CN108898459B - Commodity recommendation method and device - Google Patents
Commodity recommendation method and device Download PDFInfo
- Publication number
- CN108898459B CN108898459B CN201810665009.2A CN201810665009A CN108898459B CN 108898459 B CN108898459 B CN 108898459B CN 201810665009 A CN201810665009 A CN 201810665009A CN 108898459 B CN108898459 B CN 108898459B
- Authority
- CN
- China
- Prior art keywords
- commodities
- commodity
- target
- random walk
- module
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 29
- 238000005295 random walk Methods 0.000 claims abstract description 44
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 34
- 239000013598 vector Substances 0.000 claims description 59
- 238000004364 calculation method Methods 0.000 claims description 29
- 238000010276 construction Methods 0.000 claims description 4
- 238000010586 diagram Methods 0.000 description 16
- 238000012549 training Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 238000001914 filtration Methods 0.000 description 3
- 238000011156 evaluation Methods 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 238000004590 computer program Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
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
Abstract
The embodiment of the invention provides a commodity recommendation method and device, which are applied to the field of computer science. The commodity recommendation method provided by the invention avoids the problems of cold start and over recommendation of hot commodities for new users and new things, so that commodity recommendation is more reasonable and accurate, the commodity purchasing convenience of customers is improved, the revenue of supermarkets is increased, and effective reference is provided for commodity placement of supermarket shelf commodities. The method comprises the following steps: acquiring a historical shopping record; establishing a commodity co-purchasing relationship network according to the historical shopping record; in the commodity co-purchase relationship network, respectively calculating random walk sequences corresponding to all commodities in the co-purchase relationship network by using a random walk algorithm; determining the association degree between the target commodity and other commodities according to the random walk sequence corresponding to the commodity; and determining the commodities needing to be recommended according to the relevance between the target commodity and other commodities. The invention is applied to commodity recommendation.
Description
Technical Field
The invention relates to the field of computer science, in particular to a novel commodity recommendation method and device.
Background
With the popularization of the internet and the increasing maturity of electronic commerce, people increasingly utilize electronic commerce platforms to acquire commodity information and purchase commodities. When people hope to purchase goods, the electronic commerce platform can recommend various goods information to users, for example, recommending goods which the users may be interested in to the users. Through commodity recommendation, the path for searching the required product by the user can be shortened, and the user experience is improved.
The recommended algorithm is one of the more popular and popular algorithms at the present time. The current main recommendation algorithms mainly comprise content-based recommendation, collaborative filtering recommendation and association rule-based recommendation. In the content-based recommendation algorithm, the user's interest is first learned through the user's evaluation features of the object, and then the degree of matching between the user's interest and the item feature attributes is calculated. Finally recommending items with high matching degree to the user; the recommendation algorithm based on collaborative filtering is characterized in that a nearest neighbor algorithm is adopted to calculate a neighbor user which is closest to the favor of a target user, the favor degree of the target user to a specific commodity is predicted according to the weighted value of the nearest neighbor user to commodity evaluation, and finally commodity recommendation is carried out according to the favor degree of the user; the association rule-based recommendation takes the association rule as a basis, the purchased commodities are taken as rule heads, the recommendation object is taken as a rule body, and the co-purchase probability of the two commodities is calculated to recommend the target user.
The above common recommendation algorithms have certain limitations: the recommendation algorithm based on the content requires that the preference of the user can be expressed by the content characteristics, and the extracted characteristics have certain practical significance while ensuring the accuracy and can not be displayed to obtain the judgment condition of other users; the recommendation algorithm core based on collaborative filtering is based on historical data, the problem of 'cold start' is caused to new users and new articles, the recommendation effect depends on the amount and accuracy of data of user historical preference, in most implementations, the user historical preference is stored by using a sparse matrix, and the calculation on the sparse matrix is obvious, and the recommendation algorithm core comprises the problems that the wrong preference of a small number of people possibly has great influence on the recommendation accuracy and the like; the recommendation algorithm based on the association rule has a large calculation amount, and meanwhile, the problems of cold start and sparsity exist, and the problem that hot items are excessively recommended easily occurs. In conclusion, the three algorithms can not reasonably and accurately recommend commodities to the customer, so that the customer can conveniently shop, and the supermarket revenue is promoted.
