CN108921670A - A kind of potential interest of fusion user, the Drug trading recommended method of space-time data and classification popularity - Google Patents
A kind of potential interest of fusion user, the Drug trading recommended method of space-time data and classification popularity Download PDFInfo
- Publication number
- CN108921670A CN108921670A CN201810724191.4A CN201810724191A CN108921670A CN 108921670 A CN108921670 A CN 108921670A CN 201810724191 A CN201810724191 A CN 201810724191A CN 108921670 A CN108921670 A CN 108921670A
- Authority
- CN
- China
- Prior art keywords
- user
- drug
- matrix
- potential
- data
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0631—Item recommendations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
Abstract
The invention discloses a kind of potential interest of fusion user, the Drug trading recommended method of space-time data and classification popularity, including obtaining the purchaser record data of user's drug purchase from the data set of electric business platform, and purchaser record data are arranged to obtain user-drug rating matrix;Purchaser record based on similar users in purchaser record data establishes the potential interest model of user, and obtains the potential interesting data of user based on the potential interest model of user;The potential interesting data of user is merged into user-drug rating matrix;The popularity for the drug generic bought based on user in purchaser record data and user establish classification correlation model to the preference of the category;Matrix decomposition is carried out to the user-drug rating matrix for incorporating the potential interesting data of user, and obtained user preference prediction matrix and classification correlation model progress linear fusion generation recommendation list will be decomposed.The present invention solves the problems, such as that rating matrix sparsity impacts recommendation efficiency in the prior art.
Description
Technical field
The present invention relates to field of computer technology more particularly to a kind of potential interest of fusion user, space-time data and classification
The Drug trading recommended method of popularity.
Background technique
In recent years, e-commerce becomes more and more active with the development of internet and information technology, and more and more consumers is opened
Beginning shopping online.New business profit channel has not only been opened up in e-commerce, traditional sales mode has also been overturned, from space
The more conveniences of both parties, independence are assigned on the upper, time.Wherein, medical, as aily life necessitys, in recent years
Also start to progress into electric business field, more and more pharmaceutical manufacturers obtain the qualification for establishing electronic medicine business platform, doctor
The a piece of light of e-commerce development future of medicine industry.
Since medical electric business platform includes multiple types, the drug of big quantity, user requires a great deal of time and energy
The drug for removing to filter out needs, greatly reduces the user experience of platform.It was expended to solve user in magnanimity drug
Personalized recommendation technology is introduced into medical electric business platform and is necessary by the problem of more times lead to poor user experience.
In medical electric business platform, due to the particularity of drug, user can push away the scoring quantity of drug far below tradition
Middle user is recommended to the scoring quantity of article (music, film), user-drug rating matrix is very sparse, medical electric business platform
Recommend to be faced with the data cold start-up problem even more serious than conventional recommendation.
In face of medical electric business field magnanimity and diversified drug, how to design outstanding proposed algorithm and provide essence for user
True recommendation is a problem worth thinking deeply about.Currently, the field has existed some proposed algorithms, but these algorithms are most
It is to be carried out on original user-drug rating matrix, is influenced by the rating matrix sparsity very big.
Therefore, how effectively to alleviate rating matrix sparsity influences to be one urgently to be resolved caused by recommending efficiency
Problem.
Summary of the invention
In view of this, the present invention provides a kind of potential interest of fusion user, the drug of space-time data and classification popularity
Then recommended method of trading fills the potential interest of user by the historical purchase data study of user to the potential interest of user
Into user-drug rating matrix, efficiently solves rating matrix sparsity in the prior art and ask what recommendation efficiency impacted
Topic.
In order to realize above-mentioned purpose of the invention, the present invention provides a kind of potential interest of fusion user, space-time data and
The Drug trading recommended method of classification popularity, described method includes following steps:
S1 obtains the purchaser record data of user's drug purchase from the data set of electric business platform, and to purchaser record number
According to being arranged to obtain user-drug rating matrix;
S2, the purchaser record based on similar users in purchaser record data establish the potential interest model of user, and based on use
The potential interest model in family obtains the potential interesting data of user;
The potential interesting data of user is merged into user-drug rating matrix by S3;Alleviate matrix degree of rarefication to recommendation results
Caused by influence, improve recommend efficiency;
S4, the popularity for the drug generic bought based on user in purchaser record data and user are to the category
Preference establishes classification correlation model;
S5 carries out matrix decomposition to the user-drug rating matrix for incorporating the potential interesting data of user, and will decompose
To user preference prediction matrix and step S4 in classification correlation model carry out linear fusion generate recommendation list.
