object set, the similitude between any pair of target object.
Specifically, in target object set APP, between any pair of target object app
Step 203, it obtains in user's set, each user target object to be recommended.
Specifically, it is determined that after the target object set comprising each target object, for each use in the social networks
Family generates the subset of the target object set respectively, and the target object in the subset is to be recommended as corresponding user
Target object.
A kind of each user u as possible implementation, in user's set UrWith an APP subset App (ur),
App(ur) ∈ APP, it is urTarget object set to be recommended, to user urThe APP of recommendation is both from this set.The collection
Conjunction can simply use target object set APP, or according to certain methods, such as using traditional recommended models, determine
A subset is as the user u in target object set APPrTarget object set APP (u to be recommendedr)。
Furthermore it is also possible to by user urThe object of browsing is as target object set APP (ur)。
Step 204, the target object to be recommended according to user generates each node in figure recommended models.
Wherein, a target in the corresponding user of each node and user target object list to be recommended
Object.
Specifically, by the user u in user's set UrAnd its set APP (u to be recommendedr) in some appp, such as
appp, a user-target object pair is constituted, and by user-target object to a node being considered as in figure recommended models, mark
Knowing is N (ur,appp)。
Step 205, in the figure recommended models, there is similitude, and corresponding use between corresponding target object
There are two nodes of concern behavior between family, connect two nodes using side, generate figure recommended models.
Wherein, the direction on side is concerned from being paid close attention in the described two nodes of the corresponding node direction of user in two nodes
The corresponding node of user.
Specifically, it is made of due to the node in figure recommended models user and target object two parts, so if
There are sides must satisfy following two condition between node:
1) there is concern between the user for including in two nodes;
2) there is similitude between the target object for including in two nodes.
Meanwhile the node of the different target object for the same user of correspondence, it can be connected using double-head arrow, that is, this
Side is two-way between two nodes, is considered as each user and is defaulted as paying close attention to itself.
When the above two conditions are met, if user urPay close attention to user ut, then the direction on side is by utThe node at place is directed toward
urThe node at place.Figure recommended models can be formed by method as above.
In order to clearly illustrate the present embodiment, a kind of possible application scenarios are present embodiments provided.
Assuming that being respectively { u there are five user1,u2,u3,u4,u5, target object collection is combined into { app1,app2,…,app9}。
Wherein, social relationships are as follows:
u1Pay close attention to u3And u5;
u2And u3Mutually concern;
u4Pay close attention to u5;
u5Pay close attention to u1。
Target object similitude are as follows: (belonging to has similitude between the target object of identity set)
{app1,app3,app5};
{app1,app6};
{app2,app6};
{app1,app7};
{app4,app6,app8};
{app4,app9}。
The target object set to be recommended of each user are as follows:
u1={ app1,app3,app6,app7};
u2={ app1,app5};
u3={ app1,app3,app5,app8};
u4={ app2,app4,app6,app9};
u5={ app4,app6,app7,app8}。
To form figure recommended models as shown in Figure 3.Fig. 3 is the schematic diagram of figure recommended models, as shown in figure 3, including
Following node:
Corresponding user u1Node, i.e. N (u1,app1), N (u1,app3),N(u1,app6),N(u1,app7);
Corresponding user u2Node, i.e. N (u2,app1), N (u2,app5);
Corresponding user u3Node, i.e. N (u3,app1), N (u3,app3),N(u3,app5),N(u3,app8);
Corresponding user u4Node, i.e. N (u3,app2),N(u3,app4),N(u3,app6),N(u3,app9);
Corresponding user u5Node, i.e. N (u5,app4), N (u5,app6), N (u5,app7), N (u5,app8)。
In turn, due to u1Pay close attention to u3And u5, corresponding user u1Node and corresponding user u3Node, corresponding user u5Knot
There are side between point, corresponding user u is directed toward on side1Node.
In addition similarly, u2And u3Mutually concern, corresponding user u2Node, corresponding user u3Node between there are side,
It is directed toward corresponding user u simultaneously in side2Node, corresponding user u3Node;
u4Pay close attention to u5, corresponding user u4Node, corresponding user u5Node between there are side, corresponding user u is directed toward on side4's
Node;
u5Pay close attention to u1, corresponding user u1Node, corresponding user u5Node between there are side, corresponding user u is directed toward on side5's
Node.
In turn, it is being formed between figure recommended models, Monte Carlo Random Walk Algorithm can be used and calculate in figure each
The centrality of node is based on centrality, the target object recommended needed for determining.
In the present embodiment, by according to the social networks between user each in social networks, and each target object of acquisition
Between similitude establish figure recommended models, and then calculate in figure recommended models, the centrality of each node, so as to according to each node
Centrality, the corresponding target object of the node is recommended into the corresponding user of the node.The figure recommended models are to be based on
The graph model of Node connectedness relationship, by the instruction of Node connectedness relationship, there are correspond to target pair to be recommended in the user of social networks
Similitude as between combines the recommendation of target object with the social networks between user to realize, and solves existing
The technical problem for having the recommendation effect of target object in technology bad.
How to be recommended based on figure recommended models to be established in an embodiment in clear explanation after figure recommended models,
It present embodiments provides another recommended method based on social networks, in the present embodiment, is pushed away mainly for application program
It recommends, that is to say, that target object is specially application program.
