CN105095267A - User involving project recommendation method and apparatus - Google Patents
User involving project recommendation method and apparatus Download PDFInfo
- Publication number
- CN105095267A CN105095267A CN201410195669.0A CN201410195669A CN105095267A CN 105095267 A CN105095267 A CN 105095267A CN 201410195669 A CN201410195669 A CN 201410195669A CN 105095267 A CN105095267 A CN 105095267A
- Authority
- CN
- China
- Prior art keywords
- user
- project
- targeted customer
- value
- vector
- 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
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application discloses a user involving project recommendation method. The method comprises: acquiring a project set to be recommended to target users; acquiring a user set formed by users that have been involved in projects in the project set to be recommended to the target users; acquiring an influence value of each user in the user set ; according to projects involved in participation behaviors of the users, acquiring similarity degree between each user of the user set and the target users; considering the influence value of each user in the user set and the similarity degree between each user of the user set and the target users, calculating predicted values of degree of interest the target users have on each project in the project set to be recommended to the target users; and considering the predicted values, selecting at least one project from the project set to be recommended to the target users, and recommending the selected at least one project to the target users. According to the user involving project recommendation method and apparatus, the recommendation accuracy and the quality of recommended projects are improved, and the recommended projects are higher in representativeness and effectiveness.
Description
Technical field
The application relates to technical field of internet application, is specifically related to a kind of recommend method participating in project for user.The application relate in addition a kind of for user participate in the recommendation apparatus of project, the computing method of customer impact force value, customer impact force value calculation element, participate in the recommend method of project for user and participate in the recommendation apparatus of project for user.
Background technology
The appearance of internet and popularize and bring a large amount of information to user, meet the demand of user in the information age to information, but increasing substantially of the network information amount brought along with developing rapidly of network, make user cannot get self useful information in time when in the face of bulk information, not only take a large amount of time, and cause the wasting of resources, the way addressed this problem is commending system, it is according to user to the demand of information and interest etc., interested for user information or product etc. is recommended the system of user.Wherein, collaborative filtering recommending technology obtains larger success in current commending system.
Under prior art, the recommend method of collaborative filtering recommending technology mainly comprises following three kinds: based on the collaborative filtering of user (User-based), based on the collaborative filtering of project (Item-based) and the collaborative filtering based on model (Model-based).
Collaborative filtering based on user (User-based) comprises following three steps:
1) user's information is collected
Collection can the information of representative of consumer interest.General web station system uses the mode of scoring or evaluates, this mode is called as " initiatively scoring ", in addition, a kind of marking mode is also had to be " passive scoring ", be replace user to complete evaluation according to the behavior pattern of user by system, do not need user directly to give a mark or input evaluation data.
2) nearest neighbor search
Calculate the similarity between user and generate the core that " nearest-neighbors " user collection is the collaborative filtering based on user (User-based).Such as: find, with user A, there is n user of certain similarity, using a described n user to the scoring of project M as the score in predicting of user A to project M.Similarity is the similarity degree of characterizing consumer to the project that the level of interest of project or evaluation, the behavior of user and the behavior of user relate to.The algorithm that general meeting is different according to the different choice of data, the similarity algorithm of current more use has cosine similarity algorithm, Pearson (Pearson) related coefficient algorithm and adjustment cosine similarity algorithm.
3) recommendation results is produced
There is " nearest-neighbors " user to collect, just can predict the interest of targeted customer, produced recommendation results.According to recommending the difference of object to carry out multi-form recommendation, more common recommendation results has Top-N to recommend, and Top-N recommendation chooses the highest N number of project of the interest of targeted customer as recommended project.
The recommend method of prior art be similarity between based target user and other users to targeted customer's recommendation information, the items of interest having other users of similarity with targeted customer is recommended targeted customer.But in the middle of reality, for a project, different users for the interest-degree of this project or demand degree not identical, now, in fact commending system cannot play due effect; In this case, the accuracy of recommendation must be caused lower, and the Quality Down of recommended project.
From the angle of system, the information interaction that inappropriate project recommendation will inevitably be unnecessary between adding users, thus cause Website server or instant communication server to increase added burden, cause the waste of Internet resources.
Summary of the invention
The application provides a kind of recommend method participating in project for user, to solve prior art Problems existing.The application relate in addition a kind of for user participate in the recommendation apparatus of project, the computing method of customer impact force value, customer impact force value calculation element, participate in the recommend method of project for user and participate in the recommendation apparatus of project for user.
The application provides a kind of recommend method participating in project for user, comprising:
Obtain to be recommended to the project set of targeted customer;
Obtain and the described user be made up of the user crossing participative behavior to project in the project set of targeted customer to be recommended is gathered;
What obtain each user in this user set affects force value;
Project involved by the participative behavior of user obtains the similarity between each user of described user set and described targeted customer;
In conjunction with the similarity affected in force value and this user set between each user and described targeted customer of each user in described user's set, calculate described targeted customer to described to be recommended to the predicted value of the interest-degree of each project in the project set of targeted customer;
In conjunction with this predicted value choose described to be recommended give targeted customer project set at least one project recommend to described targeted customer;
Wherein, the force value that affects of described user is determined according to the historical operation behavior of user.
Optionally, the described acquisition project set of targeted customer of giving to be recommended comprises:
Obtain the project involved by participative behavior of described targeted customer;
The item destination aggregation (mda)s having the participative behavior of user all under obtaining project kind corresponding to this project, using described item destination aggregation (mda) as the project set giving targeted customer to be recommended.
Optionally, the force value that affects of each user in this user of described acquisition set comprises:
Obtain each user in described user set and be directed to the liveness of each project in described project set, and form the matrix of user and project;
Generate and initialising subscriber vector sum project latent vector, wherein, the element value of described user vector represent user in described user set initial affect force value, the element value of described project latent vector represents the initial value of the attribute of project in described project set;
By energy transmission iterative algorithm, iterative computation is carried out to the matrix of described user and project, user vector and project latent vector, and the element value of user vector after described iterative computation is affected force value as user in described user's set;
Wherein, described liveness characterizes each user in described user set and is directed to the degree of participation of each project in described project set.
Optionally, the liveness that in described acquisition described user set, each user is directed to each project in described project set comprises:
For each user in described user's set, obtain the set that this user is directed to all participative behaviors of each project in described project set, and the behavior weight that in this set, the participation number of times of each participative behavior is corresponding with each participative behavior;
Calculate the participation number of times of each participative behavior in described set and the product of behavior weight, and calculate the weighted value of the participation number of times of all participative behaviors in this set and the product of behavior weight, this weighted value is directed to the liveness of each project in described project set as each user in described user's set.
Optionally, describedly by energy transmission iterative algorithm, iterative computation is carried out to the matrix of described user and project, user vector and project latent vector and comprises:
Calculate the product of the transpose of a matrix matrix of described project latent vector and described user and project, using the value of this product as described user vector, and the acting force this calculating be considered as in project is applied to the primary energy propagation of user force by user behavior;
To calculate the product of the matrix of user vector and described user and the project obtained as project vector, and the influence power this calculating being considered as user is propagated to the primary energy of other users by there being the project of user's participative behavior;
Above-mentioned energy transmission calculating process is repeated according to iterations;
Wherein, described iterations is determined according to the empirical value of setting or is less than according to the difference of the mould of calculated vector after twice adjacent calculation arrange fixed value and determine.
Optionally, the similarity that described project involved by the participative behavior of user obtains in described user set between each user and described targeted customer comprises:
Similarity algorithm is utilized to calculate the similarity between each user and targeted customer of described user set;
Wherein, described similarity algorithm comprises: cosine similarity algorithm, Pearson (Pearson) related coefficient algorithm or adjustment cosine similarity algorithm.
Optionally, in user's set described in described combination each user affect similarity between each user and described targeted customer in force value and this user set, calculate described targeted customer and to be recommendedly to comprise to the predicted value of the interest-degree of each project in the project set of targeted customer described:
Calculate the product of the product affecting the force value parameter corresponding with affecting force value of each user in described user set and the described similarity parameter corresponding with similarity; And calculate the value of the two product addition gained described;
Value based on the two product addition gained described calculates described targeted customer to described to be recommended to the predicted value of the interest-degree of each project in the project set of targeted customer;
Wherein, the project kind that the value of described parameter is corresponding according to project in described project set is determined.
Optionally, described choosing in conjunction with this predicted value describedly to be recommendedly recommends to comprise to described targeted customer at least one project in the project set of targeted customer:
Choose described at least one project the highest to predicted value described in the project set of targeted customer to be recommended to recommend to described targeted customer; Or
Choose the project that predicted value described in described project set is greater than setting value to recommend to described targeted customer.
The application provides a kind of recommendation apparatus participating in project for user in addition, comprising:
Project acquiring unit, obtains to be recommended to the project set of targeted customer;
User's acquiring unit, gathers the described user be made up of the user crossing participative behavior to project in the project set of targeted customer to be recommended for obtaining;
Affect force value acquiring unit, for obtain this user set in each user affect force value;
Similarity acquiring unit, obtains the similarity between each user of described user set and described targeted customer for the project involved by the participative behavior of user;
Predictor calculation unit, for the similarity affected in force value and this user set between each user and described targeted customer in conjunction with each user in described user's set, calculate described targeted customer to described to be recommended to the predicted value of the interest-degree of each project in the project set of targeted customer;
Recommendation unit, for choose in conjunction with this predicted value described to be recommended give targeted customer project set at least one project recommend to described targeted customer.