Preferably, the step S1 includes the following steps:
S1-1, arranges the purchaser record data of user, and purchaser record data include scoring, purchase of the user to the drug of purchase
Time, drug variety are bought, user set U={ u is obtained1,u2,...,ui...,unAnd drug set D={ d1,d2,...,
dj...,dm, wherein u indicates that user, i represent the ID of user;D indicates that drug, j represent the ID of drug;
S1-2 counts the Quantity of drugs that each user buys and scores, if user buys and the Quantity of drugs to score is low
In preset value, then this user is deleted;To obtain the user for including enough user informations;
S1-3 counts the number that each drug is purchased and scores, if the frequency that drug is purchased is lower than preset value,
The relative recording of this drug is deleted;Because of the missing of data, it is easy to appear noise;
S1-4 obtains original user-drug rating matrix based on the purchaser record data put in order.
Preferably, the step S2 includes the following steps:
S2-1 merges the similar users set F of time factori:
1) using the method time discretization, it was divided into T discrete time slot 1 year, user-original in step S1
Drug rating matrix is divided into T period-user-drug rating matrix according to the purchase scoring time;
2) target user i is given, defining scoring vector of the user i in period t (t ∈ T) is:ri,t={ ri,t,1,
ri,t,2,..ri,t,m, wherein ri,t,mIndicate user i in period t to the score value of drug m.For user i, calculates the user and exist
Any two period tpAnd tqScoring vectorWithCosine similarity, then take all users at the two more than the period
The average value of string similar value is as the two period similarities, to obtain the user in discrete time slot between any two period
Similarity;
3) similarity of all users between any two period in discrete time slot is expressed as a period similarity
Matrix TS, and period-user-drug rating matrix is translated using period similar matrix TS, specific translation formula is such as
Under:
Wherein,It is the new period-user-drug rating matrix that will be used to calculate that translation obtains later;It is
Indicate period t and t*Period similarity, t*∈[1,T];It is scoring of the user i in period t* to drug j;
Then user's similarity calculation is carried out using the matrix after translation, it is highest to obtain s similarity for user i
User is as similar users Fi;
S2-2 is based on similar users FiObtain the potential interesting data of user:
It is the similar users F of the user in S2-1 step for user iiBought what still user i was not bought
Spare potential interested drug of the drug as user i, and establish the potential interest model of user to learn the potential interest of user, from
And obtain the potential interesting data of user.
Preferably, the step S3 includes the following steps:
The potential interesting data of user is filled into user original in step S1-drug rating matrix, for each by S3-1
User i, is divided into three classes drug:DiIt is the set for the drug that user bought;PiIt is the potential drug purchase set of user;UiIt is
User did not buy and the set of drug purchase non-potential, then original user-drug rating matrix turns to new scoring square
Battle array and weight matrix:
Wherein, NewR is new rating matrix, NewRi,jIndicate scoring of the user i to drug j;NewW is new weight square
Battle array, NewWi,jIt is user i to the preference of drug j;It is when drug is the potential drug purchase of user, user is to the drug
Scoring, be the numerical value between 0 to 1;μ is adjustment parameter, and taking 0.3, * here is multiplication sign.
Preferably, the step S4 includes the following steps:
S4-1 establishes a user to some drug class by user to the rating matrix of drug and the type of drug
Other rating matrix BN,|C|, wherein N is number of users, | C | it is drug variety quantity, each element representation in rating matrix is used
Scoring of the family for classification belonging to the drug bought;
S4-2 constructs a drug popularity matrix P|C|,M, wherein | C | it is drug variety quantity, M is Quantity of drugs, medicine
Each element representation drug in product popularity matrix is purchased in the popularity of generic using certain drug in a certain classification
The number bought indicates the drug in the popularity of the category;
S4-3, the classification correlation model for obtaining user's drug purchase are as follows:
Wherein, yi,jIndicate scoring of the user i to drug j under class models;Bi,c∈BN,|C|, Pc,j∈P|C|,M。
Preferably, the step S5 includes following steps:
S5-1 is decomposed, decomposable process using new rating matrix and weight matrix of the matrix decomposition algorithm to acquisition
Middle error function is as follows:
Wherein, i indicates that user, j indicate that drug, N indicate that number of users, M indicate Quantity of drugs,It is the hidden factor of user
The product of matrix and the hidden factor matrix vector of drug indicates scoring of the user i to drug j;The weight of γ expression user and drug;
| U | indicate the hidden factor matrix of user, | D | indicate the hidden factor matrix of drug,Indicate the not Luo Beini of the hidden factor matrix of user
Square of this black norm,Indicate square of this black norm of the not Luo Beini of the hidden factor matrix of drug;
S5-2, new rating matrix and weight matrix obtain the hidden eigenmatrix of user and the hidden eigenmatrix of drug after decomposing,
Two matrix multiples obtained after decomposition are obtained user preference prediction matrix, then user preference prediction matrix is related to classification
Model merges, and it is as follows to obtain final recommended models:
Wherein,It is scoring of the user i to drug j;Be update after the hidden factor matrix of user and drug it is hidden because
The product of submatrix vector indicates that user i scores to the prediction of drug j;yi,jIt indicates under class models, user i is to drug j
Scoring;∝ indicates directly proportional;* multiplication is indicated.