Disclosure of Invention
In order to overcome the defects in the prior art, the invention aims to solve the technical problems that: the commodity recommendation method and device are provided, reference is provided for commodity recommendation of e-commerce supermarkets or goods placement of supermarket shelves, the purchase convenience degree of customers is improved, and revenues of supermarkets are promoted.
In order to achieve the purpose, the invention adopts the following technical scheme:
in a first aspect, an embodiment of the present application provides a commodity recommendation method, including: acquiring a historical shopping record; establishing a commodity co-purchasing relationship network according to the historical shopping record; the commodity co-purchase relation network comprises N commodities, wherein N is more than or equal to 2; according to the historical shopping records, if two commodities are determined to be purchased simultaneously, the two commodities are connected; the N commodities comprise target commodities; in the commodity co-purchase relationship network, respectively calculating random walk sequences corresponding to the N commodities by using a random walk algorithm; determining the association degree between the target commodity and other commodities according to the random walk sequences corresponding to the N commodities; and determining the commodities needing to be recommended according to the relevance between the target commodity and other commodities.
Optionally, the method further includes: respectively calculating real number vectors of the N commodities by using a Word2vec algorithm according to the random walk sequences corresponding to the N commodities; and calculating cosine similarity between the target commodity and other commodities in the N commodities according to the real number vectors of the N commodities.
Optionally, the goods to be recommended include: and the cosine similarity between the N commodities and the target commodity is greater than a preset threshold value. The predetermined threshold may be better defined for items similar to the target item, the predetermined threshold being set to.
Optionally, the cosine similarity calculation formula is as follows:
where n represents the vector dimension of the real number vector of the commodity,a real number vector representing the target commodity,real number vectors representing the commodities other than the target commodity among the N commodities, if and only if a predetermined threshold is ≦ similarityxAt the time, reserveThe corresponding trade name, the rest is removed.
Optionally, the historical shopping record includes: and acquiring shopping records of different customers when purchasing commodities in a preset time period.
According to the commodity recommendation method provided by the invention, a commodity co-purchase relation network is constructed by acquiring historical shopping records, a random walk algorithm is utilized to obtain a random walk sequence of each commodity, the association degree between a target commodity and other commodities is calculated according to the random walk sequence corresponding to the commodity, the commodity to be recommended is determined according to the association degree, and the commodity with high similarity is recommended to a customer. The commodity recommendation method effectively avoids the problem of 'cold start' of the commodity, enables a new user to obtain reasonable recommendation when purchasing the commodity, solves the problem that hot items are excessively recommended, improves the commodity recommendation accuracy, provides reference for commodity recommendation of E-commerce supermarkets or shelf arrangement of supermarkets, improves the purchasing convenience of customers, and promotes the revenue of supermarkets.
In a second aspect, an embodiment of the present application provides a commodity recommendation device, including: the acquisition module is used for acquiring historical shopping records; the building module is used for building a commodity co-purchasing relationship network after the historical shopping record is obtained by the obtaining module; the commodity co-purchase relation network comprises N commodities, wherein N is more than or equal to 2; according to the historical shopping records, if two commodities are determined to be purchased simultaneously, the two commodities are connected; the N commodities comprise target commodities; the sequence generation module is used for respectively obtaining random walk sequences corresponding to the N commodities in the commodity co-purchase relation network by using a random walk algorithm after the construction module constructs the commodity co-purchase relation network; the determining module is used for determining the association degree between the target commodity and other commodities after the sequence generating module obtains the random walk sequence; and the recommending module is used for determining the commodities needing to be recommended after the determining module calculates the association degree between the target commodity and other commodities.
Optionally, the determining module specifically includes: a vector calculation unit and a similarity calculation unit; the vector calculation unit is used for calculating real number vectors of the N commodities respectively by using a Word2vec algorithm according to the random walk sequences corresponding to the N commodities; and the similarity calculation unit is used for calculating cosine similarities between the target commodity and other commodities in the N commodities after the vector calculation unit respectively obtains real number vectors of the N commodities.
Optionally, the goods to be recommended include: and the cosine similarity between the N commodities and the target commodity is greater than a preset threshold value.
Optionally, the cosine similarity calculation formula of the similarity calculation unit is as follows:
where n represents the vector dimension of the real number vector of the commodity,a real number vector representing the target commodity,real number vectors representing the commodities other than the target commodity among the N commodities, if and only if a predetermined threshold is ≦ similarityxAt the time, reserveThe corresponding trade name, the rest is removed.
Optionally, the historical shopping record acquired by the acquiring module includes: and acquiring shopping records of different customers when purchasing commodities in a preset time period.
Since the technical effect brought by the product recommendation device provided by the second aspect is realized based on the product recommendation method provided by the first aspect, the technical effect brought by the product recommendation device is the same as the technical effect brought by the product recommendation method, and details are not repeated here.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
Fig. 1 is a schematic step diagram of a commodity recommendation method according to an embodiment of the present invention;
fig. 2 is a schematic flow chart of a commodity recommendation method according to an embodiment of the present invention;
FIG. 3 is a schematic diagram illustrating a co-purchasing relationship between products purchased by a customer according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of a structure of a co-purchasing relationship between products purchased by customers according to an embodiment of the present invention;
FIG. 5 is a schematic diagram of a commodity purchasing relationship network construction;
FIG. 6 is a schematic diagram of a structural module of the commodity recommending device according to the present invention;
fig. 7 is a schematic structural diagram of a commodity recommendation device determination module according to the present invention.
Detailed Description
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in detail below. It is to be understood that the described embodiments are merely exemplary of the invention, and not restrictive of the full scope of the invention. All other embodiments, which can be obtained by a person skilled in the art without inventive step based on the embodiments of the present invention, are within the scope of the present invention.
Technical terms used for the embodiments of the present application are described as follows:
the Word2vec algorithm is a software tool for training Word vectors, which is opened by Google, and is used for quickly and effectively mapping each Word in a sentence into a vector with a real value in a k-dimensional space through an optimized training model according to a given corpus, and the core architecture of the Word2vec algorithm comprises CBOW and Skip-gram. CBOW is the inference of the target word from the original sentence; and the Skip-gram is just the opposite, and the original sentence is deduced from the target word.
The relevance degree of the invention is to characterize the relevance degree between two things, and mathematically refers to the degree of similarity of two functions. The similarity is a quantitative standard for measuring the similarity between articles, the articles with the highest similarity are the articles most likely to be purchased together by the customer, and there are many methods for calculating the similarity, and Cosine-based (Cosine-based) similarity calculation and Correlation-based (Correlation-based) similarity calculation are commonly used.
The first embodiment is as follows:
the embodiment of the invention provides a novel commodity recommendation method, which is applied to a scene of recommending commodities and comprises the following steps:
fig. 1 is a schematic diagram illustrating steps of a commodity recommendation method according to an embodiment of the present application. The schematic diagram comprises the steps of obtaining names of commodities purchased by customers, constructing a commodity purchasing relation network, obtaining a random walk sequence of each commodity, a Word2vec algorithm, obtaining a real number vector of each commodity, calculating cosine similarity, and carrying out bundle sale or recommendation to a user on the commodities with high similarity. The step of obtaining the commodity names purchased by the customers means that all commodity names purchased by the customer A, all commodity names purchased by the customer B, all commodity names purchased by the customer C and all commodity names purchased by the customer M are obtained through the supermarket historical data server in a certain period; constructing a commodity co-purchase relationship network refers to forming a commodity co-purchase relationship network graph by using all commodity names purchased by the M customers as points and using connecting lines among commodities appearing in the same shopping list as edges according to the shopping records of the M customers; the step of obtaining the random walk sequence of each commodity refers to that all the random walk sequences of the commodity 1, all the random walk sequences of the commodity 2, all the random walk sequences of the commodity 3 and all the random walk sequences of the commodity N are obtained by using a random walk algorithm according to the constructed commodity purchasing relation network diagram; the Word2vec algorithm is an open software tool for training Word vectors; the real number vector of each commodity is that the real number vector of the commodity 1, the real number vector of the commodity 2, the real number vector of the commodity 3 and the real number vector of the commodity N are calculated by using a Word2vec algorithm according to all random walk sequences of each commodity; the cosine similarity calculation means that cosine values between the target commodity and other commodities in the N commodities are calculated by utilizing a cosine formula according to the obtained real number vector of each commodity; the step of performing bundle sales or recommendation to the user on the commodities with high similarity refers to calculating the commodities with cosine values close to the numerical value 1 between the target commodity and other commodities in the N commodities, and performing bundle sales on the commodities and the target commodity or performing recommendation to the user who purchases the target commodity.
The detailed description is specifically made by taking an example of fig. 2;
s101, obtaining historical shopping records.
For example, the historical shopping record may be obtained by obtaining shopping records of different customers when purchasing commodities within a preset time period. For example, the shopping records of all customers in a supermarket within half a year are extracted.
And S102, constructing a commodity purchasing relation network.
Establishing a commodity co-purchasing relationship network according to the historical shopping record; the commodity co-purchase relation network comprises N commodities, wherein N is more than or equal to 2; according to the historical shopping records, if two commodities are determined to be purchased simultaneously, the two commodities are connected; the N commodities comprise target commodities.
For example, if a first customer purchases products 1, 2, 3, and 7 in the same shopping list, and the four product names are called vertices, two products are connected, that is, the product 1 is connected to the product 2, the product 1 is connected to the product 3, the product 1 is connected to the product 7, the product 2 is connected to the product 3, the product 2 is connected to the product 7, and the product 3 is connected to the product 7, so as to obtain a undirected product co-purchase relationship network diagram, as shown in fig. 3; the second customer purchases the product 3, the product 6 and the product 7 in the same shopping list, and with the three product names as vertices, two products are connected, that is, the product 3 is connected with the product 6, the product 3 is connected with the product 7, and the product 6 is connected with the product 7, so as to obtain a network schematic diagram of undirected product co-purchasing relationship, as shown in fig. 4. Similarly, a third customer purchases the commodities 4 and 6 in the same shopping list, the two commodity names are called vertexes, and every two commodities are connected, namely the commodities 4 and the commodities 6 are connected to obtain a network schematic diagram of undirected commodity co-purchasing relationship; a fourth customer purchases the commodities 2 and 4 in the same shopping list, the two commodity names are called vertexes, and every two commodities are connected, namely the commodities 2 and the commodities 4 are connected to obtain a undirected commodity co-purchase relationship network schematic diagram; the fifth customer purchases the products 4 and 5 in the same shopping list, and the two products are called as vertexes, and two products are connected, that is, the products 4 and the products 5 are connected to obtain the undirected product co-purchase relationship network schematic diagram. (the undirected product co-shopping relationship networks of the third to fifth customers are not shown in the figure because the undirected product co-shopping relationship networks of the first and second customers are generated in the same manner), the commodities purchased by the five customers are subjected to edge set combination, and finally, a co-shopping product relationship network schematic diagram, namely fig. 5, is obtained.
S103, acquiring a random walk sequence.
And respectively calculating random walk sequences corresponding to the N commodities by using a random walk algorithm in the commodity co-purchase relationship network.
Exemplarily, taking fig. 5 as an example, one node in all nodes [ product 1, product 2, product 3, product 4, product 5, product 6, and product 7] in the graph of the same-purchase product relationship network is selected as a starting point of random walk with equal probability, where the probability of each node being selected is 1/7, the starting node of random walk is obtained as product 1 through random selection, and the traversal node sequence VS is (1); selecting a target node at a moderate probability in neighbor nodes [ commodity 2, commodity 3 and commodity 7] by taking a commodity 1 as an initial node, wherein the probability of selecting each neighbor node is 1/3, after random selection, the target node is 2, and a traversal node sequence VS is (1, 2); for each step of random walk, the destination node is a point randomly selected from neighbor nodes of the current node with equal probability, and the destination node is added into the traversal node sequence.
And S104, determining the association degree between the target commodity and other commodities.
And determining the association degree between the target commodity and other commodities according to the random walk sequence corresponding to the N commodities.
Specifically, step S104 may include:
s1041, respectively calculating real number vectors of the N commodities by using a Word2vec algorithm in a natural language processing method according to the obtained random walk sequence;
s1042, calculating cosine similarity between the target commodity and other commodities in the N commodities according to the real number vectors of the N commodities.
Similarity is a quantitative standard for measuring the similarity between articles, the articles with the highest similarity are the articles most likely to be purchased together by a customer, and methods for calculating the similarity are various, and Cosine-based (Cosine-based) similarity calculation and Correlation-based (Correlation-based) similarity calculation are commonly used.
Specifically, the similarity between the commodities is measured by adopting a cosine similarity calculation method, the cosine similarity is also called as cosine similarity, the similarity between two vectors is measured by calculating the cosine value of an included angle, the cosine value range is [ -1,1], the smaller the included angle is, the closer the cosine value is to 1, the directions of the two vectors tend to be consistent, the higher the similarity is, and the cosine similarity between the target commodity and other commodities in the N commodities is calculated:
where n represents the vector dimension of the real number vector of the commodity,a real number vector representing the target commodity,real number vectors representing the commodities other than the target commodity among the N commodities, if and only if a predetermined threshold is ≦ similarityxAt the time, reserveThe corresponding trade name, the rest is removed. Through a large amount of seedsThe verification data shows that when the predetermined threshold value is 0.75, the association between the retained commodity and the target commodity is stronger.
The Word2vec algorithm is a software tool for training Word vectors, which is opened by Google corporation, and according to a given corpus, each Word in a sentence is quickly and effectively mapped into a vector with a real value in a k-dimensional space through an optimized training model, and the core architecture of the Word2vec algorithm comprises CBOW and Skip-gram. CBOW is the inference of the target word from the original sentence; and the Skip-gram is just the opposite, and the original sentence is deduced from the target word. CBOW is more appropriate for small data, whereas Skip-gram performs better in large corpora. The invention adopts a Skip-gram model, which predicts the probability of the occurrence of the context by the current word, and takes the random walk sequence obtained by S103 as a corpus.
And S105, determining the commodities needing to be recommended according to the association degree among the commodities.
Specifically, when determining the degree of association between the target product and the other product in the N products by using the cosine similarity between the target product and the other product, the product to be recommended includes:
and the cosine similarity between the N commodities and the target commodity is greater than a preset threshold value.
Specifically, if and only if ≦ similarityxThat is, when the cosine value between the target product and the other product calculated is 0.75 or more, the product is strongly associated with the target product, and the product can be recommended to the user as a recommended product.
When the number of the commodities with the cosine similarity to the target commodity being greater than the predetermined threshold value is multiple, the multiple commodities can be recommended to the user at the same time, or one or more commodities among the multiple commodities can be selected according to a predetermined rule for recommendation.
And (3) bundling and selling the commodities with high similarity: the online supermarket recommends the commodity with high similarity to the commodity to the user according to the commodity which is added into the shopping cart or purchased by the user, and recommends the network link of the commodity to the user; the physical supermarket arranges the commodities on the goods shelf according to the similarity of the commodities, and places the commodities with high similarity on the adjacent goods shelf. Therefore, the purchase convenience of the user is improved, and the revenue of the supermarket is promoted.
Example two:
the embodiment of the application provides a commodity recommendation device, which is used for executing the commodity recommendation method. Fig. 6 shows a schematic diagram of a possible module structure of the related goods recommending device. The specific commodity recommending apparatus 10 includes: the system comprises an acquisition module 101, a construction module 102, a sequence generation module 103, a determination module 104 and a recommendation module 105. Wherein
The acquisition module 101 is configured to acquire a historical shopping record.
The historical shopping records can be obtained in a mode of obtaining the shopping records of different customers when the customers purchase commodities within a preset time period. For example, the shopping records of all customers in a supermarket within half a year are extracted.
The building module 102 is configured to build a commodity co-purchasing relationship network after the obtaining module obtains the historical shopping record; the commodity co-purchase relation network comprises N commodities, wherein N is more than or equal to 2; according to the historical shopping records, if two commodities are determined to be purchased simultaneously, the two commodities are connected; the N commodities comprise target commodities.
The sequence generating module 103 is configured to, after the building module builds the commodity co-purchasing relationship network, respectively obtain the random walk sequences corresponding to the N commodities in the commodity co-purchasing relationship network by using a random walk algorithm.
A determining module 104, configured to determine, after the sequence generating module obtains the random walk sequence, a degree of association between the target product and another product.
As shown in fig. 7, the calculation module 104 may specifically include: vector calculation section 1041, similarity calculation section 1042. Wherein:
and a vector calculation unit 1041, configured to calculate real vectors of the N commodities, respectively, by using a Word2vec algorithm according to the random walk sequences corresponding to the N commodities.
The similarity calculation unit 1042 is configured to calculate cosine similarities between the target commodity and other commodities in the N commodities after the vector calculation unit obtains real vectors of the N commodities, respectively.
The cosine similarity formula is as follows:
where n represents the vector dimension of the real number vector of the commodity,a real number vector representing the target commodity,and real number vectors representing the commodities except the target commodity in the N commodities.
And the recommending module 105 is configured to determine the commodity to be recommended after the determining module determines the association degree between the target commodity and the other commodities in the N commodities.
And the cosine similarity between the N commodities and the target commodity is greater than a preset threshold value.
Specifically, if and only if ≦ similarityxThat is, when the cosine value between the target product and the other product calculated is 0.75 or more, the product is strongly associated with the target product, and the product can be recommended to the user as a recommended product.
When the number of the commodities with the cosine similarity to the target commodity being greater than the predetermined threshold value is multiple, the multiple commodities can be recommended to the user at the same time, or one or more commodities among the multiple commodities can be selected according to a predetermined rule for recommendation.
The integrated module can be realized in a hardware mode, and can also be realized in a software functional module mode. It should be noted that, in the embodiment of the present application, the division of the module is schematic, and is only one logic function division, and there may be another division manner in actual implementation. For example, the functional blocks may be divided for each function, or two or more functions may be integrated into one processing block.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by a computer program, which can be stored in a computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. The storage medium may be a magnetic disk, an optical disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), or the like.
The above description is only for the specific embodiments of the present invention, but the scope of the present invention is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present invention are included in the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the appended claims.
Claims (8)
1. A method for recommending a commodity, comprising the steps of:
acquiring a historical shopping record;
establishing a commodity co-purchasing relationship network according to the historical shopping record; the commodity co-purchase relation network comprises N commodities, wherein N is more than or equal to 2; according to the historical shopping records, if two commodities are determined to be purchased simultaneously, the two commodities are connected; the N commodities comprise target commodities;
in the commodity co-purchase relationship network, respectively calculating random walk sequences corresponding to the N commodities by using a random walk algorithm;
determining the association degree between the target commodity and other commodities according to the random walk sequences corresponding to the N commodities;
determining commodities needing to be recommended according to the relevance between the target commodity and other commodities;
determining the association degree between the target commodity and other commodities according to the random walk sequence corresponding to the N commodities, including:
respectively calculating real number vectors of the N commodities by using a Word2vec algorithm according to the random walk sequences corresponding to the N commodities;
and calculating cosine similarity between the target commodity and other commodities in the N commodities according to the real number vectors of the N commodities.
2. The article recommendation method according to claim 1,
the commodity needing to be recommended comprises: and the cosine similarity between the N commodities and the target commodity is greater than a preset threshold value.
4. The commodity recommendation method according to any one of claims 1 to 3, wherein the historical shopping record includes: and acquiring shopping records of different customers when purchasing commodities in a preset time period.
5. An article recommendation device, comprising:
the acquisition module is used for acquiring historical shopping records;
the building module is used for building a commodity co-purchasing relationship network after the historical shopping record is obtained by the obtaining module; the commodity co-purchase relation network comprises N commodities, wherein N is more than or equal to 2; according to the historical shopping records, if two commodities are determined to be purchased simultaneously, the two commodities are connected; the N commodities comprise target commodities;
the sequence generation module is used for respectively obtaining random walk sequences corresponding to the N commodities in the commodity co-purchase relation network by using a random walk algorithm after the construction module constructs the commodity co-purchase relation network;
the determining module is used for determining the association degree between the target commodity and other commodities after the sequence generating module obtains the random walk sequence;
the recommending module is used for determining the commodities needing to be recommended after the determining module determines the association degree between the target commodity and other commodities;
the determining module specifically includes: a vector calculation unit and a similarity calculation unit;
the vector calculation unit is used for calculating real number vectors of the N commodities respectively by using a Word2vec algorithm according to the random walk sequences corresponding to the N commodities;
and the similarity calculation unit is used for calculating cosine similarities between the target commodity and other commodities in the N commodities after the vector calculation unit respectively obtains real number vectors of the N commodities.
6. The article recommending device according to claim 5,
the commodity needing to be recommended comprises: and the cosine similarity between the N commodities and the target commodity is greater than a preset threshold value.
7. The item recommendation device of claim 5, wherein the cosine similarity calculation formula of the similarity unit is:
8. The item recommendation device according to any one of claims 5 to 6, wherein the historical shopping records acquired by the acquisition module include: and acquiring shopping records of different customers when purchasing commodities in a preset time period.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810665009.2A CN108898459B (en) | 2018-06-25 | 2018-06-25 | Commodity recommendation method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810665009.2A CN108898459B (en) | 2018-06-25 | 2018-06-25 | Commodity recommendation method and device |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108898459A CN108898459A (en) | 2018-11-27 |
CN108898459B true CN108898459B (en) | 2020-12-01 |
Family
ID=64346321
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810665009.2A Active CN108898459B (en) | 2018-06-25 | 2018-06-25 | Commodity recommendation method and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108898459B (en) |
Families Citing this family (23)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111523918B (en) * | 2019-02-02 | 2023-09-19 | 北京极智嘉科技股份有限公司 | Commodity clustering method, device, equipment and storage medium |
CN109885758B (en) * | 2019-01-16 | 2022-07-26 | 西北工业大学 | Random walk recommendation method based on bipartite graph |
CN110084670B (en) * | 2019-04-15 | 2022-03-25 | 东北大学 | Shelf commodity combination recommendation method based on LDA-MLP |
CN110210900B (en) * | 2019-05-23 | 2021-05-25 | 中国银行股份有限公司 | Reference product determination method, device and equipment based on transaction relevance |
CN110210905A (en) * | 2019-05-31 | 2019-09-06 | 拉扎斯网络科技(上海)有限公司 | Characteristic similarity calculation method, device, electronic equipment and computer storage medium |
CN112100221B (en) * | 2019-06-17 | 2024-02-13 | 深圳市雅阅科技有限公司 | Information recommendation method and device, recommendation server and storage medium |
CN110458637A (en) * | 2019-06-19 | 2019-11-15 | 中国平安财产保险股份有限公司 | Product method for pushing and its relevant device neural network based |
CN110264277B (en) * | 2019-06-25 | 2022-08-16 | 杭州网易再顾科技有限公司 | Data processing method and device executed by computing equipment, medium and computing equipment |
CN110782287A (en) * | 2019-10-25 | 2020-02-11 | 北京沃东天骏信息技术有限公司 | Entity similarity calculation method and device, article recommendation system, medium and equipment |
CN111178951B (en) * | 2019-12-20 | 2023-05-16 | 深圳大学 | Commodity recommendation method and device |
CN111177543A (en) * | 2019-12-23 | 2020-05-19 | 杭州火小二科技有限公司 | Object recommendation method and device, storage medium and electronic device |
CN111292164B (en) * | 2020-01-21 | 2023-08-25 | 上海风秩科技有限公司 | Commodity recommendation method and device, electronic equipment and readable storage medium |
CN111415216B (en) * | 2020-02-11 | 2023-11-07 | 广州探途网络技术有限公司 | Commodity recommendation method, commodity recommendation device, server and storage medium |
CN111340566B (en) * | 2020-03-23 | 2023-12-08 | 京东科技控股股份有限公司 | Commodity classification method and device, electronic equipment and storage medium |
CN111882404A (en) * | 2020-08-06 | 2020-11-03 | 上海风秩科技有限公司 | Association push processing method and device |
CN113763014A (en) * | 2021-01-05 | 2021-12-07 | 北京沃东天骏信息技术有限公司 | Article co-occurrence relation determining method and device and judgment model obtaining method and device |
CN112669128B (en) * | 2021-01-28 | 2022-03-22 | 天元大数据信用管理有限公司 | Real-time clustering recommendation method, device and medium |
CN113781152A (en) * | 2021-02-08 | 2021-12-10 | 北京沃东天骏信息技术有限公司 | Object recommendation method and device |
CN112989030A (en) * | 2021-03-15 | 2021-06-18 | 中国美术学院 | Visualization method and system for similar product relationship |
CN112925993B (en) * | 2021-04-08 | 2022-02-08 | 国网电子商务有限公司 | Collaborative filtering recommendation method and device, storage medium and equipment |
CN114169975A (en) * | 2021-12-17 | 2022-03-11 | 福州大学 | Shopping network commodity recommendation method and system based on random walk heterogeneous attention |
CN115544242B (en) * | 2022-12-01 | 2023-04-18 | 深圳市智加云栖科技有限公司 | Big data-based similar commodity model selection recommendation method |
CN116861323B (en) * | 2023-07-24 | 2024-02-23 | 深圳丰享信息技术有限公司 | Method and device for solving long tail effect in recommendation |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101206751A (en) * | 2007-12-25 | 2008-06-25 | 北京科文书业信息技术有限公司 | Customer recommendation system based on data digging and method thereof |
CN105427125A (en) * | 2015-10-29 | 2016-03-23 | 电子科技大学 | Goods clustering method based on goods network connection graph |
CN107730336A (en) * | 2016-08-12 | 2018-02-23 | 苏宁云商集团股份有限公司 | Commodity method for pushing and device in a kind of online transaction |
Family Cites Families (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101251859A (en) * | 2008-03-17 | 2008-08-27 | 北京科文书业信息技术有限公司 | Commercial articles clustering system and method based on shopping behaviors |
CN103208073B (en) * | 2012-01-17 | 2016-09-28 | 阿里巴巴集团控股有限公司 | Obtain Recommendations information and the method for merchandise news, device are provided |
CN103700005B (en) * | 2013-12-17 | 2016-08-31 | 南京信息工程大学 | A kind of correlation rule based on the many minimum supports of self adaptation recommends method |
KR101612482B1 (en) * | 2015-04-06 | 2016-04-14 | 주식회사 포워드벤처스 | System and method for providing shopping service |
CN107248095A (en) * | 2017-04-14 | 2017-10-13 | 北京小度信息科技有限公司 | Recommend method and device |
-
2018
- 2018-06-25 CN CN201810665009.2A patent/CN108898459B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101206751A (en) * | 2007-12-25 | 2008-06-25 | 北京科文书业信息技术有限公司 | Customer recommendation system based on data digging and method thereof |
CN105427125A (en) * | 2015-10-29 | 2016-03-23 | 电子科技大学 | Goods clustering method based on goods network connection graph |
CN107730336A (en) * | 2016-08-12 | 2018-02-23 | 苏宁云商集团股份有限公司 | Commodity method for pushing and device in a kind of online transaction |
Non-Patent Citations (1)
Title |
---|
"基于用户购买记录的改进协同过滤推荐";何有世 等;《计算机工程与设计》;20140916(第09期);第3091-3094页第1节 * |
Also Published As
Publication number | Publication date |
---|---|
CN108898459A (en) | 2018-11-27 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108898459B (en) | Commodity recommendation method and device | |
CN111369318B (en) | Recommendation method and system based on commodity knowledge graph feature learning | |
JP7105700B2 (en) | Time-division recommendation method and apparatus for service target | |
US11315164B2 (en) | Complementary product recommendation systems | |
US9223900B2 (en) | Machine optimization devices, methods, and systems | |
CN107464132B (en) | Similar user mining method and device and electronic equipment | |
CN111292164B (en) | Commodity recommendation method and device, electronic equipment and readable storage medium | |
WO2009094672A2 (en) | Belief propagation for generalized matching | |
JP2002279279A (en) | Commodity recommendation system, commodity recommendation method and commodity recommendation program | |
CN110807669B (en) | Cross-platform user information management method and device | |
CN110766428A (en) | Data value evaluation system and method | |
CN113268656A (en) | User recommendation method and device, electronic equipment and computer storage medium | |
CN110866191A (en) | Recommendation recall method, apparatus and storage medium | |
KR20210032691A (en) | Method and apparatus of recommending goods based on network | |
CN110362594A (en) | A kind of information processing method and system | |
Prasetyo | Searching cheapest product on three different e-commerce using k-means algorithm | |
CN112231548A (en) | User login behavior analysis method, device and system and storage medium | |
CN110889748B (en) | Store platform product recommendation method, store platform product recommendation device, computer equipment and storage medium | |
CN111340566B (en) | Commodity classification method and device, electronic equipment and storage medium | |
CN110020918B (en) | Recommendation information generation method and system | |
CN110781399A (en) | Cross-platform information pushing method and device | |
CN110851708A (en) | Negative sample extraction method and device, computer equipment and storage medium | |
CN115271884A (en) | Multi-source data-based commodity selection method and device and electronic equipment | |
RU2480828C1 (en) | Method of predicting target value of events based on unlimited number of characteristics | |
CN110019154A (en) | Data processing, data quality accessment, recommended products determine method and relevant device |
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 |