The application also provides a kind of computing method of customer impact force value, comprising:
Obtain to project by user's set of crossing the user of participative behavior and form, and had the project set of item design of user's participative behavior;
Each user obtaining described user set is directed to the liveness of each project in described project set, and forms the matrix of user and project;
Generate and initialising subscriber vector and project latent vector, wherein, the element value of described user vector represent user in described user set initial affect force value, the element value of described project latent vector represents the initial value of the attribute of project in described project set;
By energy transmission iterative algorithm, iterative computation is carried out to described matrix, user vector and project latent vector, and the element value of the user vector after described iterative computation is affected force value as user in described user set;
Wherein, described liveness characterizing consumer is directed to the degree of participation of project, and the force value that affects of described user is determined according to the historical operation behavior of user.
Optionally, the liveness that each user that the described user of described acquisition gathers is directed to each project in described project set comprises:
For each user in user's set, obtain the set that user is directed to all participative behaviors of each project in described project set, and the behavior weight that in the set of this participative behavior, the participation number of times of each participative behavior is corresponding with each participative behavior;
Calculate the participation number of times of each participative behavior in described set and the product of behavior weight, and calculate the weighted value of the participation number of times of all participative behaviors in this set and the product of behavior weight, this weighted value is directed to the liveness of each project in described project set as each user in described user's set.
Optionally, describedly by energy transmission iterative algorithm, iterative computation is carried out to described matrix, user vector and project latent vector and comprises:
Calculate the product of the transpose of a matrix matrix of described project latent vector and described user and project, using the value of this product as described user vector, and the acting force this calculating be considered as in project is applied to the primary energy propagation of user force by user behavior;
To calculate the product of the matrix of user vector and described user and the project obtained as project vector, and the influence power this calculating being considered as user is propagated to the primary energy of other users by there being the project of user's participative behavior;
Above-mentioned energy transmission calculating process is repeated according to iterations;
Wherein, described iterations is determined according to the empirical value of setting or is less than according to the difference of the mould of calculated vector after twice adjacent calculation arrange fixed value and determine.
The application also provides a kind of calculation element of customer impact force value, comprising:
Data capture unit, for obtaining to project by user's set of crossing the user of participative behavior and form, and had the project set of item design of user's participative behavior;
Liveness generation unit, each user for obtaining described user set is directed to the liveness of each project in described project set, and forms the matrix of user and project, and wherein, described liveness characterizing consumer is directed to the degree of participation of project;
Vector initialising unit, for generating and initialising subscriber vector and project latent vector, wherein, the element value of described user vector represent user in described user set initial affect force value, the element value of described project latent vector represents the initial value of project in described project set;
Affect force value computing unit, for carrying out iterative computation by energy transmission iterative algorithm to described matrix, user vector and project latent vector, and the element value of the user vector after described iterative computation is affected force value as user in described user set, wherein, the degree that influences each other affected in force value sign project activity between user of described user.
The application also provides a kind of recommend method participating in project for user, comprising:
User's set that the project in the project set related to the participative behavior of targeted customer that obtains be made up of the user crossing participative behavior;
What obtain each user in this user set affects force value;
Project involved by the participative behavior of user obtains the similarity between each user of described user set and described targeted customer;
In conjunction with the similarity affected in force value and described user set between each user and targeted customer of each user in described user's set, the influence power similarity during calculating targeted customer and described user gather between each user;
The project set that the participative behavior obtaining at least one the user that influence power similarity is the highest in described user set relates to, and using this project set as the project set giving targeted customer to be recommended;
Calculate targeted customer to described to be recommended to the predicted value of interest-degree of each project in the project set of targeted customer, and choose described to be recommendedly to recommend to targeted customer at least one project in the project set of targeted customer in conjunction with this predicted value;
Wherein, the force value that affects of described user is determined according to the historical operation behavior of user; Described influence power similarity characterization is in the similarity affected under the participation of force value between user of user.
Optionally, the force value that affects of each user in this user of described acquisition set comprises:
The project set of the item design that the participative behavior obtaining user in described user set relates to;
Each user obtaining described user set is directed to the liveness of each project in described project set, and forms the matrix of user and project;
Generate and initialising subscriber vector and project latent vector, wherein, the element value of described user vector represent user in described user set initial affect force value, the element value of described project latent vector represents the initial value of the attribute of project in described project set;
By energy transmission iterative algorithm, iterative computation is carried out to described matrix, user vector and project latent vector, and the element value of user vector after described iterative computation is affected force value as user in described user set;
Wherein, described liveness characterizes each user in described user set and is directed to the degree of participation of each project in described project set.
Optionally, the liveness of each project that each user that the described user of described acquisition gathers is directed in described project set comprises:
For each user in described user's set, obtain the set that this user is directed to all participative behaviors of each project in described project set, and the behavior weight that in the set of this participative behavior, the participation number of times of each participative behavior is corresponding with each participative behavior;
Calculate the participation number of times of each participative behavior in the set of described participative behavior and the product of behavior weight, and calculate the weighted value of the participation number of times of all participative behaviors in this set and the product of behavior weight, this weighted value is directed to the liveness of each project in described project set as each user in described user's set.
Optionally, describedly by energy transmission iterative algorithm, iterative computation is carried out to described matrix, user vector and project latent vector and comprises:
Calculate the product of the transpose of a matrix matrix of described project latent vector and described user and project, using the value of this product as described user vector, and the acting force this calculating be considered as in project is applied to the primary energy propagation of user force by user behavior;
To calculate the product of the matrix of user vector and described user and the project obtained as project vector, and the influence power this calculating being considered as user is propagated to the primary energy of other users by there being the project of user's participative behavior;
Above-mentioned energy transmission calculating process is repeated according to iterations;
Wherein, described iterations is determined according to the empirical value of setting or is less than according to the difference of the mould of calculated vector after twice adjacent calculation arrange fixed value and determine.
Optionally, the similarity that described project involved by the participative behavior of user obtains between each user of described user set and described targeted customer comprises:
Similarity algorithm is utilized to calculate the similarity between each user and targeted customer of described user set;
Wherein, described similarity algorithm comprises: cosine similarity algorithm, Pearson (Pearson) related coefficient algorithm or adjustment cosine similarity algorithm.
Optionally, the similarity affected in force value and described user set between each user and targeted customer of each user in user's set described in described combination, the influence power similarity during calculating targeted customer and described user gather between each user comprises:
Calculate the product affecting the force value parameter corresponding with affecting force value of each user in described user set, and the product of the described similarity parameter corresponding with similarity;
Using the value of the two product addition gained described as the influence power similarity between each user during described targeted customer and described user gather;
Wherein, the project kind that the value of described parameter is corresponding according to project in described project set is determined.
Optionally, described in conjunction with this predicted value choose described to be recommended give targeted customer project set at least one project to targeted customer recommend comprise:
Choose described at least one project the highest to predicted value described in the project set of targeted customer to be recommended to recommend to described targeted customer; Or
Choose the project that predicted value described in described project set is greater than setting value to recommend to described targeted customer.
The application also provides a kind of recommendation apparatus participating in project for user, comprising:
User's acquiring unit, for obtaining user's set that in the project set that relates to the participative behavior of targeted customer, project be made up of the user crossing participative behavior;
Affect force value acquiring unit, for obtain this user set in each user affect force value;
Similarity acquiring unit, obtains the similarity between each user of described user set and described targeted customer for the project involved by the participative behavior of user;
Influence power similarity calculated, for the similarity affected in force value and described user set between each user and targeted customer in conjunction with each user in described user's set, the influence power similarity during calculating targeted customer and described user gather between each user;
Project acquiring unit to be recommended, the project set that the participative behavior for obtaining at least one the user that influence power similarity is the highest in described user set relates to, and using this project set as the project set giving targeted customer to be recommended;
Recommendation unit, for calculating targeted customer to described to be recommended to the predicted value of interest-degree of each project in the project set of targeted customer, and choose described to be recommendedly to recommend to targeted customer at least one project in the project set of targeted customer in conjunction with this predicted value.
Compared with prior art, the application has the following advantages:
One of them aspect of the application provides a kind of recommend method participating in project for user, and the method comprises: obtain to be recommended to the project set of targeted customer; Obtain and the described user be made up of the user crossing participative behavior to project in the project set of targeted customer to be recommended is gathered; What obtain each user in this user set affects force value; Project involved by the participative behavior of user obtains the similarity between each user of described user set and described targeted customer; In conjunction with the similarity affected in force value and this user set between each user and described targeted customer of each user in described user's set, calculate described targeted customer to described to be recommended to the predicted value of the interest-degree of each project in the project set of targeted customer; In conjunction with this predicted value choose described to be recommended give targeted customer project set at least one project recommend to described targeted customer; Wherein, the degree that influences each other affected in force value sign project activity between user of described user.
