User shopping behavior identification method based on big data, storage device and mobile terminal
Technical Field
The invention relates to the field of big data analysis, in particular to a user shopping behavior identification method based on big data.
Background
With the rapid development of the mobile internet and the massive popularization of online shopping, the commodity information is explosively increased, and the problems brought by massive information are as follows:
the information utilization rate is low, and users do not have enough patience and time to browse the commodities arranged behind the users in the face of massive commodities; the timeliness is poor, and when the information is updated instantly, the user cannot obtain interested commodities in the first time; the operation difficulty is high, the interested commodities need to be found, a large number of conditions are often set and ordered, and the user is required to have considerable technical knowledge; how does the user select? For a merchant, how do users be offered items of interest?
In view of the above problems, there is an urgent need to identify the shopping behavior of the user, analyze the user's preference and taste, and recommend the goods according to the user's preference and taste, and there are some recommendation algorithms in the prior art to recommend the goods according to the user's preference and taste, but these techniques have some disadvantages: the method is high in theority, and based on a scoring mechanism of a user, quantification of shopping behaviors and shopping scenes is lacked; the real-time performance is poor, and the quick response cannot be realized through a large amount of calculation and analysis; the impact of time on user preference is not taken into account.
Therefore, the prior art needs to be further improved, and a user shopping behavior identification method based on big data, which is simple in theoretical operation and good in real-time performance, is developed.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provides a user shopping behavior identification method based on big data, which is simple in theoretical operation and good in real-time performance.
In order to achieve the above object, the present invention is realized by:
a big data-based user shopping behavior recognition method is suitable for being executed in computer equipment and comprises the following steps:
preparing basic data: establishing a label of the commodity, wherein the plurality of dimensions describe the commodity; establishing consumer group labels, reflecting user characteristics in multiple dimensions, and storing the established labels of commodities and the consumer group labels in a database;
collecting user behaviors: the method comprises the steps that behavior data of browsing, clicking, collecting, forwarding and purchasing when a user opens a webpage are collected through a collection module;
and (3) quantifying user behaviors: classifying and summarizing the collected behavior data, and establishing a functional relation among users, commodity labels and behavior attention degrees of the users on certain commodity labels, wherein the formula is as follows:
i: representing a user;
j: a label representing the article;
p: the type of the operation is shown, and n types of operations are total;
t (i, j): indicating the operation times of the user i on the commodity label j;
t (i, j) p: representing the number of times that the user i performs p operation on the commodity label j;
and (Wp): representing the weight of p operation, wherein the larger the participation of the user is, the larger the weight is;
sp: representing a scene coefficient when p operation occurs, wherein the larger the coefficient is, the lower the interference of the scene to the user behavior is;
v (i, j): the quantitative value of the behavior attention degree of the user i on the commodity label j is obtained;
establishing a user preference matrix: according to the functional relationship among the user, the commodity label and the action attention degree of the user on a certain commodity label, establishing a user preference function:
K(i,j)old: representing the preference of the user i to the history of the commodity label j;
v (i, j): the quantitative value of the behavior attention degree of the user i on the commodity label j is operated at this time;
Rj: the superposition coefficient of the commodity label is related to the consumption frequency of the commodity;
K(i,j)new: showing the preference of the user i to the commodity label j after the operation occurs;
forming a user preference matrix by a plurality of user preference degrees;
establishing user similarity: forming user similarity according to the user preference matrix, classifying the users according to the user similarity, and creating a new crowd label based on the recommendation of the users and the similarity;
and (4) displaying the commodities: and displaying the commodities with the user preference degree higher than 60% and the user similarity degree higher than 60% to the user according to the user preference matrix and the user similarity degree.
Preferably, the label of the article comprises at least a brand, a category, a specification, a price, a production date, a shelf life and a promotional means; the consumer population label includes at least gender, age, and consumption ability.
Preferably, the acquisition module collects information by embedding events in an App and a Web page.
Preferably, the commodities displayed to the user in the commodity display step are displayed in the order of high preference and low similarity.
A storage device having stored therein a plurality of instructions adapted to be loaded by a processor and to perform the operations of:
preparing basic data: establishing a label of the commodity, wherein the plurality of dimensions describe the commodity; establishing consumer group labels, reflecting user characteristics in multiple dimensions, and storing the established labels of commodities and the consumer group labels in a database;
collecting user behaviors: the method comprises the steps that behavior data of browsing, clicking, collecting, forwarding and purchasing when a user opens a webpage are collected through a collection module;
and (3) quantifying user behaviors: classifying and summarizing the collected behavior data, and establishing a functional relation among users, commodity labels and behavior attention degrees of the users on certain commodity labels, wherein the formula is as follows:
i: representing a user;
j: a label representing the article;
p: the type of the operation is shown, and n types of operations are total;
t (i, j): indicating the operation times of the user i on the commodity label j;
t (i, j) p: representing the number of times that the user i performs p operation on the commodity label j;
and (Wp): representing the weight of p operation, wherein the larger the participation of the user is, the larger the weight is;
sp: representing a scene coefficient when p operation occurs, wherein the larger the coefficient is, the lower the interference of the scene to the user behavior is;
v (i, j): the quantitative value of the behavior attention degree of the user i on the commodity label j is obtained;
establishing a user preference matrix: according to the functional relationship among the user, the commodity label and the action attention degree of the user on a certain commodity label, establishing a user preference function:
K(i,j)old: representing the preference of the user i to the history of the commodity label j;
v (i, j): the quantitative value of the behavior attention degree of the user i on the commodity label j is operated at this time;
Rj: the superposition coefficient of the commodity label is related to the consumption frequency of the commodity;
K(i,j)new: showing the preference of the user i to the commodity label j after the operation occurs;
forming a user preference matrix by a plurality of user preference degrees;
establishing user similarity: forming user similarity according to the user preference matrix, classifying the users according to the user similarity, and creating a new crowd label based on the recommendation of the users and the similarity;
and (4) displaying the commodities: and displaying the commodities with the user preference degree higher than 60% and the user similarity degree higher than 60% to the user according to the user preference matrix and the user similarity degree.
Preferably, the label of the article comprises at least a brand, a category, a specification, a price, a production date, a shelf life and a promotional means; the consumer population label includes at least gender, age, and consumption ability.
Preferably, the acquisition module collects information by embedding events in App and Web pages;
and displaying the commodities displayed to the user in the commodity displaying step according to the sequence of the preference degree and the similarity degree from high to low.
A mobile terminal comprising a processor adapted to implement instructions; comprising a memory adapted to store a plurality of instructions, the instructions adapted to be loaded by the processor and to perform the following operations:
preparing basic data: establishing a label of the commodity, wherein the plurality of dimensions describe the commodity; establishing consumer group labels, reflecting user characteristics in multiple dimensions, and storing the established labels of commodities and the consumer group labels in a database;
collecting user behaviors: the method comprises the steps that behavior data of browsing, clicking, collecting, forwarding and purchasing when a user opens a webpage are collected through a collection module;
and (3) quantifying user behaviors: classifying and summarizing the collected behavior data, and establishing a functional relation among users, commodity labels and behavior attention degrees of the users on certain commodity labels, wherein the formula is as follows:
i: representing a user;
j: a label representing the article;
p: the type of the operation is shown, and n types of operations are total;
t (i, j): indicating the operation times of the user i on the commodity label j;
t (i, j) p: representing the number of times that the user i performs p operation on the commodity label j;
and (Wp): representing the weight of p operation, wherein the larger the participation of the user is, the larger the weight is;
sp: representing a scene coefficient when p operation occurs, wherein the larger the coefficient is, the lower the interference of the scene to the user behavior is;
v (i, j): the quantitative value of the behavior attention degree of the user i on the commodity label j is obtained;
establishing a user preference matrix: according to the functional relationship among the user, the commodity label and the action attention degree of the user on a certain commodity label, establishing a user preference function:
K(i,j)old: representing the preference of the user i to the history of the commodity label j;
v (i, j): the quantitative value of the behavior attention degree of the user i on the commodity label j is operated at this time;
Rj: the superposition coefficient of the commodity label is related to the consumption frequency of the commodity;
K(i,j)new: showing the preference of the user i to the commodity label j after the operation occurs;
forming a user preference matrix by a plurality of user preference degrees;
establishing user similarity: forming user similarity according to the user preference matrix, classifying the users according to the user similarity, and creating a new crowd label based on the recommendation of the users and the similarity;
and (4) displaying the commodities: and displaying the commodities with the user preference degree higher than 60% and the user similarity degree higher than 60% to the user according to the user preference matrix and the user similarity degree.
Preferably, the label of the article comprises at least a brand, a category, a specification, a price, a production date, a shelf life and a promotional means; the consumer population label includes at least gender, age, and consumption ability.
Preferably, the acquisition module collects information by embedding events in App and Web pages;
and displaying the commodities displayed to the user in the commodity displaying step according to the sequence of the preference degree and the similarity degree from high to low.
The invention has the beneficial effects that: the invention carries out quantitative processing on the user behavior, establishes the user preference and similarity function, has simple logic theory and strong operability, considers the purchasing scene and the user participation degree in the evaluation of the preference and the similarity, has good real-time performance and more accurate analysis result.
Drawings
FIG. 1 is a block diagram of the operation flow of a big data-based user shopping behavior recognition method according to the present invention.
Detailed Description
The following further describes embodiments of the present invention with reference to the drawings.
As shown in fig. 1, a big data-based user shopping behavior recognition method is suitable for being executed in a computer device, and includes the following steps:
preparing basic data: establishing a label of a commodity, describing the commodity in multiple dimensions, wherein the label of the commodity at least comprises a brand, a class, a specification, a price, a production date, a quality guarantee period, a sales promotion mode and the like, and the sales promotion mode of the commodity is discounted, one-to-one buying, double points or WeChat red envelope reward and the like; establishing consumer group labels, wherein a plurality of dimensions represent user characteristics, and the consumer group labels at least comprise gender, age, consumption capacity and the like and are used for data analysis as the basis of purchasing and promotion; storing the established labels of the commodities and the labels of the consumer groups in a database;
collecting user behaviors: the behavior data of browsing, clicking, collecting, forwarding and purchasing when a user opens a webpage are collected through a collection module, and before data collection, event embedding is carried out in App and Web pages (embedding is a common data collection method for website analysis, and codes are implanted in the App and the Web pages to realize tracking of series behaviors of the user on each interface of a platform) to collect information;
and (3) quantifying user behaviors: classifying and summarizing the collected behavior data, and establishing a functional relation among users, commodity labels and behavior attention degrees of the users on certain commodity labels, wherein the formula is as follows:
i: representing a user;
j: a label representing the article;
p: the type of the operation is shown, and n types of operations are total;
t (i, j): indicating the operation times of the user i on the commodity label j;
t (i, j) p: representing the number of times that the user i performs p operation on the commodity label j;
and (Wp): the weight of the p operation is represented, the larger the participation of the user is, the larger the weight is, and the participation of the user is related to the self-board click and the passive click of the user;
sp: the method comprises the steps that a scene coefficient when p operation occurs is shown, the larger the coefficient is, the lower the interference of the scene on user behaviors is shown, the scene is a scene where no promotion activity exists at ordinary times, the user behaviors are a common scene, when promotion activities are realized, purchasing behaviors of users are performed under the scene of the promotion activities, and Sps of people are completely different;
v (i, j): the quantitative value of the behavior attention degree of the user i on the commodity label j is obtained;
the formula comprises the participation weight of the user and the scene coefficient during operation, so that the obtained result is more accurate;
establishing a user preference matrix: according to the functional relationship among the user, the commodity label and the action attention degree of the user on a certain commodity label, establishing a user preference function:
K(i,j)old: representing the preference of the user i to the history of the commodity label j;
v (i, j): the quantitative value of the behavior attention degree of the user i on the commodity label j is operated at this time;
Rj: the superposition coefficient of the commodity label is related to the consumption frequency of the commodity;
K(i,j)new: showing the preference of the user i to the commodity label j after the operation occurs;
the plurality of user preferences form a user preference matrix, which is a user preference matrix as shown in the following table.
Establishing user similarity: and forming user similarity according to the user preference matrix, classifying the users according to the user similarity, creating a new crowd label based on the recommendation of the users and the similarity, dividing the crowd with the similarity in a close interval into a crowd class, and giving a corresponding label to each crowd class according to the similarity.
And (4) displaying the commodities: and displaying the commodities with the user preference degree higher than 60% and the user similarity degree higher than 60% to the user according to the user preference matrix and the user similarity degree, wherein the commodities displayed to the user in the commodity displaying step are displayed according to the sequence of the preference degree and the similarity degree from high to low.
A storage device having stored therein a plurality of instructions adapted to be loaded by a processor and to perform the operations of:
preparing basic data: establishing a label of a commodity, describing the commodity in multiple dimensions, wherein the label of the commodity at least comprises a brand, a class, a specification, a price, a production date, a quality guarantee period, a sales promotion mode and the like, and the sales promotion mode of the commodity is discounted, one-to-one buying, double points or WeChat red envelope reward and the like; establishing consumer group labels, wherein a plurality of dimensions represent user characteristics, and the consumer group labels at least comprise gender, age, consumption capacity and the like and are used for data analysis as the basis of purchasing and promotion; storing the established labels of the commodities and the labels of the consumer groups in a database;
collecting user behaviors: the behavior data of browsing, clicking, collecting, forwarding and purchasing when a user opens a webpage are collected through a collection module, and before data collection, time limit is carried out on event embedding points in App and Web pages for collecting information;
and (3) quantifying user behaviors: classifying and summarizing the collected behavior data, and establishing a functional relation among users, commodity labels and behavior attention degrees of the users on certain commodity labels, wherein the formula is as follows:
i: representing a user;
j: a label representing the article;
p: the type of the operation is shown, and n types of operations are total;
t (i, j): indicating the operation times of the user i on the commodity label j;
t (i, j) p: representing the number of times that the user i performs p operation on the commodity label j;
and (Wp): the weight of the p operation is represented, the larger the participation of the user is, the larger the weight is, and the participation of the user is related to the self-board click and the passive click of the user;
sp: the method comprises the steps that a scene coefficient when p operation occurs is shown, the larger the coefficient is, the lower the interference of the scene on user behaviors is shown, the scene is a scene where no promotion activity exists at ordinary times, the user behaviors are a common scene, when promotion activities are realized, purchasing behaviors of users are performed under the scene of the promotion activities, and Sps of people are completely different;
v (i, j): the quantitative value of the behavior attention degree of the user i on the commodity label j is obtained;
the formula comprises the participation weight of the user and the scene coefficient during operation, so that the obtained result is more accurate;
establishing a user preference matrix: according to the functional relationship among the user, the commodity label and the action attention degree of the user on a certain commodity label, establishing a user preference function:
K(i,j)old: representing the preference of the user i to the history of the commodity label j;
v (i, j): the quantitative value of the behavior attention degree of the user i on the commodity label j is operated at this time;
Rj: the superposition coefficient of the commodity label is related to the consumption frequency of the commodity;
K(i,j)new: showing the preference of the user i to the commodity label j after the operation occurs;
the plurality of user preferences form a user preference matrix, which is a user preference matrix as shown in the following table.
|
User 1
|
User 2
|
User 3
|
……
|
User i
|
Merchandise tag 1
|
K(1,1)
|
|
|
|
|
Goods label 2
|
K(1,2)
|
|
|
|
|
Goods label 3
|
|
|
|
|
|
……
|
|
|
|
|
|
Commodity label j
|
|
|
|
|
K(i,j) |
Establishing user similarity: and forming user similarity according to the user preference matrix, classifying the users according to the user similarity, creating a new crowd label based on the recommendation of the users and the similarity, dividing the crowd with the similarity in a close interval into a crowd class, and giving a corresponding label to each crowd class according to the similarity.
And (4) displaying the commodities: and displaying the commodities with the user preference degree higher than 60% and the user similarity degree higher than 60% to the user according to the user preference matrix and the user similarity degree, wherein the commodities displayed to the user in the commodity displaying step are displayed according to the sequence of the preference degree and the similarity degree from high to low.
A mobile terminal comprising a processor adapted to implement instructions; comprising a memory adapted to store a plurality of instructions, the instructions adapted to be loaded by the processor and to perform the following operations:
preparing basic data: establishing a label of a commodity, describing the commodity in multiple dimensions, wherein the label of the commodity at least comprises a brand, a class, a specification, a price, a production date, a quality guarantee period, a sales promotion mode and the like, and the sales promotion mode of the commodity is discounted, one-to-one buying, double points or WeChat red envelope reward and the like; establishing consumer group labels, wherein a plurality of dimensions represent user characteristics, and the consumer group labels at least comprise gender, age, consumption capacity and the like and are used for data analysis as the basis of purchasing and promotion; storing the established labels of the commodities and the labels of the consumer groups in a database;
collecting user behaviors: the behavior data of browsing, clicking, collecting, forwarding and purchasing when a user opens a webpage are collected through a collection module, and before data collection, time limit is carried out on event embedding points in App and Web pages for collecting information;
and (3) quantifying user behaviors: classifying and summarizing the collected behavior data, and establishing a functional relation among users, commodity labels and behavior attention degrees of the users on certain commodity labels, wherein the formula is as follows:
i: representing a user;
j: a label representing the article;
p: the type of the operation is shown, and n types of operations are total;
t (i, j): indicating the operation times of the user i on the commodity label j;
t (i, j) p: representing the number of times that the user i performs p operation on the commodity label j;
and (Wp): the weight of the p operation is represented, the larger the participation of the user is, the larger the weight is, and the participation of the user is related to the self-board click and the passive click of the user;
sp: the method comprises the steps that a scene coefficient when p operation occurs is shown, the larger the coefficient is, the lower the interference of the scene on user behaviors is shown, the scene is a scene where no promotion activity exists at ordinary times, the user behaviors are a common scene, when promotion activities are realized, purchasing behaviors of users are performed under the scene of the promotion activities, and Sps of people are completely different;
v (i, j): the quantitative value of the behavior attention degree of the user i on the commodity label j is obtained;
the formula comprises the participation weight of the user and the scene coefficient during operation, so that the obtained result is more accurate;
establishing a user preference matrix: according to the functional relationship among the user, the commodity label and the action attention degree of the user on a certain commodity label, establishing a user preference function:
K(i,j)old: representing the preference of the user i to the history of the commodity label j;
v (i, j): the quantitative value of the behavior attention degree of the user i on the commodity label j is operated at this time;
Rj: the superposition coefficient of the commodity label is related to the consumption frequency of the commodity;
K(i,j)new: showing the preference of the user i to the commodity label j after the operation occurs;
the plurality of user preferences form a user preference matrix, which is a user preference matrix as shown in the following table.
|
User 1
|
User 2
|
User 3
|
……
|
User i
|
Merchandise tag 1
|
K(1,1)
|
|
|
|
|
Goods label 2
|
K(1,2)
|
|
|
|
|
Goods label 3
|
|
|
|
|
|
……
|
|
|
|
|
|
Commodity label j
|
|
|
|
|
K(i,j) |
Establishing user similarity: and forming user similarity according to the user preference matrix, classifying the users according to the user similarity, creating a new crowd label based on the recommendation of the users and the similarity, dividing the crowd with the similarity in a close interval into a crowd class, and giving a corresponding label to each crowd class according to the similarity.
And (4) displaying the commodities: and displaying the commodities with the user preference degree higher than 60% and the user similarity degree higher than 60% to the user according to the user preference matrix and the user similarity degree, wherein the commodities displayed to the user in the commodity displaying step are displayed according to the sequence of the preference degree and the similarity degree from high to low.
Variations and modifications to the above-described embodiments may occur to those skilled in the art, which fall within the scope and spirit of the above description. Therefore, the present invention is not limited to the specific embodiments disclosed and described above, and some modifications and variations of the present invention should fall within the scope of the claims of the present invention. Furthermore, although specific terms are employed herein, they are used in a generic and descriptive sense only and not for purposes of limitation.