Application program is due to being different from the physical commodities such as common clothes, tool, daily necessities, and there are following three for application program
A significant difference:
First, classification is many kinds of so that being not easy to generate hobby for different types of application program;
Second, lack enough description informations, since application program is program class product, is difficult as common product one
Sample is angularly clearly depicted from performance, availability;
Third, closely, such as many shopping class applications will answer the fitness between different application with payment class
With matching, therefore when recommending a certain shopping class to user in application, should also recommend to user and its close-fitting payment
The characteristics of class application, the correlation between this application is do not had in traditional Products Show problem.
Therefore, in the prior art, the recommendation results that the recommended method of user journal has tended not to are relied on.
In the present embodiment, it is foundation to what is generated that figure recommended models, which are with the mutual concern between user, passes through this
User is paid close attention to behavior and is combined with application program similitude by figure recommended models, thus realize application program recommended method, this
The core of kind application program recommended method is: if many users of user's concern use similar application program,
Even if the user is not directed to the interesting measure of this application program, also this application program should be recommended user.
Figure recommended models are generated based on previous embodiment in the present embodiment, for carrying out the recommendation of application program, Neng Gouyou
Effect improves the accuracy recommended.It will describe in detail below to recommended method, Fig. 4 is another provided by the embodiment of the present invention
The flow diagram of recommended method of the kind based on social networks.
As shown in figure 4, after the figure recommended models that previous embodiment generates, i.e., after step 205, further includes:
Step 301, the figure recommended models are calculated using Random Walk Algorithm, obtains the centrality of each node.
Wherein, centrality is determined according to the quantity of corresponding node institute's link node in figure recommended models, for referring to
Show the significance level of node.
Specifically, Monte Carlo Random Walk Algorithm can be used to calculate in figure recommended models, the centrality of each node,
In the present embodiment, specifically the centrality of node can be indicated using the centrality of PageRank value instruction node, PageRank value
It is also more conventional and effective.
The centrality of node, such as PageRank value are the importance that node is determined according to the connection relationship between node.
The principle of its foundation can be sketched are as follows: if a node pointed by multiple neighborhood of nodes, and the centrality of its neighborhood of nodes
When larger, then the centrality of this node is also relatively large.This property is consistent with the description in the present embodiment for node,
I.e. if there is user urWith application program appp, wherein urConcern it is multiple with per family to application program apppWith similar
The application program of function is interested, then user urTo application program apppInterested possibility be also it is very big, i.e., should
By application program apppRecommend user ur。
It follows that the centrality of node can reflect out the use recommended the application program in the node in the node
The feasibility at family.
The PageRank value that node can be very easily calculated using Monte Carlo Random Walk Algorithm, below will be to meter
Calculation process describes in detail:
Assuming that there is n node in figure recommended models, then repeated random walk m time since each node, formation nm item with
Machine migration path.Passing through node n every timeiWhen, n is randomly choosed with probability ciCertain side jump to and niThe node n closed onjOn,
Or it is jumped on any one node in figure at random with probability 1-c, and terminate random walk.
Node niThe exact value of PageRank can be calculated by following formula:
Wherein, c is to jump probability, outjFor node njOut-degree, Vi TIt is to jump to node n when jumping generation at randomi's
Probability can be set as 1 under the premise of no priori knowledge.
It should be noted that if node n is passed through in nm random walk path of noteiNumber be xi, then have node ni's
PageRank statistical value are as follows:Wherein, 1/ (1-c) is the desired length in random walk path.
Pass through PR (ni)=((1-c)/nm) E (xi) provable π (ni) desired value and PR (ni) it is equal.It can by this proof
Know, the PageRank value of node can be calculated using Monte Carlo random walk.
Step 302, it according to the centrality of each node, is ranked up for the node of the same user of correspondence.
Specifically, PageRank value is bigger, sorts more forward, that is to say, that according to ascending suitable of PageRank value
Sequence is ranked up for the node of the same user of correspondence.
Step 303, according to the sequence, the corresponding target object of the node is recommended into the corresponding use of the node
Family.
Specifically, the node for obtaining predetermined number before being ordered as, recommends the knot for the corresponding target object of these nodes
The corresponding user of point.
In order to clearly illustrate the present embodiment, provided by Fig. 3 on the basis of figure recommended models, operation is carried out, has been obtained
The PageRank value of each node.Fig. 5 is the schematic diagram of the PageRank value of each node in figure recommended models.
Based on Fig. 3 as it can be seen that due to user u1Pay close attention to u2And u3, and u2And u3Mutually concern, it means that user u1、u2And u3
With similar hobby, simultaneously because app1It appears in the application APP set to be recommended of these three users, anticipates
Taste app1A possibility that being liked by these three users should be very big, therefore should be by u1Recommend these three users.As a result
As shown in Figure 5, i.e. node N (u1,app1)、N(u2,app1) and N (u3,app1) PageRank value it is larger.This result shows that:
The mutual concern that this figure recommended models can use between user is excavated the point of interest of user and then is recommended to user interested
Application program.
It is also shown by Fig. 3, due to app1With app6It is interrelated, it means that by app1Recommend u1While, also answer
This is by related application app6U is recommended together1, the result in Fig. 5 just so, i.e. node N (u1,app6) PageRank value
Also larger.This result shows that: this figure recommended models can will have it is correlative application recommend user together, facilitate user
Use.
The above-mentioned time complexity for calculating PageRank value by random walk is O (nm/ (1-c)).According to Google
The data provided, c are generally set to 0.85, i.e. the time complexity of the calculating process is O (nm).In fact many articles are pointed out,
When m is 1, calculating effect, that is, pretty good, therefore m can be set to the numerical value within 10, i.e. the time of this calculating process is multiple
Miscellaneous degree is very low, and efficiency is very high.
In the present embodiment, by according to the social networks between user each in social networks, and each target object of acquisition
Between similitude establish figure recommended models, and then calculate in figure recommended models, the centrality of each node, so as to according to each node
Centrality, the corresponding target object of the node is recommended into the corresponding user of the node.The figure recommended models are to be based on
The graph model of Node connectedness relationship, by the instruction of Node connectedness relationship, there are correspond to target pair to be recommended in the user of social networks
Similitude as between combines the recommendation of target object with the social networks between user to realize, and solves existing
The technical problem for having the recommendation effect of target object in technology bad.
On the basis of a upper embodiment, in the formula for calculating PageRank value, priori knowledge is related to.Upper one implements
In example, priori knowledge is not considered, for each node by Vi TValue is 1.In the present embodiment, priori knowledge is introduced, for not
Same node determines V respectivelyi TValue.
It should be noted that priori knowledge is used to indicate user for the preference of each application program.
In the present embodiment, Vi TIt is according to node njPriori knowledge determine, priori knowledge includes node njCorresponding use
The family preference node njThe determination degree of corresponding target object.
Specifically, in the presence of such priori knowledge, the recommendation results obtained according to graph model will also meet this priori
The restriction of knowledge, i.e., if it is known that user urPreference appp, then the node N (u obtained after operationr,appp)
PageRank value also should be larger, to represent to user urRecommend apppA possibility that it is larger.In addition to this, this priori knowledge is also
Influence whether other users, i.e., if user upPay close attention to user ur, due to upPay close attention to ur, it can be said that upHave and urSimilar
Point of interest, it is likely that the APP of preference same type.It so will be with apppU is recommended in similar applicationpA possibility that also should be larger.
The corresponding mathematical sense of this physical meaning can be regarded as:, will also be by between node in figure under the premise of meeting priori knowledge
Interconnection this priori knowledge is handed on.
In the present embodiment, formula can be passed throughIn Vi TBy priori knowledge knot
It closes into the calculating process of PageRank value.
In above formula, Vi TFor the probability that user specifies, which represent when jumping generation at random, to node niWhat is jumped is general
Rate.Under the premise of without any priori knowledge, the probability that jumps of all nodes is generally set as 1.However, if there is priori
Knowledge, then can be to Vi TDifferent values is set, and the height of value represents the application journey to user preference its inter-node in node
The determination degree of sequence.
As a kind of possible implementation, priori knowledge Vi TValue range can for estimation several values.
Such as: priori knowledge Vi TValue range be three values { 5,3,1 }, priori knowledge Vi TIt can be according to node ni's
The determination degree of priori knowledge therefrom selects a value.5 represent very determining priori knowledge, the i.e. corresponding user preference of node
The corresponding application of the node, 3, which represent priori knowledge, certain uncertainty, and 1 represents to the node without any priori knowledge.It is false
If node N (u in Fig. 35, app6) Vi TValue is 5, node N (u4, app2) Vi TValue is 3, the V of remaining nodei TValue is 1, generation
EnterIt can obtain introducing the PageRank value of the node after priori knowledge.
Fig. 6 is that the schematic diagram for the figure recommended models for introducing priori knowledge is integrated to when by this priori knowledge as seen from Figure 6
After in the calculating process of PageRank value, node N (u5, app6) and node N (u4, app2) PageRank value it is larger, this knot
Fruit just coincide with priori knowledge.
Meanwhile all concern node N (u5, app6) and node N (u4, app2) the PageRank value of neighborhood of nodes have
Increased, that is to say, that have side from node N (u5, app6) and node N (u4, app2) issue node, these nodes
PageRank value increased represent introduce priori knowledge after bring influence.
Such as: in Fig. 6, node N (u4,app4) PageRank value increase 1/3 times compared with Fig. 5, represent and introduce priori and know
After knowledge, by app4Recommend u4A possibility that increase 1/3 times.This result meets its physical meaning, because in priori knowledge really
Determine user u5Preference app6, and u4Pay close attention to u5, and app4And app6With correlation, therefore to u4Recommend app4A possibility that
It should increase.
In order to realize above-described embodiment, the present invention also proposes a kind of recommendation apparatus based on social networks.
Fig. 7 is a kind of structural schematic diagram of the recommendation apparatus based on social networks provided in an embodiment of the present invention.
As shown in fig. 7, being somebody's turn to do the recommendation apparatus based on social networks includes: to obtain module 71, establish module 72, computing module
73 and recommending module 74.
Module 71 is obtained, for obtaining the social networks in social networks between each user, and each target object of acquisition
Between similitude.
Specifically, obtain module 71, be specifically used for: according between the content of each target object the degree of correlation and each mesh
The degree of correlation between the evaluation of object is marked, the similitude between each target object is calculated.
Module 72 is established, for establishing figure recommended models according to the social networks and the similitude.
Wherein, the connection relationship of each node is used to indicate in the user there are social networks in figure recommended models, it is corresponding to
Recommend the similitude between target object.As a kind of possible implementation, figure recommended models include each node, the node
User in the corresponding social networks, and user target object to be recommended is corresponded to, phase in the figure recommended models
There are social networks between the corresponding user of two nodes to connect, and there are phases between corresponding target object
Like property.
Computing module 73, for calculating in the figure recommended models, the centrality of each node.
Wherein, the centrality is determined according to the quantity of corresponding node institute's link node in the figure recommended models
's.
Recommending module 74 is recommended the corresponding target object of the node described for the centrality according to each node
The corresponding user of node.
In the present embodiment, by according to the social networks between user each in social networks, and each target object of acquisition
Between similitude establish figure recommended models, and then calculate in figure recommended models, the centrality of each node, so as to according to each node
Centrality, the corresponding target object of the node is recommended into the corresponding user of the node.The figure recommended models are to be based on
The graph model of Node connectedness relationship, by the instruction of Node connectedness relationship, there are correspond to target pair to be recommended in the user of social networks
Similitude as between combines the recommendation of target object with the social networks between user to realize, and solves existing
The technical problem for having the recommendation effect of target object in technology bad.
It should be noted that the aforementioned explanation to the recommended method embodiment based on social networks is also applied for the reality
The recommendation apparatus based on social networks of example is applied, details are not described herein again.
Based on the above embodiment, the embodiment of the invention also provides a kind of the possible of recommendation apparatus based on social networks
Implementation, Fig. 8 is the structural schematic diagram of another recommendation apparatus based on social networks provided in an embodiment of the present invention, upper
On the basis of one embodiment, in the recommendation apparatus based on social networks, social networks include that concern involved in concern behavior is used
Family and it is concerned user.
Based on this, module 72 is established, comprising: generation unit 721 and connection unit 722.
Generation unit 721, for generating each node in the figure recommended models.
Wherein, each node corresponds to user in the social networks and user target object to be recommended.
Connection unit 722 has similitude for being directed in the figure recommended models between corresponding target object, and
There are two nodes of concern behavior between corresponding user, connect described two nodes using side.
Wherein, quilt in described two nodes is directed toward from the corresponding node of user is paid close attention in described two nodes in the direction on side
Pay close attention to the corresponding node of user.
Further, in a kind of possible implementation of the embodiment of the present invention, recommending module 74, comprising: determination unit
741 and recommendation unit 742.
Determination unit 741 is ranked up for the centrality according to each node for the node of the same user of correspondence.
Recommendation unit 742, for according to the sequence, the corresponding target object of the node to be recommended the node pair
The user answered.
Further, recommendation apparatus, further includes: program determining module 75.
Object determining module 75, for determining the target object set comprising each target object;According to each user browsing
Target object generates the subset of the target object set for each user respectively;Using the target object in the subset as
Corresponding user target object to be recommended.
Further, computing module 73, comprising: migration unit 731 and probability calculation unit 732.
Migration unit 731, for the random walk since each of figure recommended models node, random walk road
Diameter passes through node n every timeiWhen, node n is jumped to probability c random selectioniClose on node njOn, alternatively, with probability 1-c with
Machine jumps to any node in the figure recommended models and terminates random walk.
Probability calculation unit 732, for the node n in the figure recommended modelsiPageRank value PR (ni) be
Wherein, c is to jump probability, outjFor node njOut-degree, Vi TIt is to jump to node n when jumping generation at randomi's
Probability.
As a kind of possible implementation, Vi TIt is according to node njPriori knowledge determine;The priori knowledge packet
Include node njNode n described in corresponding user preferencejThe determination degree of corresponding target object.
In the embodiment of the present invention, by according to the social networks between user each in social networks, and each target of acquisition
Similitude between object establishes figure recommended models, and then calculates in figure recommended models, the centrality of each node, so as to according to each
The corresponding target object of the node is recommended the corresponding user of the node by the centrality of node.The figure recommended models are
Based on the graph model of Node connectedness relationship, by the instruction of Node connectedness relationship, there are correspond to mesh to be recommended in the user of social networks
The similitude between object is marked, combines the recommendation of target object with the social networks between user to realize, is solved
The bad technical problem of the recommendation effect of target object in the prior art.
In order to realize above-described embodiment, the present invention also proposes a kind of computer equipment, comprising: processor, and for depositing
Store up the memory of the processor-executable instruction.
Wherein, processor is configured as:
The social networks in social networks between each user are obtained, and obtain the similitude between each target object;Root
According to the social networks and the similitude, figure recommended models are established;Wherein, in the figure recommended models each node connection
Relationship is used to indicate in the user there are social networks, the similitude between corresponding target object to be recommended.The figure is calculated to push away
It recommends in model, the centrality of each node;Wherein, the centrality is connected in the figure recommended models according to corresponding node
What the quantity of node determined;According to the centrality of each node, the corresponding target object of the node is recommended into the node pair
The user answered.
In order to realize above-described embodiment, the present invention also proposes a kind of computer readable storage medium, when the storage medium
In instruction be performed by processor, be able to carry out a kind of recommended method based on social networks, which comprises obtain
Social networks in social networks between each user, and obtain the similitude between each target object;According to the social pass
System and the similitude, establish figure recommended models;Wherein, the connection relationship of each node is used to indicate in the figure recommended models
There are the similitudes in the user of social networks, between corresponding target object to be recommended.It calculates in the figure recommended models, it is each to tie
The centrality of point;Wherein, the centrality is true according to the quantity of corresponding node institute's link node in the figure recommended models
Fixed;According to the centrality of each node, the corresponding target object of the node is recommended into the corresponding user of the node.
In order to clearly illustrate the recommended method based on social networks of previous embodiment offer, one kind is present embodiments provided
Recommender system based on social networks.
It include recommendation server, Resource Server and social networking service in the recommender system based on social networks
Device.
Specifically, Fig. 9 is the structural schematic diagram of the recommender system provided in an embodiment of the present invention based on social networks, is such as schemed
Shown in 9, including recommendation server 91, Resource Server 92 and social network server 93.
Wherein, Resource Server 92, for storing similitude between each target object and user's mesh to be recommended
Mark object.Target object is specifically as follows application program.
Social network server 93, for storing the social networks in social networks between each user.
Recommendation server 91, for obtaining the similitude between each target object from Resource Server 92 respectively, and
The social networks in social networks between each user are obtained from social network server 93.
Further, recommendation server 91 are also used to establish figure according to the social networks and the similitude and recommend
Model;Wherein, the connection relationship of each node is used to indicate in the user there are social networks in the figure recommended models, it is corresponding to
Recommend the similitude between target object.It calculates in the figure recommended models, the centrality of each node;Wherein, the centrality
It is to be determined according to the quantity of corresponding node institute's link node in the figure recommended models;It, will according to the centrality of each node
The corresponding target object of the node recommends the corresponding user of the node.
Such as: recommendation server 91 can carry out the recommendation of target object by social networks to user, in the present embodiment
The mode of recommendation is not construed as limiting.
It should be noted that user described in various embodiments of the present invention refers to user terminal, user account etc..
As a kind of possible application scenarios, after recommendation server 91 generates figure recommended models, if social networking service
Device 93 notifies that the social networks in 91 social networks of recommendation server between each user change or Resource Server 92 is logical
Know that similitude or user between 91 target object of recommendation server target object to be recommended change, recommendation server
91 need to regenerate figure recommended models.
In the present embodiment, the figure recommended models that recommendation server 91 generates are easy to be extended, i.e., if the concern of user
Or after point of interest changes or has new target object to be added, it is easy to which the structure of change figure recommended models changes in turn to be pushed away
Recommend result.
Specifically, usually there are two types of modes for the mode being changed to the structure of figure recommended models:
First way does not increase new node, but changes the structure on side.That is, will not introduce new use
Family or application, but change the concern of user to bring the adjustment on side.For example, if user urNewest concern ut, and appp
And appqWith correlation, then should be in node N (ur,appp) and node N (ut,appq) between establish one by N (ut,
appq) it is directed toward N (ur,appp) side.Similarly, if urNo longer pay close attention to ut, then should cancel side between the two.
The second way increases node.If certain new users or fresh target object are increased in model, such as using
Program will necessarily increase a series of node in figure in this way to show that certain target objects the use newly increased can be recommended
Family, or the target object newly increased is recommended into certain already present users.
The third mode deletes node.If some mesh has been determined according to certain priori knowledges or some other factor
Mark object can remove, such as application program undercarriage, then its corresponding node should be deleted from figure recommended models.
For example, if from user urAPP to be recommended set in eliminate appp, then should remove in figure recommended models
Node N (ur,appp).In fact, no matter occur it is any in above-mentioned several situations, be required to change random walk calculating side
The certain random walk paths formed in method.However, it is to be modified to be not that whole random walk paths is both needed to, i.e., only need
Modify those pass through or may node or side Jing Guo being modified random walk path, these paths needs are given birth to again
At.
Specifically, it may be considered that a kind of simplest situation, i.e., if after deleting a line, need it is to be modified with
The desired value of machine migration number of path.It can be proved that after deleting a line, the desired value of number of path that need to be to be modified isWherein, WtThe number of path for needing to be adjusted is represented, e represents total number of edges.
Next consider other several situations:
1) increase a line;
2) increase a node;
3) node is deleted.
In fact, three cases above may be incorporated in considers together.Assuming that changed node is nk, it is possible to increase
One and nkThe side of connection increases nkInto figure, or n is deleted from figurek。
No matter above-mentioned which kind of situation is occurred, as long as certain paths passes through nkIt will be modified to re-form.Assuming that and nkConnection
The number on side be cu, then known to need the desired value of number of path to be modified are as follows:
In fixed knot points and in the case where number of edges, need number of path to be modified be it is very small, therefore, can prove that figure recommends mould
Type is easily modified.
System provided in an embodiment of the present invention by according to the social networks between user each in social networks, and obtains
It takes the similitude between each target object to establish figure recommended models, and then calculates in figure recommended models, the centrality of each node, with
Just according to the centrality of each node, the corresponding target object of the node is recommended into the corresponding user of the node.The figure pushes away
Recommending model is the graph model based on Node connectedness relationship, and by the instruction of Node connectedness relationship, there are corresponding in the user of social networks
Similitude between target object to be recommended mutually ties the social networks between the recommendation of target object and user to realize
It closes, solves the bad technical problem of the recommendation effect of target object in the prior art.
In the description of this specification, reference term " one embodiment ", " some embodiments ", " example ", " specifically show
The description of example " or " some examples " etc. means specific features, structure, material or spy described in conjunction with this embodiment or example
Point is included at least one embodiment or example of the invention.In the present specification, schematic expression of the above terms are not
It must be directed to identical embodiment or example.Moreover, particular features, structures, materials, or characteristics described can be in office
It can be combined in any suitable manner in one or more embodiment or examples.In addition, without conflicting with each other, the skill of this field
Art personnel can tie the feature of different embodiments or examples described in this specification and different embodiments or examples
It closes and combines.
In addition, term " first ", " second " are used for descriptive purposes only and cannot be understood as indicating or suggesting relative importance
Or implicitly indicate the quantity of indicated technical characteristic.Define " first " as a result, the feature of " second " can be expressed or
Implicitly include at least one this feature.In the description of the present invention, the meaning of " plurality " is at least two, such as two, three
It is a etc., unless otherwise specifically defined.
Any process described otherwise above or method description are construed as in flow chart or herein, and expression includes
It is one or more for realizing custom logic function or process the step of executable instruction code module, segment or portion
Point, and the range of the preferred embodiment of the present invention includes other realization, wherein can not press shown or discussed suitable
Sequence, including according to related function by it is basic simultaneously in the way of or in the opposite order, to execute function, this should be of the invention
Embodiment person of ordinary skill in the field understood.
Expression or logic and/or step described otherwise above herein in flow charts, for example, being considered use
In the order list for the executable instruction for realizing logic function, may be embodied in any computer-readable medium, for
Instruction execution system, device or equipment (such as computer based system, including the system of processor or other can be held from instruction
The instruction fetch of row system, device or equipment and the system executed instruction) it uses, or combine these instruction execution systems, device or set
It is standby and use.For the purpose of this specification, " computer-readable medium ", which can be, any may include, stores, communicates, propagates or pass
Defeated program is for instruction execution system, device or equipment or the dress used in conjunction with these instruction execution systems, device or equipment
It sets.The more specific example (non-exhaustive list) of computer-readable medium include the following: there is the electricity of one or more wirings
Interconnecting piece (electronic device), portable computer diskette box (magnetic device), random access memory (RAM), read-only memory
(ROM), erasable edit read-only storage (EPROM or flash memory), fiber device and portable optic disk is read-only deposits
Reservoir (CDROM).In addition, computer-readable medium can even is that the paper that can print described program on it or other are suitable
Medium, because can then be edited, be interpreted or when necessary with it for example by carrying out optical scanner to paper or other media
His suitable method is handled electronically to obtain described program, is then stored in computer storage.
It should be appreciated that each section of the invention can be realized with hardware, software, firmware or their combination.Above-mentioned
In embodiment, software that multiple steps or method can be executed in memory and by suitable instruction execution system with storage
Or firmware is realized.Such as, if realized with hardware in another embodiment, following skill well known in the art can be used
Any one of art or their combination are realized: have for data-signal is realized the logic gates of logic function from
Logic circuit is dissipated, the specific integrated circuit with suitable combinational logic gate circuit, programmable gate array (PGA), scene can compile
Journey gate array (FPGA) etc..
Those skilled in the art are understood that realize all or part of step that above-described embodiment method carries
It suddenly is that relevant hardware can be instructed to complete by program, the program can store in a kind of computer-readable storage medium
In matter, which when being executed, includes the steps that one or a combination set of embodiment of the method.
It, can also be in addition, each functional unit in each embodiment of the present invention can integrate in a processing module
It is that each unit physically exists alone, can also be integrated in two or more units in a module.Above-mentioned integrated mould
Block both can take the form of hardware realization, can also be realized in the form of software function module.The integrated module is such as
Fruit is realized and when sold or used as an independent product in the form of software function module, also can store in a computer
In read/write memory medium.
Storage medium mentioned above can be read-only memory, disk or CD etc..Although having been shown and retouching above
The embodiment of the present invention is stated, it is to be understood that above-described embodiment is exemplary, and should not be understood as to limit of the invention
System, those skilled in the art can be changed above-described embodiment, modify, replace and become within the scope of the invention
Type.
Claims (14)
1. a kind of recommended method based on social networks, which comprises the following steps:
The social networks in social networks between each user are obtained, and obtain the similitude between each target object;
According to the social networks and the similitude, figure recommended models are established;Wherein, each node in the figure recommended models
Connection relationship be used to indicate in the user there are social networks, the similitude between corresponding target object to be recommended;
It calculates in the figure recommended models, the centrality of each node;Wherein, the centrality is according to corresponding node in the figure
The quantity of institute's link node determines in recommended models;
According to the centrality of each node, the corresponding target object of the node is recommended into the corresponding user of the node.
2. the recommended method according to claim 1 based on social networks, which is characterized in that the social networks include concern row
For related concern user and it is concerned user;It is described to establish figure recommended models, comprising:
Generate each node in the figure recommended models;Wherein, each node corresponds to user in the social networks and described
User's target object to be recommended;
For in the figure recommended models, there is similitude between corresponding target object, and exists between corresponding user and close
Two nodes of note behavior connect described two nodes using side;User is paid close attention to from described two nodes in the direction on the side
Corresponding node is directed toward in described two nodes and is concerned the corresponding node of user.
3. the recommended method according to claim 1 based on social networks, which is characterized in that described to obtain each target object
Between similitude, comprising:
According to the degree of correlation between the evaluation of the degree of correlation and each target object between the content of each target object, calculate
Similitude between each target object.
4. the recommended method according to claim 1 based on social networks, which is characterized in that the collection according to each node
The corresponding target object of the node is recommended the corresponding user of the node by neutrality, comprising:
According to the centrality of each node, it is ranked up for the node of the same user of correspondence;
According to the sequence, the corresponding target object of the node is recommended into the corresponding user of the node.
5. the recommended method according to claim 1 based on social networks, which is characterized in that described according to the social pass
System and the similitude, are established before figure recommended models, further includes:
Determine the target object set comprising each target object;
According to the target object that each user browses, the subset of the target object set is generated respectively for each user;
Using the target object that the target object in the subset is to be recommended as corresponding user.
6. the recommended method according to claim 1-5 based on social networks, which is characterized in that the centrality
Including PageRank value;It is described to calculate in the figure recommended models, the centrality of each node, comprising:
The figure recommended models are calculated using Random Walk Algorithm, obtain the PageRank value of each node.
7. a kind of recommendation apparatus based on social networks, which comprises the following steps:
Module is obtained, for obtaining between the social networks in social networks between each user, and each target object of acquisition
Similitude;
Module is established, for establishing figure recommended models according to the social networks and the similitude;Wherein, the figure pushes away
The connection relationship for recommending each node in model is used to indicate in the user there are social networks, between corresponding target object to be recommended
Similitude;
Computing module, for calculating in the figure recommended models, the centrality of each node;Wherein, the centrality is according to right
The quantity of node institute's link node in the figure recommended models is answered to determine;
The corresponding target object of the node is recommended the node pair for the centrality according to each node by recommending module
The user answered.
8. the recommendation apparatus according to claim 7 based on social networks, which is characterized in that the social networks include concern row
For related concern user and it is concerned user;It is described to establish module, comprising:
Generation unit, for generating each node in the figure recommended models;Wherein, each node corresponds in the social networks
User and user target object to be recommended;
Connection unit has similitude, and corresponding use for being directed in the figure recommended models between corresponding target object
There are two nodes of concern behavior between family, connect described two nodes using side;The direction on the side, from described two knots
It is paid close attention in point in the described two nodes of the corresponding node direction of user and is concerned the corresponding node of user.
9. the recommendation apparatus according to claim 7 based on social networks, which is characterized in that the acquisition module, specifically
For:
According to the degree of correlation between the evaluation of the degree of correlation and each target object between the content of each target object, calculate
Similitude between each target object.
10. the recommendation apparatus according to claim 7 based on social networks, which is characterized in that the recommending module, packet
It includes:
Determination unit is ranked up for the centrality according to each node for the node of the same user of correspondence;
Recommendation unit, for according to the sequence, the corresponding target object of the node to be recommended the corresponding use of the node
Family.
11. the recommendation apparatus according to claim 7 based on social networks, which is characterized in that described device, further includes:
Object determining module, for determining the target object set comprising each target object;The target pair browsed according to each user
As generating the subset of the target object set respectively for each user;Using the target object in the subset as to application
Family target object to be recommended.
12. according to the described in any item recommendation apparatus based on social networks of claim 7-11, which is characterized in that the concentration
Property includes PageRank value;The computing module, is specifically used for:
The figure recommended models are calculated using Random Walk Algorithm, obtain the PageRank value of each node.
13. a kind of computer equipment, which is characterized in that including memory, processor and store on a memory and can handle
The computer program run on device, which is characterized in that when the processor executes described program, realize as in claim 1-6
Any recommended method based on social networks.
14. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the program is by processor
Such as the recommended method as claimed in any one of claims 1 to 6 based on social networks is realized when execution.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710483555.XA CN109118379B (en) | 2017-06-22 | 2017-06-22 | Social network-based recommendation method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710483555.XA CN109118379B (en) | 2017-06-22 | 2017-06-22 | Social network-based recommendation method and device |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109118379A true CN109118379A (en) | 2019-01-01 |
CN109118379B CN109118379B (en) | 2023-03-21 |
Family
ID=64732967
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710483555.XA Active CN109118379B (en) | 2017-06-22 | 2017-06-22 | Social network-based recommendation method and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109118379B (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110032682A (en) * | 2019-04-17 | 2019-07-19 | 腾讯科技(上海)有限公司 | A kind of information recommendation list generation method, device and equipment |
CN112199605A (en) * | 2020-09-09 | 2021-01-08 | 福建天泉教育科技有限公司 | Social relationship recommendation method and terminal |
CN112861015A (en) * | 2019-11-27 | 2021-05-28 | 北京达佳互联信息技术有限公司 | Account associated information acquisition method and device in application program and electronic equipment |
CN113365090A (en) * | 2021-05-31 | 2021-09-07 | 北京百度网讯科技有限公司 | Object recommendation method, object recommendation device, electronic equipment and readable storage medium |
CN113609378A (en) * | 2021-07-02 | 2021-11-05 | 清华大学 | Information recommendation method and device, electronic equipment and storage medium |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20090276453A1 (en) * | 2008-04-30 | 2009-11-05 | Yahoo! Inc. | Modification of brand representations by a brand engine in a social network |
CN104021233A (en) * | 2014-06-30 | 2014-09-03 | 电子科技大学 | Social network friend recommendation method based on community discovery |
CN104899763A (en) * | 2015-05-07 | 2015-09-09 | 西安电子科技大学 | Personalized recommendation method based on bilateral diffusion of bipartite network |
CN105809558A (en) * | 2016-03-15 | 2016-07-27 | 平安科技(深圳)有限公司 | Social network based recommendation method and apparatus |
CN106056455A (en) * | 2016-06-02 | 2016-10-26 | 南京邮电大学 | Group and place recommendation method based on location and social relationship |
CN106126586A (en) * | 2016-06-21 | 2016-11-16 | 安徽师范大学 | A kind of social networks recommended models construction method trusted based on overall merit |
-
2017
- 2017-06-22 CN CN201710483555.XA patent/CN109118379B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20090276453A1 (en) * | 2008-04-30 | 2009-11-05 | Yahoo! Inc. | Modification of brand representations by a brand engine in a social network |
CN104021233A (en) * | 2014-06-30 | 2014-09-03 | 电子科技大学 | Social network friend recommendation method based on community discovery |
CN104899763A (en) * | 2015-05-07 | 2015-09-09 | 西安电子科技大学 | Personalized recommendation method based on bilateral diffusion of bipartite network |
CN105809558A (en) * | 2016-03-15 | 2016-07-27 | 平安科技(深圳)有限公司 | Social network based recommendation method and apparatus |
CN106056455A (en) * | 2016-06-02 | 2016-10-26 | 南京邮电大学 | Group and place recommendation method based on location and social relationship |
CN106126586A (en) * | 2016-06-21 | 2016-11-16 | 安徽师范大学 | A kind of social networks recommended models construction method trusted based on overall merit |
Non-Patent Citations (1)
Title |
---|
刘辉; 郭梦梦; 潘伟强: "个性化推荐***综述" * |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110032682A (en) * | 2019-04-17 | 2019-07-19 | 腾讯科技(上海)有限公司 | A kind of information recommendation list generation method, device and equipment |
CN110032682B (en) * | 2019-04-17 | 2023-08-11 | 腾讯科技(上海)有限公司 | Information recommendation list generation method, device and equipment |
CN112861015A (en) * | 2019-11-27 | 2021-05-28 | 北京达佳互联信息技术有限公司 | Account associated information acquisition method and device in application program and electronic equipment |
CN112199605A (en) * | 2020-09-09 | 2021-01-08 | 福建天泉教育科技有限公司 | Social relationship recommendation method and terminal |
CN113365090A (en) * | 2021-05-31 | 2021-09-07 | 北京百度网讯科技有限公司 | Object recommendation method, object recommendation device, electronic equipment and readable storage medium |
CN113365090B (en) * | 2021-05-31 | 2022-11-18 | 北京百度网讯科技有限公司 | Object recommendation method, object recommendation device, electronic equipment and readable storage medium |
CN113609378A (en) * | 2021-07-02 | 2021-11-05 | 清华大学 | Information recommendation method and device, electronic equipment and storage medium |
CN113609378B (en) * | 2021-07-02 | 2022-11-22 | 清华大学 | Information recommendation method and device, electronic equipment and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN109118379B (en) | 2023-03-21 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109118379A (en) | Recommended method and device based on social networks | |
CN103514204B (en) | Information recommendation method and device | |
CN108898459A (en) | A kind of Method of Commodity Recommendation and device | |
CN109559208A (en) | A kind of information recommendation method, server and computer-readable medium | |
Sumith et al. | Influence maximization in large social networks: Heuristics, models and parameters | |
US20120324004A1 (en) | Systems and methods for analyzing social network user data | |
Chelmis et al. | Social networking analysis: A state of the art and the effect of semantics | |
CN104052714B (en) | The method for pushing and server of multimedia messages | |
CN110287420B (en) | Training method and device of information recommendation system | |
CN105354202B (en) | Data push method and device | |
CN104951465A (en) | Application recommendation method and device | |
Hu et al. | Personalized tag recommendation using social influence | |
CN112258301B (en) | Commodity recommendation method, commodity recommendation device, commodity recommendation system, readable storage medium and electronic equipment | |
Lim et al. | A topological approach for detecting twitter communities with common interests | |
CN115439197A (en) | E-commerce recommendation method and system based on knowledge map deep learning | |
Razghandi et al. | A context-aware and user behavior-based recommender system with regarding social network analysis | |
Chang et al. | Link prediction in a bipartite network using Wikipedia revision information | |
Au Yeung et al. | Capturing implicit user influence in online social sharing | |
Cimini et al. | Enhancing topology adaptation in information-sharing social networks | |
Moosavi et al. | Community detection in online social networks using actions of users | |
Cai et al. | Mechanism design for personalized recommender systems | |
Sziklai et al. | Finding early adopters of innovation in social networks | |
Song et al. | Extraction and visualization of implicit social relations on social networking services | |
Saha et al. | Group proximity measure for recommending groups in online social networks | |
Karimi-Majd et al. | Extracting new ideas from the behavior of social network users |
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 |