The described recommend method participating in project for user determines project set to be recommended according to the participative behavior of targeted customer, calculate and to be recommendedly affect force value to project in the project set of targeted customer by what cross all users in user's set that the user of participative behavior form to described, targeted customer described in force value and the Similarity Measure between targeted customer and other users is affected to described to be recommended to the predicted value of the interest-degree of each project in the project set of targeted customer based on this, and in conjunction with this predicted value choose described to be recommended give targeted customer project set at least one project recommend to described targeted customer, fully take into account the relation that influences each other between user and project, thus make the accuracy of project recommendation higher, and the quality that improve to targeted customer's recommended project, the project recommending targeted customer is made to have more representativeness and validity, in addition, due to higher recommendation accuracy and validity, the problem of network resources waste can not be caused.
Accompanying drawing explanation
Fig. 1 is a kind of recommend method processing flow chart participating in project for user that the application provides;
Fig. 2 is a kind of recommendation apparatus schematic diagram participating in project for user that the application provides;
Fig. 3 is the computing method processing flow chart of a kind of customer impact force value that the application provides;
Fig. 4 is the calculation element schematic diagram of a kind of customer impact force value that the application provides;
Fig. 5 is a kind of recommend method processing flow chart participating in project for user that the application provides;
Fig. 6 is a kind of recommendation apparatus schematic diagram participating in project for user that the application provides.
Embodiment
Set forth a lot of detail in the following description so that fully understand the application.But the application can be much different from alternate manner described here to implement, those skilled in the art can when doing similar popularization without prejudice to when the application's intension, and therefore the application is by the restriction of following public concrete enforcement.
The application provides a kind of recommend method participating in project for user, to solve the problems referred to above that prior art exists.The application relate in addition a kind of for user participate in the recommendation apparatus of project, the computing method of customer impact force value, customer impact force value calculation element, participate in the recommend method of project for user and participate in the recommendation apparatus of project for user.
Be described in detail below in conjunction with the recommend method participating in project for user of drawings and Examples to the application.
Please refer to Fig. 1, it illustrates a kind of recommend method processing flow chart participating in project for user that the present embodiment provides.
Step S101, obtains to be recommended to the project set of targeted customer.
Described project to refer in internet for customer consumption, participates in and carry out the mutual data message of behavior, such as commodity, advertisement and virtual information etc.
In the present embodiment, be described for the commodity in e-commerce website; Data message and the commodity of other type in described project are similar, and provide the recommend method of following commodity with reference to the present embodiment, the present embodiment is not being listed one by one.
Obtain and to be recommendedly to comprise the following steps to the project set of targeted customer:
1) commodity that the participative behavior obtaining described targeted customer relates to;
Described participative behavior comprises user in e-commerce website shopping process, corresponding to the behaviors such as browsing, click, collect, add shopping cart, pay and/or share of commodity.
Before the commodity that the participative behavior obtaining described targeted customer relates to, carry out the preliminary work of data, that is: obtain the data corresponding to behavior such as user browses, clicks, collects, adds shopping cart, pays and shares in e-commerce website shopping process.In the present embodiment, the related data that user produces by e-commerce website in shopping process is stored in database.In addition, server meeting periodic (such as every day) is by this data importing cloud computing platform (such as Hadoop).
After completing above-mentioned Data Preparation, the commodity that the participative behavior obtaining described targeted customer relates to.It should be noted that, described commodity refer to the commodity collection comprising one or more commodity.
2) the item destination aggregation (mda)s having the participative behavior of user all under obtaining project kind corresponding to this project, using described item destination aggregation (mda) as the project set giving targeted customer to be recommended;
According to above-mentioned steps 1) after the commodity that relate to of the participative behavior that gets targeted customer, for described commodity:
Determine the type of merchandize (such as electronic product, food, books, phonotapes and videotapes etc.) that each part commodity is corresponding;
The set having the commodity of user's participative behavior all under obtaining this type of merchandize, and using the set of these commodity as commodity set of giving targeted customer to be recommended.
Step S102, obtains and gathers the described user be made up of the user crossing participative behavior to project in the project set of targeted customer to be recommended.
Step S101 determines to be recommended to the commodity set of targeted customer, for each the part commodity in this commodity set, obtains the user these commodity being had to participative behavior; Similar, obtaining all commodity in described commodity set had the user of participative behavior to gather.
Step S103, what obtain each user in this user set affects force value.
The degree that influences each other affected in force value sign project activity between user of described user.In the present embodiment, in order to react the mutual potential impact between the behavior otherness of user between different user type and user, introduce the concept affecting force value of user, and user affected force value as the otherness criterion between different user.
Concrete, user is in e-commerce website shopping process, user affects force value and can react the influence power of user in the middle of e-commerce website, and Shopping Behaviors between different user (such as senior user and little Bai user) also difference to some extent; The force value that affects of described user lies in the middle of the historical behavior of user's shopping, to a certain degree can reflect the human-subject test of user to certain class commodity or certain field, in collaborative filtering recommending, affect the higher user of force value usually represent this user to the commodity of certain type or certain field understanding darker, and affect the higher user of force value and can produce certain impact for the shopping of domestic consumer, being usually expressed as affects the higher user of force value and can provide guidance to a certain degree for the shopping of domestic consumer.
The described computation process affecting force value is as follows:
1) obtain each user in described user set and be directed to the liveness of each project in described project set, and form the matrix of user and project;
Described liveness characterizes each user in described user set and is directed to the degree of participation of each project in described project set; Liveness can show as the scoring that user is directed to a certain commodity, also can be the participative behavior that user is directed to these commodity;
Such as: user repeatedly buys a certain commodity or very high to the grading of these commodity, then think that user is directed to these commodity and has higher liveness; Otherwise.
Further, user is directed to having of the participative behavior of a certain commodity a variety of (such as browse, click, collect, add shopping cart, pay and/or share), in order to embody the otherness between the different participative behavior of user, a value is given, that is: behavior weight to each participative behavior of user;
Such as:
The behavior weight that the navigation patterns of user is corresponding is 1;
The behavior weight that the click behavior of user is corresponding is 2;
The behavior weight that the collection behavior of user is corresponding is 5;
User to add behavior weight corresponding to shopping cart behavior be 10;
The behavior weight that the payment behavior of user is corresponding is 20;
The behavior weight that the splitting glass opaque of user is corresponding is 8.
In addition, in order to characterizing consumer is directed to the number of times of the participation that same participative behavior is different between different commodity, introduce and participate in number of times;
Participate in the number of times that number of times characterizing consumer is directed to the participation of the same participative behavior of a certain commodity.
The liveness that in described user set, each user is directed to each commodity in described commodity set calculates and comprises following two steps:
A, for each user in described user set, obtain the set that this user is directed to all participative behaviors of each commodity in described commodity set, and the participation number of times of each participative behavior and behavior weight corresponding to each participative behavior in this set;
B, calculate the participation number of times of each participative behavior in described set and the product of behavior weight, and calculate the weighted value of the participation number of times of all participative behaviors in this set and the product of behavior weight, this weighted value is directed to the liveness of each commodity in described commodity set as each user in described user's set.
The participative behavior X of user
krepresent, this participative behavior X
kcorresponding time participation number T
krepresent, this participative behavior X
kcorresponding behavior weight W
krepresent, user i is directed to the liveness A of commodity j
ijrepresent;
Such as:
X
1expression is browsed, X
2represent click, X
3represent collection, X
4represent and add shopping cart, X
5represent and pay and X
6expression is shared;
Then W
1to W
6value be followed successively by: 1,2,5,10,20 and 8;
For T
k;
If user is not directed to the participative behavior of these commodity, then T
k=0;
If user has the participative behavior being directed to these commodity, then T
kvalue equal the value of participation number of times corresponding to this participative behavior;
Then user i is directed to the liveness of commodity j and is: ∑ X
k* T
k* W
k, wherein, k ∈ (1, K).
To complete in above-mentioned acquisition user set after each user is directed to the liveness step of each commodity in described commodity set, form the matrix of user and commodity;
All users in described user's set are formed a line one by one according to order; All commodity in described commodity set are in line one by one according to order; Form the matrix of user and commodity, and the liveness each user in described user's set being directed to each commodity in described commodity set corresponds to described matrix, as following table:
Item1 | Item2 | Item3 | Item4 | |
User1 | 0.063 | 0.045 | … | … |
User2 | … | 0.1 | ||
User3 | … | 0.6 |
By this matrix A
m × Nrepresent; This entry of a matrix element (that is: in described user's set, each user is directed to the liveness of each commodity in described commodity set) is used A
ijrepresent;
Wherein, M represents the number of users in described user set, and N represents the commodity number in described commodity set; I representative of consumer i, j represent commodity j;
Calculate all users in described user set and be directed to the liveness A of each commodity in described commodity set
ij, and by all A
ijafter value inserts table, just obtain complete user and the matrix A of commodity
m × N.
In addition, arranging of the behavior weight that the participative behavior of user is corresponding need follow following 2 points:
Between the participative behavior of a, user and corresponding behavior weight, there is following relation: user be directed to that the participative behavior of commodity occurs more early, then the behavior weight that this participative behavior is corresponding is higher;
B, for commodity, the participative behavior of user is not unalterable with corresponding behavior weight;
Such as: user is in e-commerce website shopping process, for commodity, after the sales volume of these commodity is greater than default sales volume threshold value, user can see following behavior to the buying behavior of these commodity, now, the buying behavior be greater than after default sales volume threshold value of the sales volume of described commodity is made and falls power process, that is: reduce the participative behavior of user and the value of corresponding behavior weight.
2) also initialising subscriber vector is generated, and commodity latent vector;
The element value of described user vector represent user in described user set initial affect force value, the element value of described project latent vector represents the initial value of the attribute of project in described project set.
User vector E represents, commodity latent vector Q represents;
Definition user vector E=(1,1,1 ..., 1) and ∈ R
m, commodity latent vector Q=(1,1,1 ... 1) ∈ R
n; And
︱E︱=M,︱Q︱=N;
Wherein, M represents the matrix A of described user and commodity
m × Nin number of users, N represents the matrix A of described user and commodity
m × Nin commodity number;
Initialising subscriber vector E:
E=(1,1,1,…,1);
Define the initial force value that affects of each user in described user set and be 1, represent that the initial effects force value of all users is all equal, that is: all users are all in same level to the degree of awareness of commodity and the active value that is directed to these commodity;
Initialization commodity latent vector Q:
Q=(1,1,1,…,1);
The initial value defining the attribute of all commodity in described commodity set is 1, represents that the initial attribute of all commodity is identical, is in same level.
3) by energy transmission iterative algorithm, iterative computation is carried out to the matrix of described user and project, user vector and project latent vector, and the element value of user vector after described iterative computation is affected force value as user in described user's set.
In the present embodiment, described energy transmission iterative algorithm is as follows:
①E=Q*A
M×N T;
②Q=E*A
M×N;
Wherein, E is user vector E, Q is commodity latent vector.
In 1. formula, by the matrix A of described commodity latent vector Q and described user and commodity
m × Ntransposed matrix A
m × N tbe multiplied, represent the intrinsic attribute (that is: the acting force of commodity) of each commodity in described commodity set be applied to described user set in each user affect in force value, this process is considered as the acting force on commodity by the participative behavior of user be applied to user force primary energy propagate;
In 2. formula, by the matrix A of described user vector E and described user and commodity
m × Nbe multiplied, represent that the force value that affects of all users in described user set is applied on other users by the commodity of the participative behavior having user, the behavior of other users in the middle of electronic business transaction website is had an impact; And after this step completes, in the commodity set that commodity latent vector Q represents, the attribute of commodity is no longer intrinsic attribute, commodity latent vector Q changes commodity vector Q into.
In addition, definition iterations k, and repeat above-mentioned energy transmission iterative algorithm according to iterations k.Described iterations k based on experience value (usually, the rational value of iterations t 4 ?between 10) to determine or by the value of the multiplicity iterations k of described energy transmission iterative algorithm during described energy transmission iterative algorithm convergence.The referring to that user vector E or commodity vector Q is less than in the difference of adjacent twice iterative computation rear mold fixed value be set of described energy transmission iterative algorithm convergence.
After described energy transmission iterative algorithm completes iterative computation according to described iterations k, using the set that affect force value of user vector E now as each user in described user's set.
Above-mentioned user and the matrix A of commodity
m × Nother method outside implementation method described in this step can be adopted to realize with energy transmission iterative algorithm, in this no limit.
Step S104, the project involved by the participative behavior of user obtains the similarity between each user of described user set and described targeted customer.
Similarity between each user of described user's set and described targeted customer utilizes similarity algorithm to calculate and obtains; In the present embodiment, the similarity adopting cosine similarity algorithm to calculate between each user of described user set and described targeted customer utilizes similarity, and described cosine similarity algorithm is as follows:
B
uvrepresent the similarity between user v and targeted customer u in described user set, r
uirepresent that targeted customer u is directed to the liveness of commodity i in described commodity set, I
uvrepresent that in user's set, user v and targeted customer u had the commodity set of common participative behavior, I
urepresent that targeted customer u had the commodity set of participative behavior, I
vrepresent that in user's set, user v had the commodity set of participative behavior.
In addition, the computation process that the algorithm realization beyond the cosine similarity algorithm described in the present embodiment is above-mentioned can also be adopted, such as: Pearson (Pearson) related coefficient algorithm and adjustment cosine similarity algorithm etc.
Represent step S105, in conjunction with the similarity affected in force value and this user set between each user and described targeted customer of each user in described user's set, calculate described targeted customer to described to be recommended to the predicted value of the interest-degree of each project in the project set of targeted customer.
1) product of the product affecting the force value parameter corresponding with affecting force value of each user in described user set and the described similarity parameter corresponding with similarity is calculated; And calculate the value of the two product addition gained described;
The set affecting force value of all users in described user set is represented, I with I
vrepresent described user set in user v affect force value; α represent described user set in user v affect parameter corresponding to force value, β represents the parameter that described similarity is corresponding, w
uvrepresent the value of the product addition gained of the product affecting the force value weight parameter corresponding with affecting force value of user v in described user set and the described similarity weight parameter corresponding with similarity, then:
w
uv=αI
v+βB
uv
Wherein, the project kind that the value of described parameter alpha and β is corresponding according to project in described project set is determined.
2) value based on the two product addition gained described calculates described targeted customer to described to be recommended to the predicted value of the interest-degree of each commodity in the commodity set of targeted customer; Described calculating targeted customer u is as follows to the predicted value algorithm of the interest-degree of commodity i in described commodity set:
Wherein, p
uirepresent the predicted value of targeted customer u to the interest-degree of commodity i in described commodity set; N
iu () expression is directed to the set that commodity i has the user of participative behavior.
Step S106, in conjunction with this predicted value choose described to be recommended give targeted customer project set at least one project recommend to described targeted customer.
Calculating the described targeted customer of gained according to above-mentioned steps S105 is directed to described to be recommended to after the predicted value of the interest-degree of each commodity in the commodity set of targeted customer, chooses described to be recommendedly to recommend to described targeted customer at least one project in the middle of the commodity set of targeted customer.
In the present embodiment, employing topN algorithm chooses commodity from described to be recommended giving the commodity set of targeted customer, that is: choose described N (N>=1) the part commodity the highest to predicted value in the commodity set of targeted customer to be recommended and recommend to targeted customer.
In addition, the commodity that predicted value in described commodity set is greater than setting value can also be chosen and recommend to described targeted customer, in this no limit.
The described recommendation apparatus embodiment for participating in project is as follows:
With reference to Fig. 2, it illustrates the recommendation apparatus schematic diagram for participating in project that the present embodiment provides.
Because device embodiment is substantially similar to embodiment of the method, so describe fairly simple, the correspondence that relevant part refers to embodiment of the method illustrates.The device embodiment of following description is only schematic.
The recommendation apparatus for participating in project described in the present embodiment, comprising:
Project acquiring unit 201, obtains to be recommended to the project set of targeted customer;
User's acquiring unit 202, gathers the described user be made up of the user crossing participative behavior to project in the project set of targeted customer to be recommended for obtaining;
Affect force value acquiring unit 203, for obtain this user set in each user affect force value;
Similarity acquiring unit 204, obtains the similarity between each user of described user set and described targeted customer for the project involved by the participative behavior of user;
Predictor calculation unit 205, for the similarity affected in force value and this user set between each user and described targeted customer in conjunction with each user in described user's set, calculate described targeted customer to described to be recommended to the predicted value of the interest-degree of each project in the project set of targeted customer;
Recommendation unit 206, for choose in conjunction with this predicted value described to be recommended give targeted customer project set at least one project recommend to described targeted customer.
The computing method embodiment of described customer impact force value is as follows:
With reference to Fig. 3, it illustrates the computing method processing flow chart of the customer impact force value that the present embodiment provides.
The embodiment of the computing method of the customer impact force value provided due to the present embodiment and the above-mentioned embodiment basic simlarity participating in the recommend method of project for user, so description is fairly simple, relevant part refers to the above-mentioned embodiment correspondence participating in the recommend method of project for user and illustrates.The embodiment of following description is only schematic.
The computing method of the customer impact force value described in the present embodiment, comprising:
Step S301; Obtain to project by user's set of crossing the user of participative behavior and form, and had the project set of item design of user's participative behavior;
Step 302; Each user obtaining described user set is directed to the liveness of each project in described project set, and forms the matrix of user and project;
Step S303; Generate and initialising subscriber vector and project latent vector, wherein, the element value of described user vector represent user in described user set initial affect force value, the element value of described project latent vector represents the initial value of the attribute of project in described project set;
Step S304; By energy transmission iterative algorithm, iterative computation is carried out to described matrix, user vector and project latent vector, and the element value of the user vector after described iterative computation is affected force value as user in described user set;
Wherein, described liveness characterizing consumer is directed to the degree of participation of project, the degree that influences each other affected in force value sign project activity between user of described user.
Optionally, the liveness that each user that the described user of described acquisition gathers is directed to each project in described project set comprises:
For each user in user's set, obtain the set that user is directed to all participative behaviors of each project in described project set, and the behavior weight that in the set of this participative behavior, the participation number of times of each participative behavior is corresponding with each participative behavior;
Calculate the participation number of times of each participative behavior in described set and the product of behavior weight, and calculate the weighted value of the participation number of times of all participative behaviors in this set and the product of behavior weight, this weighted value is directed to the liveness of each project in described project set as each user in described user's set.