S5-3, according toThe size of score value is ranked up, the drug of k before then score value being selected to come from big to small
Recommendation list is generated, user is recommended.
In conclusion the invention discloses a kind of potential interest of fusion user, the drug of space-time data and classification popularity
Transaction recommended method, obtains the purchaser record data of user's drug purchase, and to purchase first from the data set of electric business platform
Record data are arranged to obtain user-drug rating matrix;Then the purchase note based on similar users in purchaser record data
The potential interest model of user is established in record, and obtains the potential interesting data of user based on the potential interest model of user;Then by user
Potential interesting data is merged into user-drug rating matrix;And then the drug institute bought based on user in purchaser record data
The popularity and user for belonging to classification establish classification correlation model to the preference of the category;Finally to incorporating the potential interest number of user
According to user-drug rating matrix carry out matrix decomposition, and obtained user preference prediction matrix and classification relevant mode will be decomposed
Type carries out linear fusion and generates recommendation list.The present invention is learnt to the potential interest of user, so by the historical purchase data of user
The potential interest of user is filled into user-drug rating matrix afterwards, efficiently solves rating matrix sparsity pair in the prior art
The problem of recommending efficiency to impact.
Additional aspect and advantage of the invention will be set forth in part in the description, and will partially become from the following description
Obviously, or practice through the invention is recognized.
Detailed description of the invention
Above-mentioned and/or additional aspect of the invention and advantage will become from the description of the embodiment in conjunction with the following figures
Obviously and it is readily appreciated that, wherein:
Fig. 1 is the Drug trading of a kind of potential interest of fusion user disclosed by the invention, space-time data and classification popularity
The basic flow chart of recommended method;
Fig. 2 is the potential interest learning algorithm schematic diagram of user disclosed by the invention;
Fig. 3 is the establishment process schematic diagram of classification correlation model disclosed by the invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
In the description of the present invention, it is to be understood that, term " longitudinal direction ", " transverse direction ", "upper", "lower", "front", "rear",
The orientation or positional relationship of the instructions such as "left", "right", "vertical", "horizontal", "top", "bottom" "inner", "outside" is based on attached drawing institute
The orientation or positional relationship shown, is merely for convenience of description of the present invention and simplification of the description, rather than the dress of indication or suggestion meaning
It sets or element must have a particular orientation, be constructed and operated in a specific orientation, therefore should not be understood as to limit of the invention
System.
In the description of the present invention, unless otherwise specified and limited, it should be noted that term " installation ", " connected ",
" connection " shall be understood in a broad sense, for example, it may be mechanical connection or electrical connection, the connection being also possible to inside two elements can
, can also indirectly connected through an intermediary, for the ordinary skill in the art to be to be connected directly, it can basis
Concrete condition understands the concrete meaning of above-mentioned term.
The present invention provides a kind of potential interest of fusion user, the Drug trading recommendation side of space-time data and classification popularity
Method, as shown in Figure 1-3, include the following steps:
S1 obtains the purchaser record data of user's drug purchase from the data set of electric business platform, and to purchaser record number
According to being arranged to obtain user-drug rating matrix;
S2, the purchaser record based on similar users in purchaser record data establish the potential interest model of user, and based on use
The potential interest model in family obtains the potential interesting data of user;
The potential interesting data of user is merged into user-drug rating matrix by S3;Alleviate matrix degree of rarefication to recommendation results
Caused by influence, improve recommend efficiency;
S4, the popularity for the drug generic bought based on user in purchaser record data and user are to the category
Preference establishes classification correlation model;
S5 carries out matrix decomposition to the user-drug rating matrix for incorporating the potential interesting data of user, and will decompose
To user preference prediction matrix and step S4 in classification correlation model carry out linear fusion generate recommendation list.
Preferably, step S1 includes the following steps:
S1-1, arranges the purchaser record data of user, and purchaser record data include scoring, purchase of the user to the drug of purchase
Time, drug variety are bought, user set U={ u is obtained1,u2,...,ui...,unAnd drug set D={ d1,d2,...,
dj...,dm, wherein u indicates that user, i represent the ID of user;D indicates that drug, j represent the ID of drug;
S1-2 counts the Quantity of drugs that each user buys and scores, if user buys and the Quantity of drugs to score is low
In preset value, then this user is deleted;To obtain the user for including enough user informations;
S1-3 counts the number that each drug is purchased and scores, if the frequency that drug is purchased is lower than preset value,
The relative recording of this drug is deleted;Because of the missing of data, it is easy to appear noise;
S1-4 obtains original user-drug rating matrix based on the purchaser record data put in order.
Preferably, step S2 includes the following steps:
S2-1 merges the similar users set F of time factori:
1) using the method time discretization, it was divided into T discrete time slot 1 year, user-original in step S1
Drug rating matrix is divided into T period-user-drug rating matrix according to the purchase scoring time;
2) target user i is given, defining scoring vector of the user i in period t (t ∈ T) is:ri,t={ ri,t,1,
ri,t,2,..ri,t,m, wherein ri,t,mIndicate user i in period t to the score value of drug m.For user i, calculates the user and exist
Any two period tpAnd tqScoring vectorWithCosine similarity, then take all users at the two more than the period
The average value of string similar value is as the two period similarities, to obtain the user in discrete time slot between any two period
Similarity;
3) similarity of all users between any two period in discrete time slot is expressed as a period similarity
Matrix TS, and period-user-drug rating matrix is translated using period similar matrix TS, specific translation formula is such as
Under:
Wherein,It is the new period-user-drug rating matrix that will be used to calculate that translation obtains later;It is
Indicate period t and t*Period similarity, t*∈[1,T];It is scoring of the user i in period t* to drug j.
Then user's similarity calculation is carried out using the matrix after translation, it is highest to obtain s similarity for user i
User is as similar users Fi;
S2-2 is based on similar users FiObtain the potential interesting data of user:
It is the similar users F of the user in S2-1 step for user iiBought what still user i was not bought
Spare potential interested drug of the drug as user i, and establish the potential interest model of user to learn the potential interest of user, from
And obtain the potential interesting data of user.
Preferably, step S3 includes the following steps:
The potential interesting data of user is filled into user original in step S1-drug rating matrix, for each by S3-1
User i, is divided into three classes drug:DiIt is the set for the drug that user bought;PiIt is the potential drug purchase set of user;UiIt is
User did not buy and the set of drug purchase non-potential, then original user-drug rating matrix turns to new scoring square
Battle array and weight matrix:
Wherein, NewR is new rating matrix, NewRi,jIndicate scoring of the user i to drug j;NewW is new weight square
Battle array, NewWi,jIt is user i to the preference of drug j;It is when drug is the potential drug purchase of user, user is to the drug
Scoring, be the numerical value between 0 to 1;μ is adjustment parameter, and taking 0.3, * here is multiplication sign.
Preferably, step S4 includes the following steps:
S4-1 establishes a user to some drug class by user to the rating matrix of drug and the type of drug
Other rating matrix BN,|C|, wherein N is number of users, | C | it is drug variety quantity, each element representation in rating matrix is used
Scoring of the family for classification belonging to the drug bought;
S4-2 constructs a drug popularity matrix P|C|,M, wherein | C | it is drug variety quantity, M is Quantity of drugs, medicine
Each element representation drug in product popularity matrix is purchased in the popularity of generic using certain drug in a certain classification
The number bought indicates the drug in the popularity of the category;
S4-3, the classification correlation model for obtaining user's drug purchase are as follows:
Wherein, yi,jIndicate scoring of the user i to drug j under class models;Bi,c∈BN,|C|, Pc,j∈P|C|,M。
Preferably, step S5 includes following steps:
S5-1 is decomposed, decomposable process using new rating matrix and weight matrix of the matrix decomposition algorithm to acquisition
Middle error function is as follows:
Wherein, i indicates that user, j indicate that drug, N indicate that number of users, M indicate Quantity of drugs,It is the hidden factor of user
The product of matrix and the hidden factor matrix vector of drug indicates scoring of the user i to drug j;The weight of γ expression user and drug;
| U | indicate the hidden factor matrix of user, | D | indicate the hidden factor matrix of drug,Indicate the not Luo Beini of the hidden factor matrix of user
Square of this black norm,Indicate square of this black norm of the not Luo Beini of the hidden factor matrix of drug.
S5-2, new rating matrix and weight matrix obtain the hidden eigenmatrix of user and the hidden eigenmatrix of drug after decomposing,
Two matrix multiples obtained after decomposition are obtained user preference prediction matrix, then user preference prediction matrix is related to classification
Model merges, and it is as follows to obtain final recommended models:
Wherein,It is scoring of the user i to drug j;Be update after the hidden factor matrix of user and drug it is hidden because
The product of submatrix vector indicates that user i scores to the prediction of drug j;yi,jIt indicates under class models, user i is to drug j
Scoring;∝ indicates directly proportional;* multiplication is indicated.
S5-3, according toThe size of score value is ranked up, the drug of k before then score value being selected to come from big to small
Generate recommendation list.
In conclusion the invention discloses a kind of potential interest of fusion user, the drug of space-time data and classification popularity
Transaction recommended method, obtains the purchaser record data of user's drug purchase, and to purchase first from the data set of electric business platform
Record data are arranged to obtain user-drug rating matrix;Then the purchase note based on similar users in purchaser record data
The potential interest model of user is established in record, and obtains the potential interesting data of user based on the potential interest model of user;Then by user
Potential interesting data is merged into user-drug rating matrix;And then the drug institute bought based on user in purchaser record data
The popularity and user for belonging to classification establish classification correlation model to the preference of the category;Finally to incorporating the potential interest number of user
According to user-drug rating matrix carry out matrix decomposition, and obtained user preference prediction matrix and classification relevant mode will be decomposed
Type carries out linear fusion and generates recommendation list.The present invention is learnt to the potential interest of user, so by the historical purchase data of user
The potential interest of user is filled into user-drug rating matrix afterwards, efficiently solves rating matrix sparsity pair in the prior art
The problem of recommending efficiency to impact.
Specifically, in the above-described embodiments, it is potential emerging to learn user that the potential interest model of user is established in step S2-2
Interest can specifically learn the potential interest of user by following two selection algorithm:
The first selection algorithm is maximum selection rule strategy, using in the similar users of target user i for buying drug j
With the maximum preference to represent user of target user's similarity, which is expressed as follows:
Wherein, pri,jIndicate scoring of the user i to drug j,It is user i and its associated user with regard to drug
The similitude of j preference, f ∈ FiIt is the associated user of user i.
Second of selection algorithm is first routing strategy, in heterogeneous network G<V,E,A>In, V is node set, and E is side
Set, A is node species set.First path definition is the path of following formWherein, Ai
∈A,RiIndicate existing relationship between node, Ri∈{U-U,U-D,D-D}.Then for this yuan of path P, example path p if it exists
={ v1,v2...vn+1Be this yuan of path example, be the example path P of first path P all such example path definitions '.It is right
In each example path, paper defines a characteristic value concept and is used to describe node v1And vn+1Correlation, be expressed as cor
(p), it then the characteristic value in first path is exactly the sum of all example route characteristic values, is expressed as:
To example path p={ a1,a2...an+1, a1∈ U is user node, an+1∈ D is drug node, other aiIt is real
An intermediate node in example path.Indicate that the degree of association cor (p) between path p start node is the walk random used
Thought a, it is assumed that object is from node a1It sets out, in a network walk random, defining cor (p) is object according to example path p
Migration is to node an+1Probability, assume that and be independent from each other since step each in walk random is strolled,.Object is according to p
The probability strolled is equal to the product for the probability that each step is strolled, and calculation formula is as follows
Wherein Pro (ai,ai+1) indicate in random walk process from node aiDirectly arrive node ai+1Probability.In heterogeneous network
In network, formula is defined as:
Wherein N (ai) indicate be and ai+1The consistent node type of type.
End user's interest is expressed as:
pri,j=Eig (Pi,j)
Finally obtain the potential point of interest of target user.
Specifically, in the above-described embodiments, the matrix decomposition algorithm in step S5-1 is using following hidden matrix learning algorithm
Pseudocode:
It should be noted that system structure shown in Fig. 1-Fig. 3 of the present invention or method flow are of the invention some
Preferred embodiment is shown here and simply facilitates the understanding present invention and be not considered as limiting the invention, of the invention
Under thought guidance, it is within the scope of the present invention to implement the structure or method obtained according to the technique and scheme of the present invention,
Therefore not to repeat here.
In the description of this specification, reference term " one embodiment ", " some embodiments ", " example ", " specifically show
The description of example " or " some examples " etc. means specific features, structure, material or spy described in conjunction with this embodiment or example
Point is included at least one embodiment or example of the invention.In the present specification, schematic expression of the above terms are not
Centainly refer to identical embodiment or example.Moreover, particular features, structures, materials, or characteristics described can be any
One or more embodiment or examples in can be combined in any suitable manner.
The foregoing description of the disclosed embodiments enables those skilled in the art to implement or use the present invention.
Various modifications to these embodiments will be readily apparent to those skilled in the art, as defined herein
General Principle can be realized in other embodiments without departing from the spirit or scope of the present invention.Therefore, of the invention
It is not intended to be limited to the embodiments shown herein, and is to fit to and the principles and novel features disclosed herein phase one
The widest scope of cause.
Although an embodiment of the present invention has been shown and described, it will be understood by those skilled in the art that:Not
A variety of change, modification, replacement and modification can be carried out to these embodiments in the case where being detached from the principle of the present invention and objective, this
The range of invention is defined by the claims and their equivalents.
Claims (6)
1. a kind of potential interest of fusion user, the Drug trading recommended method of space-time data and classification popularity, which is characterized in that
Described method includes following steps:
S1, from the data set of electric business platform obtain user's drug purchase purchaser record data, and to purchaser record data into
Row arranges and obtains user-drug rating matrix;
S2, the purchaser record based on similar users in purchaser record data establish the potential interest model of user, and latent based on user
The potential interesting data of user is obtained in interest model;
The potential interesting data of user is merged into user-drug rating matrix by S3;
The preference of S4, the popularity of the drug generic bought based on user in purchaser record data and user to the category
Establish classification correlation model;
S5 carries out matrix decomposition to the user-drug rating matrix for incorporating the potential interesting data of user, and decomposition is obtained
Classification correlation model in user preference prediction matrix and step S4 carries out linear fusion and generates recommendation list.
2. the Drug trading recommendation of fusion user potential interest, space-time data and classification popularity according to claim 1
Method, which is characterized in that the step S1 includes the following steps:
S1-1 arranges the purchaser record data of user, when purchaser record data include scoring, purchase of the user to the drug of purchase
Between, drug variety, obtain user set U={ u1,u2,...,ui...,unAnd drug set D={ d1,d2,...,dj...,
dm, wherein u indicates that user, i represent the ID of user;D indicates that drug, j represent the ID of drug;
S1-2 counts the Quantity of drugs that each user buys and scores, if user buys and the Quantity of drugs to score is lower than in advance
If value then deletes this user;
S1-3 counts the number that each drug is purchased and scores, if the frequency that drug is purchased is lower than preset value, this
The relative recording of kind drug is deleted;
S1-4 obtains user-drug rating matrix based on the purchaser record data put in order.
3. the Drug trading recommendation of fusion user potential interest, space-time data and classification popularity according to claim 1
Method, which is characterized in that the step S2 includes the following steps:
S2-1 merges the similar users set F of time factori:
1) using the method time discretization, it was divided into T discrete time slot 1 year, user-drug original in step S1
Rating matrix is divided into T period-user-drug rating matrix according to the purchase scoring time;
2) target user i is given, defining scoring vector of the user i in period t (t ∈ T) is:ri,t={ ri,t,1,ri,t,2,
..ri,t,m, wherein ri,t,mIndicate that user i, to the score value of drug m, for user i, calculates the user any two in period t
A period tpAnd tqScoring vectorWithCosine similarity, then take all users similar in the cosine of the two periods
The average value of value is as the two period similarities, to obtain the similar of the user in discrete time slot between any two period
Degree;
3) similarity of all users between any two period in discrete time slot is expressed as a period similarity matrix
TS, and period-user-drug rating matrix is translated using period similarity matrix TS, specific translation formula is as follows:
Wherein,It is the new period-user-drug rating matrix that will be used to calculate that translation obtains later;It is to indicate
Period t and t*Period similarity, t*∈[1,T];It is scoring of the user i in period t* to drug j;
Then user's similarity calculation is carried out using the matrix after translation, the highest user of s similarity is obtained for user i
As similar users Fi;
S2-2 is based on similar users FiObtain the potential interesting data of user:
It is the similar users F of the user in S2-1 step for user iiBought the drug that still user i was not bought
As the spare potential drug interested of user i, and the potential interest model of user is established to learn the potential interest of user, thus
To the potential interesting data of user.
4. the Drug trading recommendation of fusion user potential interest, space-time data and classification popularity according to claim 1
Method, the step S3 include the following steps:
The potential interesting data of user is filled into the user in step S1-drug rating matrix by S3-1, for each user i,
Drug is divided into three classes:DiIt is the set for the drug that user bought;PiIt is the potential drug purchase set of user;UiIt is that user does not have
It bought and the set of drug purchase non-potential, then original user-drug rating matrix turns to new rating matrix and weight
Matrix:
Wherein, NewR is new rating matrix, NewRi,jIndicate scoring of the user i to drug j;NewW is new weight matrix,
NewWi,jIt is user i to the preference of drug j;It is when drug is the potential drug purchase of user, user is to the drug
Scoring, is the numerical value between 0 to 1;μ is adjustment parameter.
5. the Drug trading recommendation of fusion user potential interest, space-time data and classification popularity according to claim 1
Method, which is characterized in that the step S4 includes the following steps:
It is other to some drug class to establish a user by user to the rating matrix of drug and the type of drug by S4-1
Rating matrix BN,|C|, wherein N is number of users, | C | it is drug variety quantity, each element representation user couple in rating matrix
The scoring of the classification belonging to the drug bought;
S4-2 constructs a drug popularity matrix P|C|,M, wherein | C | it is drug variety quantity, M is Quantity of drugs, drug stream
Each element representation drug in row degree matrix is purchased in the popularity of generic using certain drug in a certain classification
Number indicates the drug in the popularity of the category;
S4-3, the classification correlation model for obtaining user's drug purchase are as follows:
Wherein, yi,jIndicate scoring of the user i to drug j under class models;Bi,c∈BN,|C|, Pc,j∈P|C|,M。
6. the Drug trading recommendation of fusion user potential interest, space-time data and classification popularity according to claim 1
Method, the step S5 includes following steps:
S5-1 is decomposed using new rating matrix and weight matrix of the matrix decomposition algorithm to acquisition, in decomposable process accidentally
Difference function is as follows:
Wherein, i indicates that user, j indicate that drug, N indicate that number of users, M indicate Quantity of drugs,It is the hidden factor matrix of user
With the product of the hidden factor matrix vector of drug, scoring of the user i to drug j is indicated;The weight of γ expression user and drug;|U|
Indicate the hidden factor matrix of user, | D | indicate the hidden factor matrix of drug,Indicate the hidden factor matrix of user not Luo Beini crow this
Square of norm,Indicate square of this black norm of the not Luo Beini of the hidden factor matrix of drug;
S5-2, new rating matrix and weight matrix obtain the hidden eigenmatrix of user and the hidden eigenmatrix of drug after decomposing, dividing
Two matrix multiples obtaining after solution obtain user preference prediction matrix, then by user preference prediction matrix and classification correlation model
It merges, it is as follows to obtain final recommended models:
Wherein,It is scoring of the user i to drug j;It is the hidden factor matrix of user and the hidden factor square of drug after updating
The product of battle array vector indicates that user i scores to the prediction of drug j;yi,jIt indicates under class models, user i comments drug j
Point;∝ indicates directly proportional;* multiplication is indicated;
S5-3, according toThe size of score value is ranked up, and the drug generation of k pushes away before then score value being selected to come from big to small
Recommend list.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810724191.4A CN108921670B (en) | 2018-07-04 | 2018-07-04 | Drug transaction recommendation method fusing potential interest, spatio-temporal data and category popularity of user |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810724191.4A CN108921670B (en) | 2018-07-04 | 2018-07-04 | Drug transaction recommendation method fusing potential interest, spatio-temporal data and category popularity of user |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108921670A true CN108921670A (en) | 2018-11-30 |
CN108921670B CN108921670B (en) | 2022-06-14 |
Family
ID=64424469
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810724191.4A Active CN108921670B (en) | 2018-07-04 | 2018-07-04 | Drug transaction recommendation method fusing potential interest, spatio-temporal data and category popularity of user |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108921670B (en) |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110085292A (en) * | 2019-04-28 | 2019-08-02 | 广东技术师范大学 | Drug recommended method, device and computer readable storage medium |
CN110442797A (en) * | 2019-08-19 | 2019-11-12 | 重庆华医康道科技有限公司 | A kind of internet hospital products configuration optimization method |
CN111311324A (en) * | 2020-02-18 | 2020-06-19 | 电子科技大学 | User-commodity preference prediction system and method based on stable neural collaborative filtering |
CN111325419A (en) * | 2018-12-13 | 2020-06-23 | 北京沃东天骏信息技术有限公司 | Method and device for identifying blacklist user |
CN111564201A (en) * | 2020-05-08 | 2020-08-21 | 深圳市万佳安人工智能数据技术有限公司 | Particle swarm optimization-based intelligent prediction method and device for children diet |
CN111815351A (en) * | 2020-05-29 | 2020-10-23 | 杭州览众数据科技有限公司 | Cooperative filtering and association rule-based clothing recommendation method |
CN113221000A (en) * | 2021-05-17 | 2021-08-06 | 上海博亦信息科技有限公司 | Talent data intelligent retrieval and recommendation method |
CN113449210A (en) * | 2021-07-01 | 2021-09-28 | 深圳市数字尾巴科技有限公司 | Personalized recommendation method and device based on space-time characteristics, electronic equipment and storage medium |
CN113569155A (en) * | 2021-07-30 | 2021-10-29 | 西南大学 | Recommendation recall method and system based on improved recurrent neural network algorithm |
CN114881689A (en) * | 2022-04-26 | 2022-08-09 | 驰众信息技术(上海)有限公司 | Building recommendation method and system based on matrix decomposition |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106339502A (en) * | 2016-09-18 | 2017-01-18 | 电子科技大学 | Modeling recommendation method based on user behavior data fragmentation cluster |
CN107463645A (en) * | 2017-07-21 | 2017-12-12 | 雷锤智能科技南京有限公司 | The personalized recommendation system and its recommendation method being oriented to based on user property scoring |
US20180075512A1 (en) * | 2016-09-13 | 2018-03-15 | Adobe Systems Incorporated | Item recommendation techniques |
-
2018
- 2018-07-04 CN CN201810724191.4A patent/CN108921670B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20180075512A1 (en) * | 2016-09-13 | 2018-03-15 | Adobe Systems Incorporated | Item recommendation techniques |
CN106339502A (en) * | 2016-09-18 | 2017-01-18 | 电子科技大学 | Modeling recommendation method based on user behavior data fragmentation cluster |
CN107463645A (en) * | 2017-07-21 | 2017-12-12 | 雷锤智能科技南京有限公司 | The personalized recommendation system and its recommendation method being oriented to based on user property scoring |
Non-Patent Citations (1)
Title |
---|
郁钢等: "基于用户兴趣模型的个性推荐算法", 《智能计算机与应用》 * |
Cited By (16)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111325419A (en) * | 2018-12-13 | 2020-06-23 | 北京沃东天骏信息技术有限公司 | Method and device for identifying blacklist user |
CN110085292A (en) * | 2019-04-28 | 2019-08-02 | 广东技术师范大学 | Drug recommended method, device and computer readable storage medium |
CN110085292B (en) * | 2019-04-28 | 2022-07-26 | 广东技术师范大学 | Medicine recommendation method and device and computer-readable storage medium |
CN110442797B (en) * | 2019-08-19 | 2022-02-08 | 重庆华医康道科技有限公司 | Internet hospital product configuration optimization method |
CN110442797A (en) * | 2019-08-19 | 2019-11-12 | 重庆华医康道科技有限公司 | A kind of internet hospital products configuration optimization method |
CN111311324A (en) * | 2020-02-18 | 2020-06-19 | 电子科技大学 | User-commodity preference prediction system and method based on stable neural collaborative filtering |
CN111311324B (en) * | 2020-02-18 | 2022-05-20 | 电子科技大学 | User-commodity preference prediction system and method based on stable neural collaborative filtering |
CN111564201A (en) * | 2020-05-08 | 2020-08-21 | 深圳市万佳安人工智能数据技术有限公司 | Particle swarm optimization-based intelligent prediction method and device for children diet |
CN111815351A (en) * | 2020-05-29 | 2020-10-23 | 杭州览众数据科技有限公司 | Cooperative filtering and association rule-based clothing recommendation method |
CN113221000A (en) * | 2021-05-17 | 2021-08-06 | 上海博亦信息科技有限公司 | Talent data intelligent retrieval and recommendation method |
CN113221000B (en) * | 2021-05-17 | 2023-02-28 | 上海博亦信息科技有限公司 | Talent data intelligent retrieval and recommendation method |
CN113449210A (en) * | 2021-07-01 | 2021-09-28 | 深圳市数字尾巴科技有限公司 | Personalized recommendation method and device based on space-time characteristics, electronic equipment and storage medium |
CN113449210B (en) * | 2021-07-01 | 2023-01-31 | 深圳市数字尾巴科技有限公司 | Personalized recommendation method and device based on space-time characteristics, electronic equipment and storage medium |
CN113569155A (en) * | 2021-07-30 | 2021-10-29 | 西南大学 | Recommendation recall method and system based on improved recurrent neural network algorithm |
CN113569155B (en) * | 2021-07-30 | 2022-05-03 | 西南大学 | Recommendation recall method and system based on improved recurrent neural network algorithm |
CN114881689A (en) * | 2022-04-26 | 2022-08-09 | 驰众信息技术(上海)有限公司 | Building recommendation method and system based on matrix decomposition |
Also Published As
Publication number | Publication date |
---|---|
CN108921670B (en) | 2022-06-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108921670A (en) | A kind of potential interest of fusion user, the Drug trading recommended method of space-time data and classification popularity | |
Chen et al. | Increasing rates of impulsive online shopping on tourism websites | |
Antéblian et al. | Consumption experience in retail environments: A literature review | |
Cai et al. | Designing not just for pleasure: effects of web site aesthetics on consumer shopping value | |
Urdea et al. | Implementing experiential marketing in the digital age for a more sustainable customer relationship | |
Wolfinbarger et al. | Shopping online for freedom, control, and fun | |
CN108460619A (en) | A kind of fusion shows the Collaborative Recommendation model of implicit feedback | |
CN108920527A (en) | A kind of personalized recommendation method of knowledge based map | |
CN103514239B (en) | A kind of integrated user behavior and the recommendation method and system of item contents | |
CN108228867A (en) | A kind of theme collaborative filtering recommending method based on viewpoint enhancing | |
Zhang et al. | Product features characterization and customers’ preferences prediction based on purchasing data | |
CN103150667A (en) | Individualization recommending method based on mechanical structure | |
CN107015659A (en) | A kind of virtual try-in method of wrist-watch and system | |
Harahap et al. | Effect of EWOM on Revisit Intention: Attitude and Destination Image as Mediation Variables (Study in Pasaman Regency Tourism Destinations) | |
Trivedi et al. | Interpretive structural modelling of website quality factors for repurchase intention in online context | |
Chidambaram et al. | Antecedents of consumers' online apparel purchase intention through Virtual Try On technology: A moderated moderated‐mediation model | |
Fu et al. | Sinicized exploration of sustainable digital fashion: Chinese game players’ intention to purchase traditional costume skins | |
Anshu et al. | Co-creation: interface for online affective experience and repurchase intention | |
Marmat | Influence of aesthetics attributes of brand Web pages on customer brand engagement | |
CN109636530A (en) | Product determines method, apparatus, electronic equipment and computer readable storage medium | |
Sharma | Modeling the role of information on the spread of online shopping | |
Che et al. | A cluster analysis of Lotte young plaza using semantic network analysis method | |
Ashraf et al. | System generated recommendation vs consumer generated recommendation: A differential effect on consumers beliefs and behavior in e-commerce transactions | |
Dang Quan et al. | Can you resist the virtual temptations? Unveiling impulsive buying in metaverse retail | |
Huang et al. | Digital modality richness drives vivid memory experience |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |