CN117172832A - Intelligent recommendation method for collagen peptide health products based on machine learning - Google Patents

Abstract

The invention relates to the technical field of electric digital data processing, in particular to an intelligent collagen peptide health care product recommendation method based on machine learning, which comprises the following steps: acquiring user information corresponding to each historical user in a user to be recommended and a historical user set; forming a target user group by using target users with the same occupation in all target users; analyzing and processing the working intensity of the occupation of each target user group; determining a physical strain index corresponding to each target user; determining the usage dissimilarity of the health products between every two target users; clustering all target users according to the body strain index, the professional work intensity and the dissimilarity of health care product use; and recommending the collagen peptide health products to the user to be recommended according to the user information corresponding to the target user in the target cluster to which the user to be recommended belongs. According to the invention, through data processing on all user information, the accuracy of recommending the collagen peptide health care products to the user is improved.

Description

Intelligent recommendation method for collagen peptide health products based on machine learning

Technical Field

The invention relates to the technical field of electric digital data processing, in particular to an intelligent recommendation method for collagen peptide health products based on machine learning.

Background

The peptide health product is a skin care health product. When a human body constructs cells by oneself, collagen peptide is often required to be used as a raw material, and researches show that collagen peptide below 1000 daltons can be absorbed by the human body with high efficiency, so that the human body can use exogenous collagen peptide health care products to supplement the collagen peptide. Because of the variety of the collagen peptide health products, in order to facilitate the selection of the collagen peptide health products by users, the recommendation of the collagen peptide health products by users can be often performed based on machine learning. Currently, when recommending an item, the following methods are generally adopted: and recommending the articles to the user according to the article use information of the user.

However, when the collagen peptide health product recommendation is performed on the user according to the use information of the collagen peptide health product of the user, the following technical problems often exist:

when a user wants to try the type of collagen peptide health products which the user has not used, the user is often difficult to accurately recommend the collagen peptide health products based on the use information of the collagen peptide health products of the user, so that the accuracy of recommending the collagen peptide health products of the user is poor.

Disclosure of Invention

The summary of the invention is provided to introduce a selection of concepts in a simplified form that are further described below in the detailed description. The summary of the invention is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used to limit the scope of the claimed subject matter.

In order to solve the technical problem of poor accuracy of recommending the collagen peptide health care products for users, the invention provides an intelligent recommending method of the collagen peptide health care products based on machine learning.

The invention provides an intelligent recommendation method of a collagen peptide health product based on machine learning, which comprises the following steps:

acquiring user information corresponding to each historical user in a user to be recommended and a historical user set;

the users to be recommended and each history user are target users, and target users with the same occupation in all target users form a target user group;

according to the user information corresponding to all the target users in each target user group, carrying out working intensity analysis processing on the occupation of each target user group to obtain the professional working intensity corresponding to each target user group;

Determining a physical strain index corresponding to each target user according to the user information corresponding to each target user and the professional work intensity corresponding to the target user group to which the target user belongs;

determining the dissimilarity of health care product usage between every two target users according to health care product usage data contained in user information corresponding to all the target users;

clustering all target users according to the body strain index corresponding to all target users, the professional work intensity corresponding to all target user groups and the dissimilarity degree of all health care product usage, so as to obtain target cluster clusters;

and recommending the collagen peptide health products to the user to be recommended according to the user information corresponding to the target user in the target cluster to which the user to be recommended belongs.

Optionally, the analyzing the working intensity of the occupation to which each target user group belongs according to the user information corresponding to all the target users in each target user group to obtain the working intensity of the occupation corresponding to each target user group includes:

determining an excess income proportion corresponding to each target user according to income data and minimum income standard of residence included in user information corresponding to each target user in the target user group;

Classifying target users in the target user group according to excess income proportions corresponding to all the target users in the target user group to obtain a preset number of reference categories;

taking age data contained in user information corresponding to a target user in each reference category as an abscissa, and taking excess income proportion corresponding to the target user in the reference category as an ordinate, and making a scatter diagram corresponding to the reference category;

performing straight line fitting on the scatter diagram corresponding to each reference category, determining a fitting straight line corresponding to each reference category, and obtaining a fitting straight line set corresponding to the target user group;

and determining the professional work intensity corresponding to the target user group according to the fitting straight line set.

Optionally, the determining, according to the income data and the minimum residential income standard included in the user information corresponding to each target user in the target user group, the excess income proportion corresponding to each target user includes:

and determining the ratio of the difference between the income data included in the user information corresponding to the target user and the minimum income standard of the residential area in the minimum income standard of the residential area included in the user information corresponding to the target user as the excess income ratio corresponding to the target user.

Optionally, the formula corresponding to the professional work intensity corresponding to the target user group is:

the method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>Is the firstiProfessional work intensity corresponding to each target user group;iis the sequence number of the target user group; />Is the firstiThe number of fitting straight lines in the fitting straight line set corresponding to each target user group;tandrare all the firstiSequence numbers of fitting straight lines in the fitting straight line sets corresponding to the target user groups; />Is the firstiThe first fitting straight line set corresponding to each target user grouptSlope of each fitting straight line; />Is the firstiThe first fitting straight line set corresponding to each target user grouprSlope of each fitting straight line; />Taking an absolute value function; />Is the firstiThe first fitting straight line set corresponding to each target user grouptIntercept of the fitting lines; />Is the firstiThe first fitting straight line set corresponding to each target user grouprIntercept of the fitting lines; />Is the firstiAnd the average value of all fitting straight line intercept in the fitting straight line set corresponding to each target user group.

Optionally, the determining the physical strain index corresponding to each target user according to the user information corresponding to each target user and the professional work intensity corresponding to the target user group to which the target user belongs includes:

and determining the physical strain index corresponding to each target user according to the age data, the excess income proportion corresponding to each target user and the professional work intensity and the fitting straight line set corresponding to the target user group to which each target user belongs, which are included in the user information corresponding to each target user.

Optionally, the formula corresponding to the physical strain index corresponding to the target user is:

the method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>Is the firstiThe first target user groupjPhysical strain indexes corresponding to the target users;iis the sequence number of the target user group;jis the firstiSequence numbers of target users in the target user groups; />Is the firstiThe first target user groupjExcess income proportion corresponding to each target user; />Is the firstiProfessional work intensity corresponding to each target user group; />Is the firstiThe average value of all fitting straight line slopes in the fitting straight line sets corresponding to the target user groups;is the firstiThe first target user groupjAge data included in user information corresponding to each target user; />Is the firstiThe average value of all fitting straight line intercept in the fitting straight line set corresponding to each target user group; />Is the average value of professional work intensity corresponding to all target user groups; />Is the firstiThe first target user groupjThe corresponding excess income proportion of each target user belongs to the slope of the fitting straight line; />Is the firstiThe first target user groupjFitting of excess income proportion corresponding to each target userIntercept of a straight line.

Optionally, the determining the dissimilarity of health care product usage between every two target users according to the health care product usage data included in the user information corresponding to all the target users includes:

Determining a purchased health product information set which is corresponding to each target user and is included in health product use data corresponding to each target user, wherein the purchased health product information in the purchased health product information set comprises the following components: the product expenditure weight, the product source and the product organization are formed, and the information of one purchased health product corresponds to one collagen peptide health product;

and determining the dissimilarity of health care product use between every two target users according to the product expenditure weight, the product source and the product organization composition included in all the purchased health care product information in the purchased health care product information set corresponding to every two target users.

Optionally, the formula corresponding to the dissimilarity of the health products between the two target users is:

the method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>Is the firsthTarget user and the firstmThe dissimilarity of health care product usage among individual target users;handmis the sequence number of the target user; />；/>Is the firsthThe number of the purchased health product information in the purchased health product information set corresponding to the target users; />Is the firstmThe number of the purchased health product information in the purchased health product information set corresponding to the target users; />Taking an absolute value function; aIs the firsthThe serial numbers of the purchased health product information in the purchased health product information sets corresponding to the target users;bis the firstmThe serial numbers of the purchased health product information in the purchased health product information sets corresponding to the target users; />Is the firsthThe first item in the purchased health product information set corresponding to the target useraThe information of the purchased health products comprises the product expenditure weight; />Is the firstmThe first item in the purchased health product information set corresponding to the target userbThe information of the purchased health products comprises the product expenditure weight; when the first isaThe information of the purchased health products comprises the tissue composition of the productsbSetting +.>Is a first preset value; when the first isaThe information of the purchased health products comprises the tissue composition of the productsbSetting +.>Is a second preset value; the first preset value is smaller than the second preset value; the value range of the grade to which the product source belongs is [1,Q]；Qrepresenting the highest grade of the grades to which the source of the product belongs; when the first isaThe information of the purchased health products comprises the source and the first of the productsbThe information of the purchased health products comprises the lowest grade of the same product sourcesqSetting +. >Is thatq-1；qThe value range of (2) is [1 ],Q]the method comprises the steps of carrying out a first treatment on the surface of the When the first isaThe information of the purchased health products comprises the source and the first of the productsbThe information of the purchased health products comprises the productsWhen the sources are not the same in all grades, setting +.>Is thatQ。

Optionally, clustering all target users according to the physical strain index corresponding to all target users, the professional work intensity corresponding to all target user groups and the dissimilarity of all health products, to obtain a target cluster, including:

correcting the excess income proportion and the physical strain index corresponding to each target user and the professional work intensity corresponding to the target user group to which the target user belongs to obtain the corrected income proportion and the corrected strain degree corresponding to each target user and the corrected work intensity corresponding to the target user group to which the target user belongs respectively, wherein the excess income proportion and the corrected income proportion are positively correlated, the physical strain index and the corrected strain degree are positively correlated, and the professional work intensity and the corrected work intensity are positively correlated;

combining the corrected income proportion and the corrected strain degree corresponding to each target user, the corrected working intensity corresponding to the target user group to which the target user belongs, and age data and sex data included in the corresponding user information into professional physiological feature vectors corresponding to each target user;

Determining the square sum of the dissimilarity between the Euclidean distance between the occupational physiological feature vectors corresponding to each two target users and the health care product usage dissimilarity between the two target users as the target dissimilarity feature between the two target users;

and carrying out hierarchical clustering on all target users according to the target dissimilar characteristics among all target users, and taking each cluster obtained as a target cluster.

Optionally, the recommending the collagen peptide health product to the user to be recommended according to the user information corresponding to the target user in the target cluster to which the user to be recommended belongs includes:

the collagen peptide health products corresponding to the purchased health product information included in the user information corresponding to all the target users in the target cluster with the smallest belonged to the user to be recommended are determined to be candidate health products;

and screening collagen peptide health products which are not purchased by the user to be recommended from all candidate health products, taking the collagen peptide health products as target health products, and recommending the target health products to the user to be recommended.

The invention has the following beneficial effects:

according to the intelligent recommendation method for the collagen peptide health products based on machine learning, through data processing on all user information, the technical problem that the accuracy of recommending the collagen peptide health products to users is poor is solved, and the accuracy of recommending the collagen peptide health products to users is improved. Firstly, user information corresponding to the user to be recommended is acquired, so that preference of the user to be recommended for using the collagen peptide health product can be conveniently analyzed later, and the user to be recommended can be conveniently recommended for recommending the collagen peptide health product later. Secondly, user information corresponding to the historical users is acquired, so that the health care product use similarity condition between the users to be recommended and the historical users can be conveniently analyzed later, the collagen peptide health care product recommendation can be conveniently carried out on the users to be recommended according to the user information corresponding to the historical users with higher similarity in a follow-up auxiliary mode, and the unused collagen peptide health care product types of the users to be recommended can be recommended more accurately to a certain extent. Then, because professional requirements corresponding to different professional types are often different, the working intensity of the target users is often different, so that physical exercise conditions corresponding to users of different professional types are often different, and the types of required collagen peptide health care products are often different. Then, since one target user group corresponds to one occupation, based on the user information corresponding to all the target users in the target user group, the occupation work intensity of the occupation to which the target user group belongs can be quantified. Continuing, since the self situation of the target user and the professional work intensity of the occupation are often related to the hard and tired situation of the target user, the physical strain index corresponding to the target user can be quantified based on the user information corresponding to the target user and the professional work intensity corresponding to the target user group to which the target user belongs. And then, comprehensively considering the health care product use data included in the user information corresponding to all the target users, and quantifying the use dissimilarity of the health care products between every two target users, wherein the smaller the value is, the closer the health care product use preference of the two target users is. And then, based on the body strain index corresponding to all the target users, the professional working intensity corresponding to all the target user groups and the use dissimilarity of all the health products, clustering all the target users, so that the target users with similar health product use preference, body strain condition and working intensity can be divided into the same target cluster as far as possible. Finally, because the target users in the target cluster to which the user to be recommended belongs are often target users similar to the health care product use preference, the physical strain condition and the working strength of the user to be recommended, compared with the method for recommending the collagen peptide health care product by only considering the user information corresponding to the user to be recommended, the method and the device for recommending the collagen peptide health care product based on the user information corresponding to the target users in the target cluster to which the user to be recommended belongs, comprehensively consider the user information corresponding to the user to be recommended and the user information corresponding to the health care product use preference, the physical strain condition and the working strength of the user to be recommended, the method and the device can reasonably recommend the collagen peptide health care product to the user to be recommended, so that the accuracy of recommending the collagen peptide health care product to the user to be recommended is improved. Secondly, compared with the method for recommending the collagen peptide health products based on the content by manually labeling different types of collagen peptide health products with fixed efficacy, the method can avoid the problem of poor accuracy of recommending the collagen peptide health products for users to be recommended due to the fact that the manual labeling is wrong due to various types of the collagen peptide health products to a certain extent.

Drawings

In order to more clearly illustrate the embodiments of the invention or the technical solutions and advantages of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are only some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.

Fig. 1 is a flowchart of the intelligent recommendation method of the collagen peptide health care product based on machine learning.

Detailed Description

In order to further describe the technical means and effects adopted by the present invention to achieve the preset purpose, the following detailed description is given below of the specific implementation, structure, features and effects of the technical solution according to the present invention with reference to the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" means that the embodiments are not necessarily the same. Furthermore, the particular features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

The invention provides an intelligent recommendation method of a collagen peptide health product based on machine learning, which comprises the following steps:

acquiring user information corresponding to each historical user in a user to be recommended and a historical user set;

recording the users to be recommended and each history user as target users, and forming a target user group by using target users with the same occupation in all target users;

according to the user information corresponding to all the target users in each target user group, carrying out working intensity analysis processing on the occupation of each target user group to obtain the professional working intensity corresponding to each target user group;

determining a physical strain index corresponding to each target user according to the user information corresponding to each target user and the professional work intensity corresponding to the target user group to which the target user belongs;

determining the dissimilarity of health care product usage between every two target users according to health care product usage data contained in user information corresponding to all the target users;

clustering all target users according to the body strain index corresponding to all target users, the professional work intensity corresponding to all target user groups and the dissimilarity degree of all health care product usage, so as to obtain target cluster clusters;

And recommending the collagen peptide health products to the user to be recommended according to the user information corresponding to the target user in the target cluster to which the user to be recommended belongs.

The following detailed development of each step is performed:

referring to fig. 1, a flow of some embodiments of a machine learning based collagen peptide health care product intelligent recommendation method according to the present invention is shown. The intelligent collagen peptide health product recommending method based on machine learning comprises the following steps:

step S1, user information corresponding to each historical user in a user to be recommended and a historical user set is obtained.

In some embodiments, user information corresponding to each historical user in the set of users to be recommended and historical users may be obtained.

The user to be recommended can be a user to be recommended for collagen peptide health products. The collagen peptide is a small molecular extract of animal tissues, belongs to a health-care product, and can be recorded as a collagen peptide health-care product. The historical users in the historical user collection can be users recorded with information related to the collagen peptide health care products. The user to be recommended and each history user are recorded as target users, and the user information corresponding to the target users can be information related to the collagen peptide health products of the target users. The user information may include: revenue data, minimum residential revenue criteria, age data, gender data, occupation data, and health care usage data. Revenue data may be data characterizing revenue. For example, the revenue data may be the monthly revenue of the target user. The residence minimum revenue criteria may be a month minimum payroll criteria for the residence where the target user is located. For example, the residence minimum revenue criterion may be the month minimum payroll criterion for the city in which the target user is located. The age data may be the age of the target user. The gender data may be the gender of the target user. The occupation data may be the occupation of the target user. The health product use data can be recorded data of the condition that the target user uses the collagen peptide health product. The health product usage data may include a collection of purchased health product information. The purchased health product information in the purchased health product information set may be information of the target user purchasing each collagen peptide health product. One piece of information of the purchased health care product can correspond to one collagen peptide health care product. The purchased healthcare product information in the purchased healthcare product information set may include: product expense weight, product source and product organization. The product expenditure weight included in the purchased healthcare product information may be a total amount spent by the target user purchasing the collagen peptide healthcare products corresponding to the purchased healthcare product information, and a ratio of the total amount spent by the target user purchasing all the collagen peptide healthcare products. The purchased health product information can comprise animal sources of the collagen peptide health products corresponding to the purchased health product information, and can represent which animal the collagen peptide health products are extracted from. The product tissue composition included in the purchased health product information may be an animal tissue composition of the collagen peptide health product corresponding to the purchased health product information, and may indicate that the collagen peptide health product is extracted from the animal tissue.

It should be noted that, the user information corresponding to the user to be recommended is obtained, so that the preference of the user to be recommended for using the collagen peptide health product can be conveniently analyzed later, and the user to be recommended can be conveniently recommended for recommending the collagen peptide health product later. Secondly, user information corresponding to the historical users is acquired, so that the health care product use similarity condition between the users to be recommended and the historical users can be conveniently analyzed later, the collagen peptide health care product recommendation can be conveniently carried out on the users to be recommended according to the user information corresponding to the historical users with higher similarity in a follow-up auxiliary mode, and the unused collagen peptide health care product types of the users to be recommended can be recommended more accurately to a certain extent.

As an example, the user information corresponding to the user to be recommended and each history user may be filled according to the autonomous willingness of the user to be recommended and each history user, and whether to be anonymous may be selected in the filling process.

And S2, recording the users to be recommended and each history user as target users, and forming a target user group by using target users with the same occupation in all target users.

In some embodiments, the user to be recommended and each history user may be remembered as target users, and target users with the same occupation in all target users may be formed into a target user group.

It should be noted that, because professional requirements corresponding to different professional types are often different, the working intensity of the target users is often different, so that physical exercise conditions corresponding to users of different professional types are often different, and the types of required collagen peptide health care products are often different, therefore, the target users with the same profession in all target users form a target user group, and the professional working intensity of each profession can be conveniently analyzed later.

As an example, target users whose professional data included in the user information is the same may be divided into the same target user group.

And S3, according to the user information corresponding to all the target users in each target user group, analyzing and processing the working intensity of the occupation to which each target user group belongs, and obtaining the working intensity of the occupation corresponding to each target user group.

In some embodiments, according to user information corresponding to all target users in each target user group, a job intensity analysis process may be performed on a job to which each target user group belongs, so as to obtain a job intensity corresponding to each target user group.

The occupation to which the target user group belongs may be the occupation of the target user in the target user group.

It should be noted that, since one target user group corresponds to one occupation, based on the user information corresponding to all the target users in the target user group, the occupation work intensity of the occupation to which the target user group belongs can be quantified.

As an example, this step may include the steps of:

and a first step of determining an excess income proportion corresponding to each target user according to income data and minimum residential income standard included in the user information corresponding to each target user in the target user group.

For example, the formula for determining the ratio of the sum of the income data included in the user information corresponding to the target user and the minimum income criterion of the residence, which is the ratio of the sum of the income data included in the user information corresponding to the target user and the minimum income criterion of the residence, may be:

the method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>Is the firstiThe first target user groupjAnd the excess income proportion corresponding to each target user.iIs the sequence number of the target user group.jIs the firstiSequence numbers of target users in the individual target user groups. />Is the firstiThe first target user groupjAnd the user information corresponding to each target user comprises income data. />Is the firstiThe first target user group jThe user information corresponding to the individual target users includes minimum residential revenue criteria.

When the following is performedThe larger the tends to explain the firstiThe first target user groupjThe higher the revenue of the individual target user, the more often the description of the firstiThe first target user groupjThe more likely an individual target user is at a residence to belong to a high-income group. Because the demand of the people with different incomes for the acceptable price of the collagen peptide health care products is often differentThe demand of the people with similar income for the collagen peptide health care products is likely to be the same, and secondly, the excessive income proportion is likely to have a certain relation with the work type and is likely to be related with the work intensity, so that the physical condition of people is likely to be influenced, and the demand of the people with similar excessive income proportion for the collagen peptide health care products is likely to be the same, so that the excessive income proportion corresponding to the target users is quantified, and the collagen peptide health care products which are more suitable for the users to be recommended can be conveniently and assisted by the target users with similar excessive income proportion corresponding to the users to be recommended.

And secondly, classifying the target users in the target user group according to the excess income proportion corresponding to all the target users in the target user group to obtain a preset number of reference categories.

The preset number may be a preset number. For example, the preset number may be 10.

For example, all target users in the target user group are ordered according to the order of the excess income proportion corresponding to the target users from large to small, the obtained sequence is used as a target user sequence corresponding to the target user group, the target user sequence is equally divided into a preset number of subsequences, and the target users in each subsequence are used as target users in a reference category.

And thirdly, taking age data included in user information corresponding to the target user in each reference category as an abscissa, and taking excess income proportion corresponding to the target user in the reference category as an ordinate to make a scatter diagram corresponding to the reference category.

And fourthly, performing straight line fitting on the scatter diagram corresponding to each reference category, and determining a fitting straight line corresponding to each reference category to obtain a fitting straight line set corresponding to the target user group.

The number of the fitting straight lines in the fitting straight line set corresponding to the target user group may be a preset number.

Fifthly, determining a formula corresponding to the professional work intensity corresponding to the target user group according to the fitting straight line set, wherein the formula can be as follows:

The method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>Is the firstiProfessional work intensity corresponding to each target user group.iIs the sequence number of the target user group. />Is the firstiThe number of fitting lines in the fitting line set corresponding to each target user group.tAndrare all the firstiSequence numbers of the fitting straight lines in the fitting straight line sets corresponding to the target user groups. />Is the firstiThe first fitting straight line set corresponding to each target user grouptThe slope of the fitted line. />Is the firstiThe first fitting straight line set corresponding to each target user grouprThe slope of the fitted line. />Taking an absolute value function. />Is the firstiThe first fitting straight line set corresponding to each target user grouptThe intercept of the fitting line. />Is the firstiThe first fitting straight line set corresponding to each target user grouprThe intercept of the fitting line. />Is the firstiAnd the average value of all fitting straight line intercept in the fitting straight line set corresponding to each target user group.

It should be noted that the slope of the fitted line can be often characterizedRevenue growth rate, intercept of the fitted line may often characterize the initial revenue. When (when)The larger the tends to explain the firstiThe more uneven the allocation of revenue to the profession the individual target user group belongs to. When->The larger the tends to explain the firstiThe more obvious the occupation of the target user group increases with age, the more uneven the income gap will be, which tends to indicate that the income distribution is likely to be uneven. When- >The larger the tends to explain the firstiThe more uneven the allocation of revenue to the profession the target user group belongs to. Thus, when->The larger the tends to explain the firstiThe more uneven the revenue distribution of the occupation of the target user group is, the more often the description isiThe higher the sinking cost of the profession to which the individual target user group belongs. Generally, whether the income distribution of the profession is uniform and stable is often related to the working intensity, and often influences the working attitude and the effort degree of people, and often influences the physical condition of people, so that the demands of people on the collagen peptide health care product are influenced, and the demands of target users with similar professional working intensity on the collagen peptide health care product are often similar. Therefore, the professional work intensity is quantified, and the more appropriate collagen peptide health care product can be conveniently recommended to the user to be recommended in a follow-up auxiliary manner according to the target user which is similar to the professional work intensity of the work to which the user to be recommended belongs.

And S4, determining the body strain index corresponding to each target user according to the user information corresponding to each target user and the professional work intensity corresponding to the target user group to which the target user belongs.

In some embodiments, the physical strain index corresponding to each target user may be determined according to the user information corresponding to each target user and the professional work intensity corresponding to the target user group to which the target user belongs.

It should be noted that, since the self situation of the target user and the professional work intensity of the occupation are often related to the hard and tired situation of the target user, the physical strain index corresponding to the target user can be quantified based on the user information corresponding to the target user and the professional work intensity corresponding to the target user group to which the target user belongs.

As an example, according to age data included in the user information corresponding to each target user, an excess income proportion corresponding to the target user, and a professional work intensity and a fitting straight line set corresponding to the target user group to which the target user belongs, a formula corresponding to the physical strain index corresponding to each target user may be determined as follows:

the method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>Is the firstiThe first target user groupjAnd the physical strain index corresponding to each target user.iIs the sequence number of the target user group.jIs the firstiSequence numbers of target users in the individual target user groups. />Is the firstiThe first target user groupjAnd the excess income proportion corresponding to each target user. />Is the firstiProfessional work intensity corresponding to each target user group. />Is the firstiAnd the average value of the slopes of all the fitting straight lines in the fitting straight line set corresponding to each target user group.Is the firstiThe first target user groupjAge data included in the user information corresponding to the individual target users. / >Is the firstiAnd the average value of all fitting straight line intercept in the fitting straight line set corresponding to each target user group. />Is the average value of the occupational work intensity corresponding to all the target user groups. />Is the firstiThe first target user groupjThe excess income proportion corresponding to each target user belongs to the slope of the fitting straight line. />Is the firstiThe first target user groupjThe corresponding excess income proportion of each target user belongs to the intercept of the fitting straight line.

It should be noted that, since the excess income proportion is often related to the work type, is often related to the work intensity, and thus affects the physical condition of people, and the professional work intensity often affects the work attitude and effort of people, thus affecting the physical condition of people, soCan characterize the firstjThe physical condition corresponding to the individual target user. />Often can characterize the firstjThe occupation of the target user is in the first placejAverage excess income level of age group of the individual target users,/->Can characterize the average level of all occupational work intensities, soCan characterize the firstjThe occupation of the target user is in the first placejAverage physical condition of the population of age groups of the individual target users. />Mainly for correcting- >. Thus, the first and second substrates are bonded together,can characterize the firstjPhysical condition and physical strain condition of the individual target users. Generally, the physical strain often affects the demands of people on collagen peptide health care products, and the demands of target users with similar physical strain often are similar.

And S5, determining the dissimilarity of the health care product usage between every two target users according to the health care product usage data contained in the user information corresponding to all the target users.

In some embodiments, the dissimilarity of health care product usage between every two target users may be determined according to health care product usage data included in user information corresponding to all target users.

It should be noted that, comprehensively considering the health care product usage data included in the user information corresponding to all the target users, the dissimilarity of health care product usage between every two target users can be quantified, and the smaller the value, the closer the health care product usage preferences of the two target users are often described.

As an example, this step may include the steps of:

the method comprises the steps of firstly, determining a purchased health product information set which is included in health product use data and is included in user information corresponding to each target user as the purchased health product information set corresponding to each target user.

Secondly, according to the product expenditure weight, the product source and the product organization composition contained in all purchased health product information in the purchased health product information set corresponding to each two target users, the formula corresponding to the dissimilarity of health product use between each two target users can be determined as follows:

the method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>Is the firsthTarget user and the firstmThe health care products among the target users are not similar.hAndmis the sequence number of the target user. />。/>Is the firsthThe number of the purchased health care product information in the purchased health care product information set corresponding to each target user. />Is the firstmThe number of the purchased health care product information in the purchased health care product information set corresponding to each target user. />Is a function of absolute value.aIs the firsthAnd the serial numbers of the purchased health product information in the purchased health product information sets corresponding to the target users.bIs the firstmAnd the serial numbers of the purchased health product information in the purchased health product information sets corresponding to the target users. />Is the firsthThe first item in the purchased health product information set corresponding to the target useraThe information of the purchased health products comprises the product expenditure weight. />Is the firstmThe first item in the purchased health product information set corresponding to the target userbThe information of the purchased health products comprises the product expenditure weight. When the first is aThe information of the purchased health products comprises the tissue composition of the productsbSetting +.>Is a first preset value. When the first isaThe information of the purchased health products comprises the tissue composition of the productsbSetting +.>Is a second preset value. The first preset value is smaller than the second preset value. For example, the first preset value may be 0 and the second preset value may be 2. The value range of the grade to which the product source belongs is [1,Q]。Qrepresenting the highest grade of the grades to which the source of the product belongs. Because the product source is the animal source of the collagen peptide health care product, namely the animal, the grade of the product source can be 7, and the product source can be sequentially from low to high: "species", "genus", "family", "order", "class", "phylum" and "kingdom", at this pointQMay be 7 and may represent a "boundary" in the hierarchy. When the first isaThe information of the purchased health products comprises the source and the first of the productsbThe information of the purchased health products comprises the lowest grade of the same product sourcesqSetting +.>Is thatq-1. If at firstaThe information of the purchased health products comprises the source and the first of the products bThe source of the products included in the information of the purchased health products is different in the case of 'species', and the sources are the same in the case of 'genus', 'family', 'order', 'class', 'door' and 'kingdom', so that +_can be set at the moment>1.qThe value range of (2) is [1 ],Q]. When the first isaThe information of the purchased health products comprises the source and the first of the productsbSetting +.A. when the source of products included in the information of the purchased health products is different in all grades>Is thatQ。

When the following is performedThe smaller the time, the more description of the firsthTarget user pair(s)aThe product expenditure weight and the first product expenditure weight of the collagen peptide health products corresponding to the information of the purchased health productsmTarget user pair(s)bThe more similar the product expenditure weight of the collagen peptide health products corresponding to the information of the purchased health products is, the description of the first step is oftenhTarget user pair(s)aThe favorability degree and the first of the collagen peptide health products corresponding to the information of the purchased health productsmTarget user pair(s)bThe more similar the favorability of the collagen peptide health products corresponding to the information of the purchased health products. When->The smaller the time, the more description of the firstaThe information of the purchased health products comprises the source and the first of the productsbThe more similar the source of products included in the information of the purchased health products. When->The smaller the time, the more description of the first aThe information of the purchased health products comprises the tissue composition of the productsbThe more similar the organization composition of the products included in the information of the purchased health products. So when->、/>And->The smaller the time, the more description of the firsthTarget user and the firstmThe more similar the preference degree of the target users to the collagen peptide health care products composed of similar product sources and product tissues. Thus, when->The smaller the time, the more description of the firsthTarget user and the firstmThe more similar the target users like the raw material composition of the collagen peptide health care product.

And S6, clustering all the target users according to the body strain index corresponding to all the target users, the professional work intensity corresponding to all the target user groups and the dissimilarity of all the health care products, and obtaining a target cluster.

In some embodiments, the target clusters may be obtained by clustering all target users according to the physical strain indexes corresponding to all target users, the professional work intensities corresponding to all target user groups, and the dissimilarity of use of all health products.

It should be noted that, based on the physical strain indexes corresponding to all the target users, the professional working intensities corresponding to all the target user groups, and the dissimilarity of use of all the health products, clustering all the target users can divide the target users with similar health product use preference, physical strain condition, and working intensity into the same target cluster as much as possible.

As an example, this step may include the steps of:

the first step, correcting the excess income proportion and the physical strain index corresponding to each target user and the professional working intensity corresponding to the target user group to which the target user belongs to obtain the corrected income proportion and the corrected strain degree corresponding to each target user and the corrected working intensity corresponding to the target user group to which the target user belongs respectively.

Wherein the excess revenue ratio may be positively correlated with the corrected revenue ratio. The physical strain indicator may be positively correlated with the degree of corrected strain. The professional work intensity may be positively correlated with the corrected work intensity.

For example, the formulas corresponding to the corrected income proportion and the corrected strain degree of the target user and the corrected working intensity of the target user group to which the target user belongs may be:

；/>；/>the method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>Is the firstiThe first target user groupjAnd correcting the income proportion corresponding to each target user.iIs the sequence number of the target user group.jIs the firstiSequence numbers of target users in the individual target user groups. />Is a base 10 logarithmic function. />Is the firstiThe first target user groupjAnd the excess income proportion corresponding to each target user. />Is the firstiThe first target user groupjAnd correcting the strain degree corresponding to each target user. Is the firstiThe first target user groupjAnd the physical strain index corresponding to each target user. />Is the firstiAnd correcting the working intensity corresponding to each target user group. />Is the firstiProfessional work intensity corresponding to each target user group.

When the following is performedThe larger the tends to explain the firstiThe first target user groupjThe higher the revenue for the individual target user. Thus, when->The larger the tends to explain the firstiThe first target user groupjThe higher the revenue of the individual target user, the more often the description of the firstiThe first target user groupjThe more likely an individual target user is at a residence to belong to a high-income group. When->The larger the tends to explain the firstiThe more uneven the revenue distribution of the occupation of the target user group is, the more often the description isiThe higher the sinking cost of the occupation to which the individual target user group belongs, therefore, when +.>The larger the tends to explain the firstiThe more uneven the revenue distribution of the occupation of the target user group is, the more often the description isiThe higher the sinking cost of the profession to which the individual target user group belongs. />Can characterize the firstjPhysical condition and physical strain condition of the individual target users. Thus->Can characterize the firstjPhysical condition and physical strain condition of the individual target users.

And secondly, combining the corrected income proportion and the corrected strain degree corresponding to each target user, the corrected working intensity corresponding to the target user group to which the target user belongs, and age data and sex data included in the corresponding user information into professional physiological feature vectors corresponding to each target user.

For example, the firstiThe first target user groupjOccupational physiological feature vectors corresponding to individual target users may be {，，/>，/>，/>}, wherein->Is the firstiThe first target user groupjThe corrected income proportion corresponding to each target user; />Is the firstiCorrecting working intensity corresponding to each target user group; />Is the firstiThe first target user groupjCorrecting strain degrees corresponding to the target users; />Is the firstiThe first target user groupjGender data included in the user information corresponding to the target users; />Is the firstiThe first target user groupjAge data included in the user information corresponding to the individual target users.

And thirdly, determining the square sum of the Euclidean distance between occupational physiological feature vectors corresponding to each two target users and the dissimilarity degree of the health care product usage between the two target users as the target dissimilarity feature between the two target users.

Fourth, hierarchical clustering is carried out on all target users according to target dissimilar characteristics among all target users, and each cluster is obtained and is used as a target cluster.

For example, the target dissimilar characteristics among the target users can be used as a measure in the clustering process, hierarchical clustering is performed on all the target users, and each obtained cluster is used as a target cluster.

Optionally, product expenditure weights, product sources and product organizations included in all purchased health product information in the purchased health product information sets corresponding to all target users can be used as input, all target users are clustered through a hierarchical clustering algorithm, in the clustering process, target dissimilar characteristics between two target users can be calculated according to the product expenditure weights, the product sources and the product organizations included in all purchased health product information in the purchased health product information sets corresponding to every two target users, and the sum of the calculated Euclidean distance between occupational physiological characteristic vectors corresponding to every two target users and the target dissimilar characteristics between the two target users can be used as a measure in the clustering process; and outputting a tree structure which is hierarchical clustering, and taking each cluster obtained at the moment as a target cluster.

And S7, recommending the collagen peptide health products to the user to be recommended according to the user information corresponding to the target user in the target cluster to which the user to be recommended belongs.

In some embodiments, the collagen peptide health product recommendation of the user to be recommended may be performed according to the user information corresponding to the target user in the target cluster to which the user to be recommended belongs, so that the collagen peptide health product recommendation of the user to be recommended may be implemented.

As an example, this step may include the steps of:

firstly, determining collagen peptide health products corresponding to each purchased health product information included in user information corresponding to all target users in the target cluster with the smallest belonged to the user to be recommended as candidate health products.

And secondly, screening collagen peptide health products which are not purchased by the user to be recommended from all candidate health products, taking the collagen peptide health products as target health products, and recommending the target health products to the user to be recommended.

For example, the recommendation of the target health product to the user to be recommended may be achieved by sending the name corresponding to the target health product to the target terminal. The target terminal may be a mobile phone.

Optionally, if more collagen peptide health products are to be recommended to the user to be recommended, the recommending of the collagen peptide health products to the user to be recommended may be performed according to user information corresponding to all target users in a target cluster of a larger hierarchy to which the user to be recommended belongs, which may specifically be: the target cluster of a larger hierarchy, to which the user to be recommended belongs, is used as the smallest target cluster to which the user to be recommended belongs, and the first step to the second step, which are included as examples in the step S7, are executed, so that more collagen peptide health products can be recommended to the user to be recommended.

In summary, because the target users in the target cluster to which the user to be recommended belongs are often target users similar to the health care product use preference, the physical strain condition and the working strength of the user to be recommended, compared with the method for recommending the collagen peptide health care product by only considering the user information corresponding to the user to be recommended, the method for recommending the collagen peptide health care product by the user to be recommended based on the user information corresponding to the target users in the target cluster to which the user to be recommended belongs comprehensively considers the user information corresponding to the user to be recommended and the user information corresponding to the health care product use preference, the physical strain condition and the working strength of the user to be recommended, the method for recommending the collagen peptide health care product by the user to be recommended can reasonably recommended, and accordingly accuracy of recommending the collagen peptide health care product by the user to be recommended is improved. Secondly, compared with the method for recommending the collagen peptide health products based on the content by manually labeling different types of collagen peptide health products with fixed efficacy, the method can avoid the problem of poor accuracy of recommending the collagen peptide health products for users to be recommended due to the fact that the manual labeling is wrong due to various types of the collagen peptide health products to a certain extent.

The above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit of the invention and are intended to be included within the scope of the invention.

Claims (10)

1. The intelligent collagen peptide health product recommending method based on machine learning is characterized by comprising the following steps of:

acquiring user information corresponding to each historical user in a user to be recommended and a historical user set;

the users to be recommended and each history user are target users, and target users with the same occupation in all target users form a target user group;

according to the user information corresponding to all the target users in each target user group, carrying out working intensity analysis processing on the occupation of each target user group to obtain the professional working intensity corresponding to each target user group;

Determining a physical strain index corresponding to each target user according to the user information corresponding to each target user and the professional work intensity corresponding to the target user group to which the target user belongs;

determining the dissimilarity of health care product usage between every two target users according to health care product usage data contained in user information corresponding to all the target users;

clustering all target users according to the body strain index corresponding to all target users, the professional work intensity corresponding to all target user groups and the dissimilarity degree of all health care product usage, so as to obtain target cluster clusters;

and recommending the collagen peptide health products to the user to be recommended according to the user information corresponding to the target user in the target cluster to which the user to be recommended belongs.

2. The intelligent recommendation method of collagen peptide health products based on machine learning according to claim 1, wherein the performing the working strength analysis processing on the occupation of each target user group according to the user information corresponding to all the target users in each target user group to obtain the professional working strength corresponding to each target user group comprises:

determining an excess income proportion corresponding to each target user according to income data and minimum income standard of residence included in user information corresponding to each target user in the target user group;

Classifying target users in the target user group according to excess income proportions corresponding to all the target users in the target user group to obtain a preset number of reference categories;

taking age data contained in user information corresponding to a target user in each reference category as an abscissa, and taking excess income proportion corresponding to the target user in the reference category as an ordinate, and making a scatter diagram corresponding to the reference category;

performing straight line fitting on the scatter diagram corresponding to each reference category, determining a fitting straight line corresponding to each reference category, and obtaining a fitting straight line set corresponding to the target user group;

and determining the professional work intensity corresponding to the target user group according to the fitting straight line set.

3. The intelligent recommendation method for collagen peptide health products based on machine learning according to claim 2, wherein the determining the excess income ratio corresponding to each target user according to the income data and the minimum income standard of the residence included in the user information corresponding to each target user in the target user group comprises:

and determining the ratio of the difference between the income data included in the user information corresponding to the target user and the minimum income standard of the residential area in the minimum income standard of the residential area included in the user information corresponding to the target user as the excess income ratio corresponding to the target user.

4. The intelligent recommendation method for collagen peptide health products based on machine learning according to claim 2, wherein the formula corresponding to the professional work intensity corresponding to the target user group is:

the method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>Is the firstiProfessional work intensity corresponding to each target user group;iis the sequence number of the target user group; />Is the firstiThe number of fitting straight lines in the fitting straight line set corresponding to each target user group;tandrare all the firstiSequence numbers of fitting straight lines in the fitting straight line sets corresponding to the target user groups; />Is the firstiThe first fitting straight line set corresponding to each target user grouptSlope of each fitting straight line; />Is the firstiThe first fitting straight line set corresponding to each target user grouprSlope of each fitting straight line; />Taking an absolute value function; />Is the firstiThe first fitting straight line set corresponding to each target user grouptIntercept of the fitting lines; />Is the firstiThe first fitting straight line set corresponding to each target user grouprIntercept of the fitting lines;is the firstiAnd the average value of all fitting straight line intercept in the fitting straight line set corresponding to each target user group.

5. The intelligent recommendation method for collagen peptide health products based on machine learning according to claim 2, wherein the determining the physical strain index corresponding to each target user according to the user information corresponding to each target user and the professional work intensity corresponding to the target user group to which the target user belongs comprises the following steps:

And determining the physical strain index corresponding to each target user according to the age data, the excess income proportion corresponding to each target user and the professional work intensity and the fitting straight line set corresponding to the target user group to which each target user belongs, which are included in the user information corresponding to each target user.

6. The intelligent recommendation method for collagen peptide health products based on machine learning according to claim 5, wherein the formula corresponding to the physical strain index corresponding to the target user is:

the method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>Is the firstiThe first target user groupjPhysical strain indexes corresponding to the target users;iis the sequence number of the target user group;jis the firstiSequence numbers of target users in the target user groups; />Is the firstiThe first target user groupjExcess income proportion corresponding to each target user; />Is the firstiProfessional work intensity corresponding to each target user group; />Is the firstiThe average value of all fitting straight line slopes in the fitting straight line sets corresponding to the target user groups; />Is the firstiThe first target user groupjAge data included in user information corresponding to each target user; />Is the firstiThe average value of all fitting straight line intercept in the fitting straight line set corresponding to each target user group; />Is the average value of professional work intensity corresponding to all target user groups; / >Is the firstiThe first target user groupjThe corresponding excess income proportion of each target user belongs to the slope of the fitting straight line; />Is the firstiThe first target user groupjThe corresponding excess income proportion of each target user belongs to the intercept of the fitting straight line.

7. The intelligent recommendation method for collagen peptide health products based on machine learning according to claim 1, wherein the determining the dissimilarity of health product usage between every two target users according to health product usage data included in user information corresponding to all target users comprises:

determining a purchased health product information set which is corresponding to each target user and is included in health product use data corresponding to each target user, wherein the purchased health product information in the purchased health product information set comprises the following components: the product expenditure weight, the product source and the product organization are formed, and the information of one purchased health product corresponds to one collagen peptide health product;

and determining the dissimilarity of health care product use between every two target users according to the product expenditure weight, the product source and the product organization composition included in all the purchased health care product information in the purchased health care product information set corresponding to every two target users.

8. The intelligent recommendation method for collagen peptide health products based on machine learning according to claim 7, wherein the formula corresponding to the dissimilarity of health products used by two target users is:

the method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>Is the firsthTarget user and the firstmThe dissimilarity of health care product usage among individual target users;handmis the sequence number of the target user; />；/>Is the firsthThe number of the purchased health product information in the purchased health product information set corresponding to the target users; />Is the firstmThe number of the purchased health product information in the purchased health product information set corresponding to the target users; />Taking an absolute value function;ais the firsthThe serial numbers of the purchased health product information in the purchased health product information sets corresponding to the target users;bis the firstmThe serial numbers of the purchased health product information in the purchased health product information sets corresponding to the target users; />Is the firsthThe first item in the purchased health product information set corresponding to the target useraThe information of the purchased health products comprises the product expenditure weight; />Is the firstmThe first item in the purchased health product information set corresponding to the target userbThe information of the purchased health products comprises the product expenditure weight; when the first isaThe information of the purchased health products comprises the tissue composition of the products bSetting +.>Is a first preset value; when the first isaThe information of the purchased health products comprises the tissue composition of the productsbSetting +.>Is a second preset value; the first preset value is smaller than the second preset value; the value range of the grade to which the product source belongs is [1,Q]；Qrepresenting the highest grade of the grades to which the source of the product belongs; when the first isaThe information of the purchased health products comprises the source and the first of the productsbThe information of the purchased health products comprises the lowest grade of the same product sourcesqSetting +.>Is thatq-1；qThe value range of (2) is [1 ],Q]the method comprises the steps of carrying out a first treatment on the surface of the When the first isaThe information of the purchased health products comprises the source and the first of the productsbSetting +.A. when the source of products included in the information of the purchased health products is different in all grades>Is thatQ。

9. The intelligent recommendation method for collagen peptide health products based on machine learning according to claim 1, wherein the clustering of all target users according to the physical strain index corresponding to all target users, the professional work intensity corresponding to all target user groups and the dissimilarity of all health products to obtain target cluster comprises the following steps:

Correcting the excess income proportion and the physical strain index corresponding to each target user and the professional work intensity corresponding to the target user group to which the target user belongs to obtain the corrected income proportion and the corrected strain degree corresponding to each target user and the corrected work intensity corresponding to the target user group to which the target user belongs respectively, wherein the excess income proportion and the corrected income proportion are positively correlated, the physical strain index and the corrected strain degree are positively correlated, and the professional work intensity and the corrected work intensity are positively correlated;

combining the corrected income proportion and the corrected strain degree corresponding to each target user, the corrected working intensity corresponding to the target user group to which the target user belongs, and age data and sex data included in the corresponding user information into professional physiological feature vectors corresponding to each target user;

determining the square sum of the dissimilarity between the Euclidean distance between the occupational physiological feature vectors corresponding to each two target users and the health care product usage dissimilarity between the two target users as the target dissimilarity feature between the two target users;

and carrying out hierarchical clustering on all target users according to the target dissimilar characteristics among all target users, and taking each cluster obtained as a target cluster.

10. The intelligent recommendation method for collagen peptide health products based on machine learning according to claim 9, wherein the recommending the collagen peptide health products to the user to be recommended according to the user information corresponding to the target user in the target cluster to which the user to be recommended belongs comprises:

the collagen peptide health products corresponding to the purchased health product information included in the user information corresponding to all the target users in the target cluster with the smallest belonged to the user to be recommended are determined to be candidate health products;

and screening collagen peptide health products which are not purchased by the user to be recommended from all candidate health products, taking the collagen peptide health products as target health products, and recommending the target health products to the user to be recommended.