Optionally, describedly by energy transmission iterative algorithm, iterative computation is carried out to described matrix, user vector and project latent vector and comprises:
Calculate the product of the transpose of a matrix matrix of described project latent vector and described user and project, using the value of this product as described user vector, and the acting force this calculating be considered as in project is applied to the primary energy propagation of user force by user behavior;
To calculate the product of the matrix of user vector and described user and the project obtained as project vector, and the influence power this calculating being considered as user is propagated to the primary energy of other users by there being the project of user's participative behavior;
Above-mentioned energy transmission calculating process is repeated according to iterations;
Wherein, described iterations is determined according to the empirical value of setting or is less than according to the difference of the mould of calculated vector after twice adjacent calculation arrange fixed value and determine.
The calculation element embodiment of described customer impact force value is as follows:
With reference to Fig. 4, it illustrates the calculation element schematic diagram of the customer impact force value that the present embodiment provides.
Because device embodiment is substantially similar to embodiment of the method, so describe fairly simple, the correspondence that relevant part refers to embodiment of the method illustrates.The device embodiment of following description is only schematic.
The calculation element of the customer impact force value described in the present embodiment, comprising:
Data capture unit 401, for obtaining to project by user's set of crossing the user of participative behavior and form, and had the project set of item design of user's participative behavior;
Liveness generation unit 402, each user for obtaining described user set is directed to the liveness of each project in described project set, and forms the matrix of user and project, and wherein, described liveness characterizing consumer is directed to the degree of participation of project;
Vector initialising unit 403, for generating and initialising subscriber vector and project latent vector, wherein, the element value of described user vector represent user in described user set initial affect force value, the element value of described project latent vector represents the initial value of project in described project set;
Affect force value computing unit 404, for carrying out iterative computation by energy transmission iterative algorithm to described matrix, user vector and project latent vector, and the element value of the user vector after described iterative computation is affected force value as user in described user set, wherein, the degree that influences each other affected in force value sign project activity between user of described user.
It is described that to participate in the recommend method embodiment of project for user as follows:
With reference to Fig. 5, it illustrates the recommend method processing flow chart of the project that participates in for user that the present embodiment provides.
Due to the present embodiment and the above-mentioned embodiment basic simlarity participating in the recommend method of project for user, so description is fairly simple, relevant part refers to the above-mentioned embodiment correspondence participating in the recommend method of project for user and illustrates.The embodiment of following description is only schematic.
The recommend method of the project that participates in for user described in the present embodiment, comprising:
Step S501; User's set that the project in the project set related to the participative behavior of targeted customer that obtains be made up of the user crossing participative behavior;
Step S502; What obtain each user in this user set affects force value;
Step S503; Project involved by the participative behavior of user obtains the similarity between each user of described user set and described targeted customer;
Step S504; In conjunction with the similarity affected in force value and described user set between each user and targeted customer of each user in described user's set, the influence power similarity during calculating targeted customer and described user gather between each user;
Step S505; The project set that the participative behavior obtaining at least one the user that influence power similarity is the highest in described user set relates to, and using this project set as the project set giving targeted customer to be recommended;
Step S506; Calculate targeted customer to described to be recommended to the predicted value of interest-degree of each project in the project set of targeted customer, and choose described to be recommendedly to recommend to targeted customer at least one project in the project set of targeted customer in conjunction with this predicted value;
Wherein, described user affect force value characterizing consumer influence degree to other people in project activity; Described influence power similarity characterization is in the similarity affected under the participation of force value between user of user.
Optionally, the force value that affects of each user in this user of described acquisition set comprises:
The project set of the item design that the participative behavior obtaining user in described user set relates to;
Each user obtaining described user set is directed to the liveness of each project in described project set, and forms the matrix of user and project;
Generate and initialising subscriber vector and project latent vector, wherein, the element value of described user vector represent user in described user set initial affect force value, the element value of described project latent vector represents the initial value of the attribute of project in described project set;
By energy transmission iterative algorithm, iterative computation is carried out to described matrix, user vector and project latent vector, and the element value of user vector after described iterative computation is affected force value as user in described user set;
Wherein, described liveness characterizes each user in described user set and is directed to the degree of participation of each project in described project set.
Optionally, the liveness of each project that each user that the described user of described acquisition gathers is directed in described project set comprises:
For each user in described user's set, obtain the set that this user is directed to all participative behaviors of each project in described project set, and the behavior weight that in the set of this participative behavior, the participation number of times of each participative behavior is corresponding with each participative behavior;
Calculate the participation number of times of each participative behavior in the set of described participative behavior and the product of behavior weight, and calculate the weighted value of the participation number of times of all participative behaviors in this set and the product of behavior weight, this weighted value is directed to the liveness of each project in described project set as each user in described user's set.
Optionally, describedly by energy transmission iterative algorithm, iterative computation is carried out to described matrix, user vector and project latent vector and comprises:
Calculate the product of the transpose of a matrix matrix of described project latent vector and described user and project, using the value of this product as described user vector, and the acting force this calculating be considered as in project is applied to the primary energy propagation of user force by user behavior;
To calculate the product of the matrix of user vector and described user and the project obtained as project vector, and the influence power this calculating being considered as user is propagated to the primary energy of other users by there being the project of user's participative behavior;
Above-mentioned energy transmission calculating process is repeated according to iterations;
Wherein, described iterations is determined according to the empirical value of setting or is less than according to the difference of the mould of calculated vector after twice adjacent calculation arrange fixed value and determine.
Optionally, the similarity that described project involved by the participative behavior of user obtains between each user of described user set and described targeted customer comprises:
Similarity algorithm is utilized to calculate the similarity between each user and targeted customer of described user set;
Wherein, described similarity algorithm comprises: cosine similarity algorithm, Pearson (Pearson) related coefficient algorithm or adjustment cosine similarity algorithm.
Optionally, the similarity affected in force value and described user set between each user and targeted customer of each user in user's set described in described combination, the influence power similarity during calculating targeted customer and described user gather between each user comprises:
Calculate the product affecting the force value parameter corresponding with affecting force value of each user in described user set, and the product of the described similarity parameter corresponding with similarity;
Using the value of the two product addition gained described as the influence power similarity between each user during described targeted customer and described user gather;
Wherein, the project kind that the value of described parameter is corresponding according to project in described project set is determined.
Optionally, described in conjunction with this predicted value choose described to be recommended give targeted customer project set at least one project to targeted customer recommend comprise:
Choose described at least one project the highest to predicted value described in the project set of targeted customer to be recommended to recommend to described targeted customer; Or
Choose the project that predicted value described in described project set is greater than setting value to recommend to described targeted customer.
The described recommendation apparatus embodiment for participating in project is as follows:
With reference to Fig. 6, it illustrates the recommendation apparatus schematic diagram for participating in project that the present embodiment provides.
Because device embodiment is substantially similar to embodiment of the method, so describe fairly simple, the correspondence that relevant part refers to embodiment of the method illustrates.The device embodiment of following description is only schematic.
The recommendation apparatus for participating in project described in the present embodiment, comprising:
User's acquiring unit 601, for obtaining user's set that in the project set that relates to the participative behavior of targeted customer, project be made up of the user crossing participative behavior;
Affect force value acquiring unit 602, for obtain this user set in each user affect force value;
Similarity acquiring unit 603, obtains the similarity between each user of described user set and described targeted customer for the project involved by the participative behavior of user;
Influence power similarity calculated 604, for the similarity affected in force value and described user set between each user and targeted customer in conjunction with each user in described user's set, the influence power similarity during calculating targeted customer and described user gather between each user;
Project acquiring unit 605 to be recommended, the project set that the participative behavior for obtaining at least one the user that influence power similarity is the highest in described user set relates to, and using this project set as the project set giving targeted customer to be recommended;
Recommendation unit 606, for calculating targeted customer to described to be recommended to the predicted value of interest-degree of each project in the project set of targeted customer, and choose described to be recommendedly to recommend to targeted customer at least one project in the project set of targeted customer in conjunction with this predicted value.
Although the application with preferred embodiment openly as above; but it is not for limiting the application; any those skilled in the art are not departing from the spirit and scope of the application; can make possible variation and amendment, the scope that therefore protection domain of the application should define with the application's claim is as the criterion.
In one typically configuration, computing equipment comprises one or more processor (CPU), input/output interface, network interface and internal memory.
Internal memory may comprise the volatile memory in computer-readable medium, and the forms such as random access memory (RAM) and/or Nonvolatile memory, as ROM (read-only memory) (ROM) or flash memory (flashRAM).Internal memory is the example of computer-readable medium.
1, computer-readable medium comprises permanent and impermanency, removable and non-removable media can be stored to realize information by any method or technology.Information can be computer-readable instruction, data structure, the module of program or other data.The example of the storage medium of computing machine comprises, but be not limited to phase transition internal memory (PRAM), static RAM (SRAM), dynamic RAM (DRAM), the random access memory (RAM) of other types, ROM (read-only memory) (ROM), Electrically Erasable Read Only Memory (EEPROM), fast flash memory bank or other memory techniques, read-only optical disc ROM (read-only memory) (CD-ROM), digital versatile disc (DVD) or other optical memory, magnetic magnetic tape cassette, tape magnetic rigid disk stores or other magnetic storage apparatus or any other non-transmitting medium, can be used for storing the information can accessed by computing equipment.According to defining herein, computer-readable medium does not comprise non-temporary computer readable media (transitorymedia), as data-signal and the carrier wave of modulation.
2, it will be understood by those skilled in the art that the embodiment of the application can be provided as method, system or computer program.Therefore, the application can adopt the form of complete hardware embodiment, completely software implementation or the embodiment in conjunction with software and hardware aspect.And the application can adopt in one or more form wherein including the upper computer program implemented of computer-usable storage medium (including but not limited to magnetic disk memory, CD-ROM, optical memory etc.) of computer usable program code.
Claims (21)
1. participate in a recommend method for project for user, it is characterized in that, comprising:
Obtain to be recommended to the project set of targeted customer;
Obtain and the described user be made up of the user crossing participative behavior to project in the project set of targeted customer to be recommended is gathered;
What obtain each user in this user set affects force value;
Project involved by the participative behavior of user obtains the similarity between each user of described user set and described targeted customer;
In conjunction with the similarity affected in force value and this user set between each user and described targeted customer of each user in described user's set, calculate described targeted customer to described to be recommended to the predicted value of the interest-degree of each project in the project set of targeted customer;
In conjunction with this predicted value choose described to be recommended give targeted customer project set at least one project recommend to described targeted customer;
Wherein, the force value that affects of described user is determined according to the historical operation behavior of user.
2. the recommend method participating in project for user according to claim 1, is characterized in that, the described acquisition project set of targeted customer of giving to be recommended comprises:
Obtain the project involved by participative behavior of described targeted customer;
The item destination aggregation (mda)s having the participative behavior of user all under obtaining project kind corresponding to this project, using described item destination aggregation (mda) as the project set giving targeted customer to be recommended.
3. the recommend method participating in project for user according to claim 1, is characterized in that, the force value that affects of each user in this user of described acquisition set comprises:
Obtain each user in described user set and be directed to the liveness of each project in described project set, and form the matrix of user and project;
Generate and initialising subscriber vector sum project latent vector, wherein, the element value of described user vector represent user in described user set initial affect force value, the element value of described project latent vector represents the initial value of the attribute of project in described project set;
By energy transmission iterative algorithm, iterative computation is carried out to the matrix of described user and project, user vector and project latent vector, and the element value of user vector after described iterative computation is affected force value as user in described user's set;
Wherein, described liveness characterizes each user in described user set and is directed to the degree of participation of each project in described project set.
4. the recommend method participating in project for user according to claim 3, is characterized in that, the liveness that in described acquisition described user set, each user is directed to each project in described project set comprises:
For each user in described user's set, obtain the set that this user is directed to all participative behaviors of each project in described project set, and the behavior weight that in this set, the participation number of times of each participative behavior is corresponding with each participative behavior;
Calculate the participation number of times of each participative behavior in described set and the product of behavior weight, and calculate the weighted value of the participation number of times of all participative behaviors in this set and the product of behavior weight, this weighted value is directed to the liveness of each project in described project set as each user in described user's set.
5. the recommend method of the project that participates in for user according to claim 3 or 4, is characterized in that, describedly carries out iterative computation by energy transmission iterative algorithm to the matrix of described user and project, user vector and project latent vector and comprises:
Calculate the product of the transpose of a matrix matrix of described project latent vector and described user and project, using the value of this product as described user vector, and the acting force this calculating be considered as in project is applied to the primary energy propagation of user force by user behavior;
To calculate the product of the matrix of user vector and described user and the project obtained as project vector, and the influence power this calculating being considered as user is propagated to the primary energy of other users by there being the project of user's participative behavior;
Above-mentioned energy transmission calculating process is repeated according to iterations;
Wherein, described iterations is determined according to the empirical value of setting or is less than according to the difference of the mould of calculated vector after twice adjacent calculation arrange fixed value and determine.
6. the recommend method participating in project for user according to claim 1, is characterized in that, the similarity that described project involved by the participative behavior of user obtains in described user set between each user and described targeted customer comprises:
Similarity algorithm is utilized to calculate the similarity between each user and targeted customer of described user set;
Wherein, described similarity algorithm comprises: cosine similarity algorithm, Pearson (Pearson) related coefficient algorithm or adjustment cosine similarity algorithm.
7. the recommend method participating in project for user according to claim 1, it is characterized in that, in user's set described in described combination each user affect similarity between each user and described targeted customer in force value and this user set, calculate described targeted customer and to be recommendedly to comprise to the predicted value of the interest-degree of each project in the project set of targeted customer described:
Calculate the product of the product affecting the force value parameter corresponding with affecting force value of each user in described user set and the described similarity parameter corresponding with similarity; And calculate the value of the two product addition gained described;
Value based on the two product addition gained described calculates described targeted customer to described to be recommended to the predicted value of the interest-degree of each project in the project set of targeted customer;
Wherein, the project kind that the value of described parameter is corresponding according to project in described project set is determined.
8. the recommend method participating in project for user according to claim 1, is characterized in that, described choosing in conjunction with this predicted value describedly to be recommendedly recommends to comprise to described targeted customer at least one project in the project set of targeted customer:
Choose described at least one project the highest to predicted value described in the project set of targeted customer to be recommended to recommend to described targeted customer; Or
Choose the project that predicted value described in described project set is greater than setting value to recommend to described targeted customer.
9. participate in a recommendation apparatus for project for user, it is characterized in that, comprising:
Project acquiring unit, obtains to be recommended to the project set of targeted customer;
User's acquiring unit, gathers the described user be made up of the user crossing participative behavior to project in the project set of targeted customer to be recommended for obtaining;
Affect force value acquiring unit, for obtain this user set in each user affect force value;
Similarity acquiring unit, obtains the similarity between each user of described user set and described targeted customer for the project involved by the participative behavior of user;
Predictor calculation unit, for the similarity affected in force value and this user set between each user and described targeted customer in conjunction with each user in described user's set, calculate described targeted customer to described to be recommended to the predicted value of the interest-degree of each project in the project set of targeted customer;
Recommendation unit, for choose in conjunction with this predicted value described to be recommended give targeted customer project set at least one project recommend to described targeted customer.
10. computing method for customer impact force value, is characterized in that, comprising:
Obtain to project by user's set of crossing the user of participative behavior and form, and had the project set of item design of user's participative behavior;
Each user obtaining described user set is directed to the liveness of each project in described project set, and forms the matrix of user and project;
Generate and initialising subscriber vector and project latent vector, wherein, the element value of described user vector represent user in described user set initial affect force value, the element value of described project latent vector represents the initial value of the attribute of project in described project set;
By energy transmission iterative algorithm, iterative computation is carried out to described matrix, user vector and project latent vector, and the element value of the user vector after described iterative computation is affected force value as user in described user set;
Wherein, described liveness characterizing consumer is directed to the degree of participation of project, and the force value that affects of described user is determined according to the historical operation behavior of user.
The computing method of 11. customer impact force value according to claim 10, is characterized in that, the liveness that each user that the described user of described acquisition gathers is directed to each project in described project set comprises:
For each user in user's set, obtain the set that user is directed to all participative behaviors of each project in described project set, and the behavior weight that in the set of this participative behavior, the participation number of times of each participative behavior is corresponding with each participative behavior;
Calculate the participation number of times of each participative behavior in described set and the product of behavior weight, and calculate the weighted value of the participation number of times of all participative behaviors in this set and the product of behavior weight, this weighted value is directed to the liveness of each project in described project set as each user in described user's set.
12., according to the computing method of the arbitrary described customer impact force value of claim 9 to 11, is characterized in that, describedly carry out iterative computation by energy transmission iterative algorithm to described matrix, user vector and project latent vector and comprise:
Calculate the product of the transpose of a matrix matrix of described project latent vector and described user and project, using the value of this product as described user vector, and the acting force this calculating be considered as in project is applied to the primary energy propagation of user force by user behavior;
To calculate the product of the matrix of user vector and described user and the project obtained as project vector, and the influence power this calculating being considered as user is propagated to the primary energy of other users by there being the project of user's participative behavior;
Above-mentioned energy transmission calculating process is repeated according to iterations;
Wherein, described iterations is determined according to the empirical value of setting or is less than according to the difference of the mould of calculated vector after twice adjacent calculation arrange fixed value and determine.
The calculation element of 13. 1 kinds of customer impact force value, is characterized in that, comprising:
Data capture unit, for obtaining to project by user's set of crossing the user of participative behavior and form, and had the project set of item design of user's participative behavior;
Liveness generation unit, each user for obtaining described user set is directed to the liveness of each project in described project set, and forms the matrix of user and project, and wherein, described liveness characterizing consumer is directed to the degree of participation of project;
Vector initialising unit, for generating and initialising subscriber vector and project latent vector, wherein, the element value of described user vector represent user in described user set initial affect force value, the element value of described project latent vector represents the initial value of project in described project set;
Affect force value computing unit, for carrying out iterative computation by energy transmission iterative algorithm to described matrix, user vector and project latent vector, and the element value of the user vector after described iterative computation is affected force value as user in described user set, wherein, the force value that affects of described user is determined according to the historical operation behavior of user.
14. 1 kinds participate in the recommend method of project for user, it is characterized in that, comprising:
User's set that the project in the project set related to the participative behavior of targeted customer that obtains be made up of the user crossing participative behavior;
What obtain each user in this user set affects force value;
Project involved by the participative behavior of user obtains the similarity between each user of described user set and described targeted customer;
In conjunction with the similarity affected in force value and described user set between each user and targeted customer of each user in described user's set, the influence power similarity during calculating targeted customer and described user gather between each user;
The project set that the participative behavior obtaining at least one the user that influence power similarity is the highest in described user set relates to, and using this project set as the project set giving targeted customer to be recommended;
Calculate targeted customer to described to be recommended to the predicted value of interest-degree of each project in the project set of targeted customer, and choose described to be recommendedly to recommend to targeted customer at least one project in the project set of targeted customer in conjunction with this predicted value;
Wherein, described user affect force value characterizing consumer influence degree to other people in project activity; Described influence power similarity characterization is in the similarity affected under the participation of force value between user of user.
15. recommend methods participating in project for user according to claim 14, is characterized in that, the force value that affects of each user in this user of described acquisition set comprises:
The project set of the item design that the participative behavior obtaining user in described user set relates to;
Each user obtaining described user set is directed to the liveness of each project in described project set, and forms the matrix of user and project;
Generate and initialising subscriber vector and project latent vector, wherein, the element value of described user vector represent user in described user set initial affect force value, the element value of described project latent vector represents the initial value of the attribute of project in described project set;
By energy transmission iterative algorithm, iterative computation is carried out to described matrix, user vector and project latent vector, and the element value of user vector after described iterative computation is affected force value as user in described user set;
Wherein, described liveness characterizes each user in described user set and is directed to the degree of participation of each project in described project set.
16. recommend methods participating in project for user according to claim 15, is characterized in that, the liveness of each project that each user that the described user of described acquisition gathers is directed in described project set comprises:
For each user in described user's set, obtain the set that this user is directed to all participative behaviors of each project in described project set, and the behavior weight that in the set of this participative behavior, the participation number of times of each participative behavior is corresponding with each participative behavior;
Calculate the participation number of times of each participative behavior in the set of described participative behavior and the product of behavior weight, and calculate the weighted value of the participation number of times of all participative behaviors in this set and the product of behavior weight, this weighted value is directed to the liveness of each project in described project set as each user in described user's set.
The recommend method of 17. projects that participate in for user according to claim 15 or 16, is characterized in that, describedly carries out iterative computation by energy transmission iterative algorithm to described matrix, user vector and project latent vector and comprises:
Calculate the product of the transpose of a matrix matrix of described project latent vector and described user and project, using the value of this product as described user vector, and the acting force this calculating be considered as in project is applied to the primary energy propagation of user force by user behavior;
To calculate the product of the matrix of user vector and described user and the project obtained as project vector, and the influence power this calculating being considered as user is propagated to the primary energy of other users by there being the project of user's participative behavior;
Above-mentioned energy transmission calculating process is repeated according to iterations;
Wherein, described iterations is determined according to the empirical value of setting or is less than according to the difference of the mould of calculated vector after twice adjacent calculation arrange fixed value and determine.
18. recommend methods participating in project for user according to claim 14, is characterized in that, the similarity that described project involved by the participative behavior of user obtains between each user of described user set and described targeted customer comprises:
Similarity algorithm is utilized to calculate the similarity between each user and targeted customer of described user set;
Wherein, described similarity algorithm comprises: cosine similarity algorithm, Pearson (Pearson) related coefficient algorithm or adjustment cosine similarity algorithm.
19. recommend methods participating in project for user according to claim 14, it is characterized in that, the similarity affected in force value and described user set between each user and targeted customer of each user in user's set described in described combination, the influence power similarity during calculating targeted customer and described user gather between each user comprises:
Calculate the product affecting the force value parameter corresponding with affecting force value of each user in described user set, and the product of the described similarity parameter corresponding with similarity;
Using the value of the two product addition gained described as the influence power similarity between each user during described targeted customer and described user gather;
Wherein, the project kind that the value of described parameter is corresponding according to project in described project set is determined.
20. recommend methods participating in project for user according to claim 14, is characterized in that, described in conjunction with this predicted value choose described to be recommended give targeted customer project set at least one project to targeted customer recommend comprise:
Choose described at least one project the highest to predicted value described in the project set of targeted customer to be recommended to recommend to described targeted customer; Or
Choose the project that predicted value described in described project set is greater than setting value to recommend to described targeted customer.
21. 1 kinds participate in the recommendation apparatus of project for user, it is characterized in that, comprising:
User's acquiring unit, for obtaining user's set that in the project set that relates to the participative behavior of targeted customer, project be made up of the user crossing participative behavior;
Affect force value acquiring unit, for obtain this user set in each user affect force value;
Similarity acquiring unit, obtains the similarity between each user of described user set and described targeted customer for the project involved by the participative behavior of user;
Influence power similarity calculated, for the similarity affected in force value and described user set between each user and targeted customer in conjunction with each user in described user's set, the influence power similarity during calculating targeted customer and described user gather between each user;
Project acquiring unit to be recommended, the project set that the participative behavior for obtaining at least one the user that influence power similarity is the highest in described user set relates to, and using this project set as the project set giving targeted customer to be recommended;
Recommendation unit, for calculating targeted customer to described to be recommended to the predicted value of interest-degree of each project in the project set of targeted customer, and choose described to be recommendedly to recommend to targeted customer at least one project in the project set of targeted customer in conjunction with this predicted value.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410195669.0A CN105095267B (en) | 2014-05-09 | 2014-05-09 | Recommendation method and device for user participation items |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410195669.0A CN105095267B (en) | 2014-05-09 | 2014-05-09 | Recommendation method and device for user participation items |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105095267A true CN105095267A (en) | 2015-11-25 |
CN105095267B CN105095267B (en) | 2019-12-31 |
Family
ID=54575721
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410195669.0A Active CN105095267B (en) | 2014-05-09 | 2014-05-09 | Recommendation method and device for user participation items |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105095267B (en) |
Cited By (20)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105468786A (en) * | 2015-12-28 | 2016-04-06 | 湖南蚁坊软件有限公司 | Microblog user activeness analysis method |
CN106651427A (en) * | 2016-10-13 | 2017-05-10 | 中国工商银行股份有限公司 | Data association method based on user behavior |
CN106709076A (en) * | 2017-02-27 | 2017-05-24 | 华南理工大学 | Social network recommendation device and method based on collaborative filtering |
CN106952130A (en) * | 2017-02-27 | 2017-07-14 | 华南理工大学 | Common user item based on collaborative filtering recommends method |
CN107704997A (en) * | 2017-09-12 | 2018-02-16 | 广东万丈金数信息技术股份有限公司 | Assess the method, apparatus and storage medium of client liveness |
CN107886028A (en) * | 2016-09-29 | 2018-04-06 | 九阳股份有限公司 | The food materials input method and food materials input device of a kind of refrigerator |
CN108039952A (en) * | 2017-11-15 | 2018-05-15 | 黄敬易 | Resource transmission method, device, server and computer-readable recording medium |
CN108985806A (en) * | 2017-05-31 | 2018-12-11 | 北京京东尚科信息技术有限公司 | Method and apparatus for selecting item property |
CN109145223A (en) * | 2018-09-28 | 2019-01-04 | 合肥工业大学 | A kind of social recommendation method propagated based on social influence power |
CN109214848A (en) * | 2017-07-06 | 2019-01-15 | 云义科技股份有限公司 | Method and system for analyzing influence similarity of virtual commodities on recommendation system |
CN109218769A (en) * | 2018-09-30 | 2019-01-15 | 武汉斗鱼网络科技有限公司 | A kind of recommended method and relevant device of direct broadcasting room |
CN109727109A (en) * | 2018-12-28 | 2019-05-07 | 深圳市华云中盛科技有限公司 | Concerning taxes service organization recommended method, device, computer equipment and storage medium |
CN110020211A (en) * | 2017-10-23 | 2019-07-16 | 北京京东尚科信息技术有限公司 | The appraisal procedure and device of user property influence power |
CN110222257A (en) * | 2019-05-13 | 2019-09-10 | 特斯联(北京)科技有限公司 | A kind of method, apparatus and data-link node for recommending business information |
CN110264337A (en) * | 2019-05-28 | 2019-09-20 | 阿里巴巴集团控股有限公司 | A kind of building of prediction model and bank card belonging country prediction technique and device |
CN110929172A (en) * | 2019-11-27 | 2020-03-27 | 中科曙光国际信息产业有限公司 | Information selection method and device, electronic equipment and readable storage medium |
CN111191882A (en) * | 2019-12-17 | 2020-05-22 | 安徽大学 | Method and device for identifying influential developers in heterogeneous information network |
CN111815197A (en) * | 2020-07-24 | 2020-10-23 | 上海风秩科技有限公司 | Influence index calculation method, device, equipment and storage medium |
CN112949322A (en) * | 2021-04-27 | 2021-06-11 | 李蕊男 | E-commerce opinion mining recommendation system driven by online text comments |
CN115048615A (en) * | 2021-03-09 | 2022-09-13 | 中核武汉核电运行技术股份有限公司 | Data recommendation method based on nuclear power plant platform |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100076857A1 (en) * | 2008-09-25 | 2010-03-25 | Harshal Ulhas Deo | Methods and systems for activity-based recommendations |
CN102479202A (en) * | 2010-11-26 | 2012-05-30 | 卓望数码技术(深圳)有限公司 | Recommendation system based on domain expert |
CN102576438A (en) * | 2009-09-21 | 2012-07-11 | 瑞典爱立信有限公司 | Method and apparatus for executing a recommendation |
CN102915307A (en) * | 2011-08-02 | 2013-02-06 | 腾讯科技(深圳)有限公司 | Device and method for recommending personalized information and information processing system |
US20130304731A1 (en) * | 2010-12-31 | 2013-11-14 | Yahoo! Inc. | Behavior targeting social recommendations |
-
2014
- 2014-05-09 CN CN201410195669.0A patent/CN105095267B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100076857A1 (en) * | 2008-09-25 | 2010-03-25 | Harshal Ulhas Deo | Methods and systems for activity-based recommendations |
CN102576438A (en) * | 2009-09-21 | 2012-07-11 | 瑞典爱立信有限公司 | Method and apparatus for executing a recommendation |
CN102479202A (en) * | 2010-11-26 | 2012-05-30 | 卓望数码技术(深圳)有限公司 | Recommendation system based on domain expert |
US20130304731A1 (en) * | 2010-12-31 | 2013-11-14 | Yahoo! Inc. | Behavior targeting social recommendations |
CN102915307A (en) * | 2011-08-02 | 2013-02-06 | 腾讯科技(深圳)有限公司 | Device and method for recommending personalized information and information processing system |
Non-Patent Citations (3)
Title |
---|
SYLVAIN SENECAL ETAL.: "The influence of online product recommendations on consumers" online choices", 《JOURNAL OF RETAILING》 * |
吴杉: "引入用户权威性的分类协同过滤推荐算法的研究", 《焦作市高等专科学校学报》 * |
王喆等: "基于权威度的指导者挖掘与个性化推荐方法", 《中国科技论文在线》 * |
Cited By (31)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105468786A (en) * | 2015-12-28 | 2016-04-06 | 湖南蚁坊软件有限公司 | Microblog user activeness analysis method |
CN107886028A (en) * | 2016-09-29 | 2018-04-06 | 九阳股份有限公司 | The food materials input method and food materials input device of a kind of refrigerator |
CN106651427B (en) * | 2016-10-13 | 2020-02-11 | 中国工商银行股份有限公司 | Data association method based on user behaviors |
CN106651427A (en) * | 2016-10-13 | 2017-05-10 | 中国工商银行股份有限公司 | Data association method based on user behavior |
CN106709076A (en) * | 2017-02-27 | 2017-05-24 | 华南理工大学 | Social network recommendation device and method based on collaborative filtering |
CN106952130A (en) * | 2017-02-27 | 2017-07-14 | 华南理工大学 | Common user item based on collaborative filtering recommends method |
CN106709076B (en) * | 2017-02-27 | 2023-09-29 | 华南理工大学 | Social network recommendation device and method based on collaborative filtering |
CN106952130B (en) * | 2017-02-27 | 2020-10-27 | 华南理工大学 | General article recommendation method based on collaborative filtering |
CN108985806A (en) * | 2017-05-31 | 2018-12-11 | 北京京东尚科信息技术有限公司 | Method and apparatus for selecting item property |
CN109214848B (en) * | 2017-07-06 | 2020-10-27 | 云义科技股份有限公司 | Method and system for analyzing influence similarity of virtual commodities on recommendation system |
CN109214848A (en) * | 2017-07-06 | 2019-01-15 | 云义科技股份有限公司 | Method and system for analyzing influence similarity of virtual commodities on recommendation system |
CN107704997A (en) * | 2017-09-12 | 2018-02-16 | 广东万丈金数信息技术股份有限公司 | Assess the method, apparatus and storage medium of client liveness |
CN110020211B (en) * | 2017-10-23 | 2021-08-17 | 北京京东尚科信息技术有限公司 | Method and device for evaluating influence of user attributes |
CN110020211A (en) * | 2017-10-23 | 2019-07-16 | 北京京东尚科信息技术有限公司 | The appraisal procedure and device of user property influence power |
CN108039952B (en) * | 2017-11-15 | 2020-07-21 | 黄敬易 | Resource propagation method, device, server and computer readable storage medium |
CN108039952A (en) * | 2017-11-15 | 2018-05-15 | 黄敬易 | Resource transmission method, device, server and computer-readable recording medium |
CN109145223B (en) * | 2018-09-28 | 2021-03-09 | 合肥工业大学 | Social recommendation method based on social influence propagation |
CN109145223A (en) * | 2018-09-28 | 2019-01-04 | 合肥工业大学 | A kind of social recommendation method propagated based on social influence power |
CN109218769B (en) * | 2018-09-30 | 2021-01-01 | 武汉斗鱼网络科技有限公司 | Recommendation method for live broadcast room and related equipment |
CN109218769A (en) * | 2018-09-30 | 2019-01-15 | 武汉斗鱼网络科技有限公司 | A kind of recommended method and relevant device of direct broadcasting room |
CN109727109A (en) * | 2018-12-28 | 2019-05-07 | 深圳市华云中盛科技有限公司 | Concerning taxes service organization recommended method, device, computer equipment and storage medium |
CN110222257B (en) * | 2019-05-13 | 2020-01-31 | 特斯联(北京)科技有限公司 | method and device for recommending service information and data link node |
CN110222257A (en) * | 2019-05-13 | 2019-09-10 | 特斯联(北京)科技有限公司 | A kind of method, apparatus and data-link node for recommending business information |
CN110264337A (en) * | 2019-05-28 | 2019-09-20 | 阿里巴巴集团控股有限公司 | A kind of building of prediction model and bank card belonging country prediction technique and device |
CN110929172A (en) * | 2019-11-27 | 2020-03-27 | 中科曙光国际信息产业有限公司 | Information selection method and device, electronic equipment and readable storage medium |
CN110929172B (en) * | 2019-11-27 | 2022-11-18 | 中科曙光国际信息产业有限公司 | Information selection method and device, electronic equipment and readable storage medium |
CN111191882A (en) * | 2019-12-17 | 2020-05-22 | 安徽大学 | Method and device for identifying influential developers in heterogeneous information network |
CN111191882B (en) * | 2019-12-17 | 2022-11-25 | 安徽大学 | Method and device for identifying influential developers in heterogeneous information network |
CN111815197A (en) * | 2020-07-24 | 2020-10-23 | 上海风秩科技有限公司 | Influence index calculation method, device, equipment and storage medium |
CN115048615A (en) * | 2021-03-09 | 2022-09-13 | 中核武汉核电运行技术股份有限公司 | Data recommendation method based on nuclear power plant platform |
CN112949322A (en) * | 2021-04-27 | 2021-06-11 | 李蕊男 | E-commerce opinion mining recommendation system driven by online text comments |
Also Published As
Publication number | Publication date |
---|---|
CN105095267B (en) | 2019-12-31 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105095267A (en) | User involving project recommendation method and apparatus | |
Davoudi et al. | Social trust model for rating prediction in recommender systems: Effects of similarity, centrality, and social ties | |
Negahban et al. | Agent-based simulation applications in marketing research: an integrated review | |
Jiang et al. | Impacts of knowledge on online brand success: an agent-based model for online market share enhancement | |
Uitz | Social Media-Is It Worth the Trouble? | |
Weideman et al. | Parallel search engine optimisation and pay-per-click campaigns: A comparison of cost per acquisition | |
CN108416625A (en) | The recommendation method and apparatus of marketing product | |
US20170300939A1 (en) | Optimizing promotional offer mixes using predictive modeling | |
Arenas-Márquez et al. | Electronic word-of-mouth communities from the perspective of social network analysis | |
CN103399858A (en) | Socialization collaborative filtering recommendation method based on trust | |
CN105096152A (en) | Commodity popularity-based operation execution method and device | |
Yeo et al. | Predicting online purchase conversion for retargeting | |
WO2012064553A1 (en) | System and method for optimizing marketing effectiveness | |
Almajali et al. | Antecedents of consumers’ adoption of electronic commerce in developing countries | |
CN107292648A (en) | A kind of user behavior analysis method and device | |
Chan et al. | Continuous model selection for large-scale recommender systems | |
Najafi et al. | Investigation of the correlation between trust and reputation in B2C e-commerce using Alexa ranking | |
CN107993088A (en) | A kind of Buying Cycle Forecasting Methodology and device, electronic equipment | |
CN108446297A (en) | A kind of recommendation method and device, electronic equipment | |
CN103971256A (en) | Information push method and device | |
Tian et al. | A practical model to predict the repeat purchasing pattern of consumers in the C2C e-commerce | |
Hruschka | Comparing Small‐and Large‐Scale Models of Multicategory Buying Behavior | |
Avinadav et al. | Contracting under information superiority in a supply chain of subscription-based apps: a comparative analysis | |
JP6810184B2 (en) | Calculation device, calculation method and calculation program | |
Däs et al. | Customer lifetime network value: customer valuation in the context of network effects |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |