CN116823361B - Jewelry collocation detection and pushing method based on artificial intelligence - Google Patents
Jewelry collocation detection and pushing method based on artificial intelligence Download PDFInfo
- Publication number
- CN116823361B CN116823361B CN202311108051.1A CN202311108051A CN116823361B CN 116823361 B CN116823361 B CN 116823361B CN 202311108051 A CN202311108051 A CN 202311108051A CN 116823361 B CN116823361 B CN 116823361B
- Authority
- CN
- China
- Prior art keywords
- user
- ornament
- ornaments
- seed point
- index
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 22
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 14
- 238000001514 detection method Methods 0.000 title claims abstract description 13
- 238000012216 screening Methods 0.000 claims abstract description 6
- 238000004364 calculation method Methods 0.000 claims description 15
- 238000010606 normalization Methods 0.000 claims description 4
- 238000012163 sequencing technique Methods 0.000 claims description 4
- 238000004458 analytical method Methods 0.000 abstract description 3
- 238000011160 research Methods 0.000 abstract description 3
- 238000005034 decoration Methods 0.000 description 8
- 239000000243 solution Substances 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 238000013139 quantization Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000002347 injection Methods 0.000 description 1
- 239000007924 injection Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000036651 mood Effects 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The application relates to the technical field of analysis and research of buyers, in particular to a jewelry collocation detection and pushing method based on artificial intelligence, which comprises the following steps: determining a wearing and taking style index of a user and a target image with characters similar to the wearing and taking style of the user in the image; comparing the wearing style index of the person in the target image with the wearing style index of the user, and the characteristic index of the same kind of characteristic label corresponding to each ornament worn by the person in the target image, screening out seed point ornaments from each ornament, and clustering the ornaments of interest to the user according to the difference of the characteristic indexes of the same kind of characteristic label corresponding to the ornaments of interest to the user and the seed point ornaments to obtain each cluster; and determining the user attention degree corresponding to various feature labels in the cluster, further determining the recommendation sequence of the seed point ornaments, and recommending the seed point ornaments to the user. The application can accurately identify the proper ornaments of the user and improve the recommending rationality of the ornaments.
Description
Technical Field
The application relates to the technical field of analysis and research of buyers, in particular to an artificial intelligence-based ornament collocation detection and pushing method.
Background
The jewelry collocation is one of means for improving the appearance of the jewelry collocation in the daily life of people at present, and is closely related to the life of people, and when people purchase jewelry, a jewelry selling shop is generally required to provide personalized recommendation of the jewelry collocation for users. Along with the rapid development of the information age, jewelry collocation recommendation is gradually combined with artificial intelligence, and an intelligent jewelry special cabinet is generated, which aims to recommend proper jewelry for users so as to improve the purchasing power of the users, thereby greatly increasing the sales volume of the jewelry.
When recommending jewelry collocation, the existing intelligent jewelry special cabinets usually recommend one or more jewelry to a user according to the current day of the user wearing and combining with the established hot goods of a merchant. However, because the preference of the user is affected by age, mood and other factors and changes with time, the user cannot accurately acquire the real purchasing trend of the user through the daily wearing of a single reference user, so that the recommended jewelry cannot meet the needs of the user and the recommendation rationality is poor.
Disclosure of Invention
The application aims to provide an artificial intelligence-based jewelry collocation detection and pushing method, which is used for solving the problem that the recommendation of the existing jewelry is unreasonable.
In order to solve the technical problems, the application provides an artificial intelligence-based jewelry collocation detection and pushing method, which comprises the following steps:
determining a wearing style index of the user according to the current wearing image of the user;
acquiring a target image in a database, and determining the wearing style index of a person in the target image and the characteristic index of various characteristic labels corresponding to each ornament worn by the person, wherein the wearing style of the person in the target image is similar to the wearing style of a user in the current wearing image;
determining characteristic indexes of various characteristic labels corresponding to each ornament to be recommended, determining a seed point priority index of each ornament to be recommended according to the difference between the wearing style index of the person in the target image and the wearing style index of the user and the characteristic index of the same characteristic label corresponding to each ornament to be recommended, and screening out at least two seed point ornaments from the various ornaments to be recommended according to the seed point priority index;
determining the ornaments of interest of a user in various ornaments to be recommended according to the user observation time length of the various ornaments to be recommended, determining the similarity degree index between each ornament of interest of the user and each seed point ornament according to the difference of characteristic indexes of the same characteristic label corresponding to each ornament of interest of the user and each seed point ornament, and clustering the ornaments of interest of the user according to the similarity degree index to obtain at least two clusters;
determining the user attention degree corresponding to various feature labels in each cluster according to the difference of feature indexes of the feature labels corresponding to the user interested in each cluster and the user observation time length of the feature labels corresponding to the user interested in each cluster;
determining a recommendation sequence of the seed point ornaments according to the attention degree of the user and the difference of characteristic indexes of the same characteristic label corresponding to each of the seed point ornaments in each cluster, and recommending the seed point ornaments to the user according to the recommendation sequence.
Further, before determining each item to be recommended as a seed point priority indicator, the method further includes:
acquiring the recommendation priority of various ornaments to be recommended, and determining recommendation priority index values of the various ornaments to be recommended according to the recommendation priority, wherein the higher the recommendation priority is, the larger the value of the recommendation priority index values is;
and determining the priority index of each ornament to be recommended as a seed point according to the recommendation priority index value of each ornament to be recommended, the difference between the wearing style index of the person in the target image and the wearing style index of the user, and the difference between the characteristic index of the same characteristic label corresponding to each ornament to be recommended and each ornament worn by the person in the target image.
Further, determining each ornament to be recommended as a seed point priority index, wherein the corresponding calculation formula is as follows:
wherein,the method comprises the steps of representing an ith ornament to be recommended as a seed point priority index; />A recommendation priority index value representing an ith ornament to be recommended; />A wearing style index representing a user; />A pull-in style index representing a person in the target image; />Characteristic indexes of the j-th characteristic label corresponding to the x-th ornament worn by the person in the target image are represented; />Characteristic indexes of j characteristic labels corresponding to the ith ornament to be recommended are represented; the absolute value sign is taken; a represents the total number of accessories worn by the person in the target image; />Representing the total number of the same characteristic labels corresponding to the x-th ornament worn by the person in the target image and the i-th ornament to be recommended; />Representing an adjustment factor greater than 0.
Further, determining a similarity degree index between each ornament which is interested by the user and each seed point ornament, wherein a corresponding calculation formula is as follows:
wherein,a similarity index between the kth ornament of interest to the user and the nth seed point ornament; />Representing the characteristic index of the jth characteristic label corresponding to the nth seed point ornament; />The characteristic index of the j characteristic label corresponding to the kth ornament which is interested by the user is represented; the absolute value sign is taken; />Representing the total number of the same characteristic labels corresponding to the kth ornaments of interest to the user and the nth seed point ornaments; />Representing an adjustment factor greater than 0.
Further, determining the user attention degree corresponding to various feature labels in each cluster, wherein the corresponding calculation formula is as follows:
wherein,representing the user attention degree of the jth feature tag corresponding to the nth cluster; />The characteristic index of the jth characteristic label corresponding to the ornament which is interested by the user and is represented by the z in the nth cluster; />Representing the average value of the characteristic indexes of the j-th characteristic label corresponding to all the ornaments interested by the user in the nth cluster; the absolute value sign is taken; />Representing the user observation time length of the ornament of interest to the z-th user in the nth cluster; />Representing the total number of accessories of interest to all the users in the nth cluster; />Representing an adjustment factor greater than 0; />Representing the normalization function.
Further, determining a recommended order of the seed point ornaments, comprising:
determining a feature label corresponding to a user attention degree greater than a set attention degree threshold value in each cluster as a user favorite feature label, and determining the cluster with the user favorite feature label as a target cluster;
determining the number of user favorite feature labels corresponding to each seed point ornament according to the difference of feature indexes of the ornament which is interested by each user in each target cluster and the feature label which is corresponding to the same type of the user favorite feature label of each seed point ornament;
and sequencing the seed point ornaments according to the sequence from the big to the small of the number of the favorite feature labels of the user, thereby obtaining the recommended sequence of the seed point ornaments.
Further, determining the number of user favorite feature labels corresponding to each seed point ornament comprises:
calculating the sum of the absolute values of the differences of the characteristic indexes of the user favorite characteristic labels corresponding to the ornaments interested by the users and the seed point ornaments in each target cluster, so as to obtain characteristic index difference values;
determining the minimum value in the characteristic index difference value corresponding to each user favorite characteristic label in each target cluster, and classifying the corresponding user favorite characteristic label as the user favorite characteristic label of the seed point ornament corresponding to the minimum value;
and counting the number of the user favorite feature labels classified by each seed point ornament, thereby obtaining the number of the user favorite feature labels corresponding to each seed point ornament.
Further, at least two seed point ornaments are selected from various ornaments to be recommended, including:
according to the seed point priority index of each ornament to be recommended, determining the set number of ornaments to be recommended as seed point ornaments, wherein the seed point priority index of each seed point ornament is not smaller than the seed point priority index of other ornaments to be recommended which do not belong to the seed point ornaments.
Further, determining the ornaments of interest to the user among various ornaments to be recommended includes:
and determining the ornaments to be recommended corresponding to the user observation time length greater than the set time length threshold as the ornaments interested by the user according to the user observation time length of the various ornaments to be recommended.
Further, clustering the ornaments of interest to the user to obtain at least two clusters, including:
classifying each ornament of interest to the user into the category of the seed point ornament corresponding to the maximum similarity index according to the similarity index between each ornament of interest to the user and each seed point ornament, thereby obtaining at least two clusters.
The application has the following beneficial effects: according to the application, through analysis research of the buyer, the most suitable ornaments of the user can be accurately identified, and the recommendation rationality of the ornaments is improved. Specifically, in order to ensure that recommended ornaments are matched with the user wearing styles, the possibility of receiving recommended ornaments by the user is improved, all ornaments to be recommended of a special cabinet are respectively compared with characteristic indexes of the same characteristic label corresponding to each ornament worn by a person in a target image of the user similar wearing styles, and the difference between the wearing styles indexes of the person in the target image and the wearing styles indexes of the user is combined, so that seed point ornaments matched with the user wearing styles are screened out from all ornaments to be recommended. Meanwhile, combining actual observation time lengths of different jewelry by a user, determining jewelry which is most likely to be interested by the user, comparing similar conditions of the jewelry which is most likely to be interested with the seed point jewelry, and classifying the jewelry which is most likely to be interested into categories in which the seed point jewelry is located, so that a plurality of clusters are obtained. Because the interest of the customer to the ornaments is usually that the preference degree of one or a plurality of characteristics of the ornaments is higher, the difference of characteristic indexes of the ornaments which are interested by each user and the ornaments corresponding to the same characteristic label in each seed point in each cluster is analyzed, and the attention degree of the user to a certain characteristic label in each cluster is analyzed by combining the user observation time length of the ornaments which are interested by the user, so that the attention degree of the user corresponding to various characteristic labels in each cluster is determined. According to the attention degree of the user, analyzing the difference condition of the characteristic indexes of the characteristic labels of the ornaments of each seed point ornament and the ornaments of interest of the user, corresponding to the user, so as to determine the most reasonable recommending order of each seed point ornament, and recommending each seed point ornament to the user. According to the application, through accurately analyzing and researching the preference of the user, the proper ornaments of the user can be accurately identified finally, and the recommending rationality of the ornaments is effectively improved.
Drawings
In order to more clearly illustrate the embodiments of the application 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 application, 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 an artificial intelligence-based jewelry collocation detection and pushing method according to an embodiment of the present application.
Detailed Description
In order to further describe the technical means and effects adopted by the present application 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 application 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 application belongs. In addition, all parameters or indices in the formulas referred to herein are values after normalization that eliminate the dimensional effects.
In order to solve the problem of unreasonable recommendation of the existing ornaments, the ornaments can be different wearing articles such as ornaments, bags and scarves, and the ornaments are recommended as an example, the embodiment provides an artificial intelligence-based ornament collocation detection and pushing method, and a flow chart corresponding to the method is shown in fig. 1, and the method comprises the following steps:
step S1: and determining the wearing style index of the user according to the current wearing image of the user.
When a user enters the intelligent ornament special cabinet, the intelligent ornament special cabinet in the implementation refers to the intelligent ornament special cabinet, and the camera set at the proper position of the intelligent ornament special cabinet is used for collecting the image worn by the user on the same day, so that the current worn image of the user is obtained. According to the current wearing image of the user, searching the most similar wearing image in the Internet by utilizing a neural network technology, so as to determine the wearing style of the user, and determining the corresponding wearing style index according to the wearing style. The wearing style refers to the type of clothing wearing, including gentlewoman, college, garden, hip hop, neutrality and the like, and each wearing style is quantized, so that wearing style indexes corresponding to each wearing style are obtained. For example, the various wearing styles may be sorted in a gradual change order, and the sort numbers corresponding to the various wearing styles may be used as the corresponding wearing style index. It should be understood that, since the purpose of quantifying the type of the wearing style is to ensure that the larger the wearing style difference is, the larger the corresponding quantization index is, that is, the wearing style index difference is, the specific quantization mode of the wearing style type is not limited here.
Step S2: and acquiring a target image in a database, and determining the wearing style index of the person in the target image and the characteristic index of each ornament worn by the person corresponding to various characteristic labels, wherein the wearing style of the person in the target image is similar to the wearing style of the user in the current wearing image.
And searching a target image similar to the wearing style of the user in the current wearing image in the Internet by utilizing a neural network technology. When the target image is determined, the similarity degree of each picture stored in the internet and the wearing style of the user in the current wearing image and the praise condition of each picture can be comprehensively considered, and pictures with larger similarity degree and more praise number are selected as the target image, wherein the bias degree of the similarity degree and the praise number can be set according to the needs, and the method is not limited.
After the target image is determined, the punch-through style index of the model in the target image is determined in the same manner as the punch-through style index of the user is determined. Meanwhile, various characteristic labels of each ornament worn by the model in the target image, namely, the ornaments, are identified by utilizing an image detection technology, so that characteristic indexes of the various characteristic labels are determined. The characteristic tag refers to other detail characteristics such as wearing position, color, shape and the like of the ornament. And (3) assigning a plurality of characteristic labels to each ornament, and quantifying the characteristic labels to obtain the characteristic index corresponding to each characteristic label. Taking the characteristic label as an example, numbering a plurality of common colors of the ornaments according to a color gradual change sequence, and normalizing each number to be in a range of 0-1, thereby obtaining the characteristic index of each color corresponding to the characteristic label of the color.
Step S3: determining characteristic indexes of various characteristic labels corresponding to each ornament to be recommended, determining a seed point priority index of each ornament to be recommended according to the difference between the wearing style index of the person in the target image and the wearing style index of the user and the characteristic index of the same characteristic label corresponding to each ornament to be recommended, and screening at least two seed point ornaments from the various ornaments to be recommended according to the seed point priority index.
The method comprises the steps of obtaining images of all ornaments in an intelligent ornament special cabinet, namely, the ornaments to be recommended, and determining the characteristic indexes of each ornament to be recommended corresponding to various characteristic labels according to the same mode of determining the characteristic indexes of each ornament corresponding to various characteristic labels worn by a model in a target image.
Through the steps, the feature indexes of various feature labels corresponding to each ornament to be recommended can be determined, and as each ornament has various feature indexes, a user may like one or more features of any ornament, and at the moment, the most suitable ornament can be recommended for the user by analyzing the favorite ornament features of the user. In order to achieve the purpose of recommending the most suitable ornaments for the user, the ornaments interested by the user can be clustered according to the characteristic attention degree of the user to different types of ornaments by using a clustering mode. To achieve clustering, it is necessary to determine individual seed point ornaments as a cluster center.
Because a plurality of different ornaments are placed in the intelligent ornament special cabinet by a merchant, and the recommendation degree of each ornament is different according to the marketing scheme of the merchant, in order to meet the marketing means of the merchant while promoting the purchasing desire of a user, the ornaments in the special cabinet are required to be firstly subjected to recommendation priority sorting according to the demands of the merchant, the recommendation priority sorting is performed according to the order of the recommendation priorities from high to low, so that the recommendation priority sorting of various ornaments is obtained, the recommendation priority index value of various ornaments is determined according to the recommendation priority sorting, and the corresponding calculation formula is as follows:
wherein,indicating the i-th ornament to be recommended, namely, the recommendation priority index value of the ornament; />Representing the total number of accessories to be recommended; />And the serial number of the ith ornament to be recommended in the recommendation priority ranking is represented.
For the calculation formula of the recommendation priority index value of the ith ornament to be recommended, when the recommendation priority of the ornament is higher, the smaller the serial number of the ornament in the recommendation priority sorting is, and at the moment, the larger the recommendation priority index value corresponding to the ornament is.
Through the steps, for all ornaments to be recommended in the intelligent ornament special cabinet, the intelligent ornament special cabinet not only has a plurality of characteristic indexes corresponding to various characteristic labels, but also has a recommendation priority index value. At this time, by comparing the difference between the wearing style index of the model in the target image and the wearing style index of the user, and the difference between the characteristic indexes of the same characteristic tag corresponding to each ornament worn by the model in the target image and each ornament to be recommended, and combining the recommendation priority index values of various ornaments, determining that each ornament to be recommended is used as a seed point priority index, and the corresponding calculation formula is as follows:
wherein,indicating the ith ornament to be recommended, namely, the ornament as a seed point priority index; />A recommendation priority index value representing an ith ornament to be recommended; />A wearing style index representing a user; />A pull-in style index representing a person in the target image; />Characteristic indexes of the j-th characteristic label corresponding to the x-th ornament worn by the person in the target image are represented; />Characteristic indexes of j characteristic labels corresponding to the ith ornament to be recommended are represented; the absolute value sign is taken; a represents the total number of accessories worn by the person in the target image; />Representing the total number of the same characteristic labels corresponding to the x-th ornament worn by the person in the target image and the i-th ornament to be recommended; />Represents an adjustment coefficient greater than 0 for preventing the denominator from being 0.
In the calculation formula of the ith ornament to be recommended as the seed point priority index, when the wearing style of the character in the target image is more similar to the current wearing style of the user, namelyThe larger the recommendation priority index value ++>The larger the ornament is, and the smaller the cumulative difference degree of the same type of characteristic indexes of the ornament is, namelyWhen the weight is larger, the higher the priority of the ith ornament in the special cabinet as a clustering center is, the larger the value of the priority index corresponding to the seed point is.
It should be understood that, when the marketing scheme is not formulated by the merchant, that is, the recommended degrees of all ornaments of the special cabinet are the same, as other embodiments, in determining the seed point priority index of each ornament, the recommended priority index value in the calculation formula as the seed point priority index may be set to 1, and at this time, the determined seed point priority index of each ornament is no longer affected by different recommended degrees of each ornament.
After determining each ornament to be recommended as a seed point priority index, screening at least two seed point ornaments from various ornaments to be recommended according to the seed point priority index, namely: according to the seed point priority index of each ornament to be recommended, determining the set number of ornaments to be recommended as seed point ornaments, wherein the seed point priority index of each seed point ornament is not smaller than the seed point priority index of other ornaments to be recommended which do not belong to the seed point ornaments. That is, each ornament to be recommended is arranged according to the order from the big to the small as the seed point priority index, n ornaments which are set before are selected as optimal recommended ornaments according to the merchant demand, and the n optimal recommended ornaments are used as seed point ornaments. The set number n can be set according to the requirement, and the value of the set number n is set to be 10 in the embodiment.
Step S4: determining the ornaments of interest of a user in various ornaments to be recommended according to the user observation time length of the various ornaments to be recommended, determining the similarity degree index between each ornament of interest of the user and each seed point ornament according to the difference of characteristic indexes of the same characteristic label corresponding to each ornament of interest of the user and each seed point ornament, and clustering the ornaments of interest of the user according to the similarity degree index to obtain at least two clusters.
Before a user purchases an accessory, when the user is interested in a certain accessory, the user stays in front of the special cabinet for a long time, and carefully observes the appearance of the certain accessory. And when the user is interested in a certain characteristic of the ornaments, other ornaments with similar characteristics with ornaments with longer stay time are searched in the special cabinet. Thus, by existing line-of-sight acquisition techniques, the jewelry object observed by the user each time is determined and the time of each observation is recorded. However, considering that the user stays for a long time to have contingency, namely, because the sight line may sweep a certain uninteresting ornament during searching, record error data is generated, in order to eliminate contingency, the sight line is gathered to any ornament for 1s and then is used as a record time node, and record observation time is started, so that the user observation time of various ornaments to be recommended can be obtained. For a good that is not observed by the user, the user's observation period of the good may be considered to be 0.
Determining the ornaments of interest to the user according to the user observation time length of various ornaments to be recommended, namely: and determining the ornaments to be recommended corresponding to the user observation time length greater than the set time length threshold as the ornaments interested by the user according to the user observation time length of the various ornaments to be recommended. The set duration threshold may be set empirically, and in this embodiment, the value of the set duration threshold is set to 5s. According to the mode, N ornaments with longer stay time of the current special cabinet, namely N ornaments interested by the users, can be determined.
On the basis of the steps, according to the difference between each characteristic index of the ornaments interested by each user and the characteristic index of the same characteristic label corresponding to each seed point ornament, determining the similarity index between the ornaments interested by each user and each seed point ornament, wherein the corresponding calculation formula is as follows:
wherein,a similarity index between the kth ornament of interest to the user and the nth seed point ornament; />Representing the characteristic index of the jth characteristic label corresponding to the nth seed point ornament; />The characteristic index of the j characteristic label corresponding to the kth ornament which is interested by the user is represented; the absolute value sign is taken; />Representing the total number of the same characteristic labels corresponding to the kth ornaments of interest to the user and the nth seed point ornaments; />Represents an adjustment coefficient greater than 0 for preventing the denominator from being 0.
For the calculation formula of the similarity index between the decoration of interest of the kth user and the decoration of the nth seed point, when the average characteristic index difference of the same characteristic label corresponding to the decoration of interest of the user and the decoration of the seed point is smaller, the more similar the decoration of interest of the user and the decoration of the seed point are, the more likely the decoration of interest of the user and the decoration of the seed point are, and the larger the value of the corresponding similarity index is.
After determining the similarity index between the ornaments of interest of each user and each seed point ornament, clustering all the ornaments of interest of the user to obtain at least two clusters, namely: classifying each ornament of interest to the user into the category of the seed point ornament corresponding to the maximum similarity index according to the similarity index between each ornament of interest to the user and each seed point ornament, thereby obtaining at least two clusters. In this embodiment, since the number of seed point ornaments is 10, 10 clusters can be finally obtained.
Step S5: and determining the user attention degree corresponding to various feature labels in each cluster according to the difference of the feature indexes of the feature labels corresponding to the user interested ornaments in each cluster and the user observation time length of the feature labels corresponding to the user interested ornaments in each cluster.
Although a plurality of clusters are determined in the above manner, in practice, the user has a longer residence time in front of a particular ornament because of a higher degree of attention to one or more characteristics of the ornament, and the user needs to select his own favorite ornament by comparing and selecting. Therefore, in order to recommend the most suitable ornaments to promote the purchase of users and to purchase the hot ornaments of merchants, according to the clustering result, the attention degree of the users to a certain characteristic of a certain ornament can be obtained, and the corresponding calculation formula is as follows:
wherein,representing the user attention degree of the jth feature tag corresponding to the nth cluster; />Indicating that the jth ornament of interest to the z-th user in the nth cluster corresponds to the jth ornamentCharacteristic indexes of the characteristic labels; />Representing the average value of the characteristic indexes of the j-th characteristic label corresponding to all the ornaments interested by the user in the nth cluster; the absolute value sign is taken; />Representing the user observation time length of the ornament of interest to the z-th user in the nth cluster; />Representing the total number of accessories of interest to all the users in the nth cluster; />Represents an adjustment coefficient greater than 0 for preventing the denominator from being 0; />Representing the normalization function.
For the calculation formula of the user attention degree corresponding to the jth feature tag in the nth cluster, when the difference degree between the feature index corresponding to the jth feature tag and the feature index mean value of all the ornaments interested by the user in the cluster is smaller, the user is shown to pay more attention to the jth feature tag, so that the difference degree is mapped in a negative correlation manner, and a negative correlation mapping result is obtained. At the same time, the user observation time length is utilized to obtain the negative correlation mapping resultWeighting is performed, and the longer the user observation time is, the greater the attention degree of the user to a certain feature in the ornament is indicated. Finally, when the difference degree between the characteristic index and the mean value of the ornament possibly liked by the user in the cluster is smaller, and the observation time is longer, the overall value is longer, which means that the attention degree of the user to a certain characteristic of the ornament is longer, and the corresponding user is closedThe greater the value of the degree of injection.
Step S6: determining a recommendation sequence of the seed point ornaments according to the attention degree of the user and the difference of characteristic indexes of the same characteristic label corresponding to each of the seed point ornaments in each cluster, and recommending the seed point ornaments to the user according to the recommendation sequence.
After determining the user attention degree corresponding to various feature labels in each cluster, comparing the user attention degree with a set attention degree threshold, and combining the differences of feature indexes of the same feature label corresponding to each seed point ornament and the ornaments interested by each user in each cluster, thereby determining the recommendation sequence of each seed point ornament, wherein the implementation steps comprise:
determining a feature label corresponding to a user attention degree greater than a set attention degree threshold value in each cluster as a user favorite feature label, and determining the cluster with the user favorite feature label as a target cluster;
determining the number of user favorite feature labels corresponding to each seed point ornament according to the difference of feature indexes of the ornament which is interested by each user in each target cluster and the feature label which is corresponding to the same type of the user favorite feature label of each seed point ornament;
and sequencing the seed point ornaments according to the sequence from the big to the small of the number of the favorite feature labels of the user, thereby obtaining the recommended sequence of the seed point ornaments.
Preferably, determining the number of user favorite feature labels corresponding to each seed point ornament includes:
calculating the sum of the absolute values of the differences of the characteristic indexes of the user favorite characteristic labels corresponding to the ornaments interested by the users and the seed point ornaments in each target cluster, so as to obtain characteristic index difference values;
determining the minimum value in the characteristic index difference value corresponding to each user favorite characteristic label in each target cluster, and classifying the corresponding user favorite characteristic label as the user favorite characteristic label of the seed point ornament corresponding to the minimum value;
and counting the number of the user favorite feature labels classified by each seed point ornament, thereby obtaining the number of the user favorite feature labels corresponding to each seed point ornament.
Specifically, the set attention degree threshold is set reasonably, and the value of the set attention degree threshold is set to be 0.7 in this embodiment. And screening the feature labels in each cluster by using the set attention degree threshold, and determining the feature labels corresponding to the user attention degree larger than the set attention degree threshold in each cluster as the user favorite feature labels. Traversing all the cluster clusters to obtain the feature indexes of a plurality of the same or different user favorite feature labels of each cluster. Determining the cluster with the user favorite feature labels as a target cluster, and calculating the feature index of each user favorite feature label corresponding to the adornment interested by each user in the target cluster for any target cluster, and the sum of the absolute value of the difference values of the feature index of the seed point adornment corresponding to the same user favorite feature label in each cluster to obtain the feature index difference value of the seed point adornment corresponding to the user favorite feature label. Since the number of the seed point ornaments in the embodiment is 10, 10 characteristic index difference values can be obtained for any user favorite characteristic label in the target cluster, and the minimum value in the 10 characteristic index difference values is selected, and the seed point ornaments corresponding to the minimum value obtain a vote. According to the characteristic index difference value corresponding to each user favorite characteristic label in each target cluster, the voting number of each seed point ornament, namely the number of the user favorite characteristic labels corresponding to each seed point ornament, can be determined by voting the seed point ornaments. And sequencing each seed point ornament according to the number of votes, namely the sequence from big to small of the number of favorite feature labels of the user, so as to obtain the recommended sequence of each seed point ornament. And sequentially recommending each seed point ornament to the user according to the recommendation sequence.
According to the application, all ornaments to be recommended of the special cabinet are respectively compared with the characteristic indexes of the same characteristic label corresponding to each ornament worn by the person in the target image of the similar wearing style of the user, so that seed point ornaments matched with the wearing style of the user are screened out from all ornaments to be recommended. Meanwhile, combining actual observation time lengths of different ornaments by a user, determining the most possibly interested ornaments of the user, comparing the most possibly interested ornaments with similar conditions of the seed point ornaments, and classifying the most possibly interested ornaments into categories of the seed point ornaments, so as to obtain a plurality of clusters. The attention degree of the user to a certain characteristic label in each cluster is analyzed, and finally, the most probable favorite ornaments of the user in the special cabinet are judged and recommended to the user, so that the problem that pushed ornaments cannot be attached to the favorite ornaments of the user is solved, and the purchase possibility of the user and the practicability of the intelligent ornament special cabinet are improved.
It should be noted that: the above embodiments are only for illustrating the technical solution of the present application, and not for limiting the same; although the application 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 and scope of the technical solutions of the embodiments of the present application, and are intended to be included in the scope of the present application.
Claims (5)
1. The jewelry collocation detection and pushing method based on artificial intelligence is characterized by comprising the following steps of:
determining a wearing style index of the user according to the current wearing image of the user;
acquiring a target image in a database, and determining the wearing style index of a person in the target image and the characteristic index of various characteristic labels corresponding to each ornament worn by the person, wherein the wearing style of the person in the target image is similar to the wearing style of a user in the current wearing image;
determining characteristic indexes of various characteristic labels corresponding to each ornament to be recommended, determining a seed point priority index of each ornament to be recommended according to the difference between the wearing style index of the person in the target image and the wearing style index of the user and the characteristic index of the same characteristic label corresponding to each ornament to be recommended, and screening out at least two seed point ornaments from the various ornaments to be recommended according to the seed point priority index;
determining the ornaments of interest of a user in various ornaments to be recommended according to the user observation time length of the various ornaments to be recommended, determining the similarity degree index between each ornament of interest of the user and each seed point ornament according to the difference of characteristic indexes of the same characteristic label corresponding to each ornament of interest of the user and each seed point ornament, and clustering the ornaments of interest of the user according to the similarity degree index to obtain at least two clusters;
determining the user attention degree corresponding to various feature labels in each cluster according to the difference of feature indexes of the feature labels corresponding to the user interested in each cluster and the user observation time length of the feature labels corresponding to the user interested in each cluster;
determining a recommendation order of the seed point ornaments according to the attention degree of the user and the difference of characteristic indexes of the same characteristic label corresponding to each of the ornaments interested by the user in each cluster and each seed point ornament, and recommending the seed point ornaments to the user according to the recommendation order;
before determining each item to be recommended as a seed point priority indicator, the method further comprises:
acquiring the recommendation priority of various ornaments to be recommended, and determining recommendation priority index values of the various ornaments to be recommended according to the recommendation priority, wherein the higher the recommendation priority is, the larger the value of the recommendation priority index values is;
determining the priority index of each ornament to be recommended as a seed point according to the recommendation priority index value of each ornament to be recommended, the difference between the wearing style index of the person in the target image and the wearing style index of the user, and the difference between the characteristic index of the same characteristic label corresponding to each ornament to be recommended and each ornament worn by the person in the target image;
determining each ornament to be recommended as a seed point priority index, wherein the corresponding calculation formula is as follows:
wherein,the method comprises the steps of representing an ith ornament to be recommended as a seed point priority index; />A recommendation priority index value representing an ith ornament to be recommended; />A wearing style index representing a user; />A pull-in style index representing a person in the target image; />Characteristic indexes of the j-th characteristic label corresponding to the x-th ornament worn by the person in the target image are represented;characteristic indexes of j characteristic labels corresponding to the ith ornament to be recommended are represented; the absolute value sign is taken; a represents the person wearing the target imageThe total number of ornaments; />Representing the total number of the same characteristic labels corresponding to the x-th ornament worn by the person in the target image and the i-th ornament to be recommended; />Representing an adjustment factor greater than 0;
determining a similarity degree index between each ornament which is interested by the user and each seed point ornament, wherein a corresponding calculation formula is as follows:
wherein,a similarity index between the kth ornament of interest to the user and the nth seed point ornament; />Representing the characteristic index of the jth characteristic label corresponding to the nth seed point ornament; />The characteristic index of the j characteristic label corresponding to the kth ornament which is interested by the user is represented; the absolute value sign is taken; />Representing the total number of the same characteristic labels corresponding to the kth ornaments of interest to the user and the nth seed point ornaments; />Representing an adjustment factor greater than 0;
determining the user attention degree corresponding to various feature labels in each cluster, wherein the corresponding calculation formula is as follows:
wherein,representing the user attention degree of the jth feature tag corresponding to the nth cluster; />The characteristic index of the jth characteristic label corresponding to the ornament which is interested by the user and is represented by the z in the nth cluster; />Representing the average value of the characteristic indexes of the j-th characteristic label corresponding to all the ornaments interested by the user in the nth cluster; the absolute value sign is taken; />Representing the user observation time length of the ornament of interest to the z-th user in the nth cluster; />Representing the total number of accessories of interest to all the users in the nth cluster; />Representing an adjustment factor greater than 0; />Representing a normalization function;
determining a recommended order of the seed point ornaments, comprising:
determining a feature label corresponding to a user attention degree greater than a set attention degree threshold value in each cluster as a user favorite feature label, and determining the cluster with the user favorite feature label as a target cluster;
determining the number of user favorite feature labels corresponding to each seed point ornament according to the difference of feature indexes of the ornament which is interested by each user in each target cluster and the feature label which is corresponding to the same type of the user favorite feature label of each seed point ornament;
and sequencing the seed point ornaments according to the sequence from the big to the small of the number of the favorite feature labels of the user, thereby obtaining the recommended sequence of the seed point ornaments.
2. The method for detecting and pushing jewelry collocation based on artificial intelligence according to claim 1, wherein determining the number of user favorite feature labels corresponding to each seed point jewelry comprises:
calculating the sum of the absolute values of the differences of the characteristic indexes of the user favorite characteristic labels corresponding to the ornaments interested by the users and the seed point ornaments in each target cluster, so as to obtain characteristic index difference values;
determining the minimum value in the characteristic index difference value corresponding to each user favorite characteristic label in each target cluster, and classifying the corresponding user favorite characteristic label as the user favorite characteristic label of the seed point ornament corresponding to the minimum value;
and counting the number of the user favorite feature labels classified by each seed point ornament, thereby obtaining the number of the user favorite feature labels corresponding to each seed point ornament.
3. The artificial intelligence-based jewelry collocation detection and pushing method of claim 1, wherein at least two seed point ornaments are selected from various ornaments to be recommended, comprising:
according to the seed point priority index of each ornament to be recommended, determining the set number of ornaments to be recommended as seed point ornaments, wherein the seed point priority index of each seed point ornament is not smaller than the seed point priority index of other ornaments to be recommended which do not belong to the seed point ornaments.
4. The artificial intelligence-based jewelry collocation detection and pushing method according to claim 1, wherein determining the jewelry of interest to the user among various types of jewelry to be recommended comprises:
and determining the ornaments to be recommended corresponding to the user observation time length greater than the set time length threshold as the ornaments interested by the user according to the user observation time length of the various ornaments to be recommended.
5. The jewelry collocation detection and pushing method based on artificial intelligence of claim 1, wherein clustering the jewelry of interest to the user to obtain at least two clusters comprises:
classifying each ornament of interest to the user into the category of the seed point ornament corresponding to the maximum similarity index according to the similarity index between each ornament of interest to the user and each seed point ornament, thereby obtaining at least two clusters.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311108051.1A CN116823361B (en) | 2023-08-31 | 2023-08-31 | Jewelry collocation detection and pushing method based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311108051.1A CN116823361B (en) | 2023-08-31 | 2023-08-31 | Jewelry collocation detection and pushing method based on artificial intelligence |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116823361A CN116823361A (en) | 2023-09-29 |
| CN116823361B true CN116823361B (en) | 2023-12-12 |
Family
ID=88117055
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311108051.1A Active CN116823361B (en) | 2023-08-31 | 2023-08-31 | Jewelry collocation detection and pushing method based on artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116823361B (en) |
Citations (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20090074720A (en) * | 2009-06-29 | 2009-07-07 | 최명렬 | Program recording medium |
| CN107993131A (en) * | 2017-12-27 | 2018-05-04 | 广东欧珀移动通信有限公司 | Wear to take and recommend method, apparatus, server and storage medium |
| JP2018120527A (en) * | 2017-01-27 | 2018-08-02 | 株式会社リコー | Image processing apparatus, image processing method, and image processing system |
| CN108564439A (en) * | 2018-03-28 | 2018-09-21 | 深圳市零度智控科技有限公司 | Dress ornament recommends method and system, terminal and storage medium |
| CN109117779A (en) * | 2018-08-06 | 2019-01-01 | 百度在线网络技术（北京）有限公司 | One kind, which is worn, takes recommended method, device and electronic equipment |
| CN109146626A (en) * | 2018-08-14 | 2019-01-04 | 中山大学 | A kind of fashion clothing collocation recommended method based on user's dynamic interest analysis |
| CN109165997A (en) * | 2018-07-19 | 2019-01-08 | 阿里巴巴集团控股有限公司 | It does shopping under a kind of line the generation method and device of recommendation |
| WO2019134560A1 (en) * | 2018-01-08 | 2019-07-11 | Oppo广东移动通信有限公司 | Method for constructing matching model, clothing recommendation method and device, medium, and terminal |
| CN110110181A (en) * | 2019-05-09 | 2019-08-09 | 湖南大学 | A kind of garment coordination recommended method based on user styles and scene preference |
| KR20190115510A (en) * | 2018-03-15 | 2019-10-14 | 주식회사 비주얼 | Method and electric apparatus for recommending jewelry product |
| KR102178961B1 (en) * | 2020-04-21 | 2020-11-13 | 주식회사 스타일쉐어 | Artificial neural network apparatus and method for recommending fashion item using user clustering |
| CN112150239A (en) * | 2020-09-10 | 2020-12-29 | 浙江网安文化发展有限公司 | Wearing image information recommendation method and device |
| CN112163930A (en) * | 2020-09-27 | 2021-01-01 | 深圳莱尔托特科技有限公司 | Matching recommendation method and matching recommendation system |
| CN113935790A (en) * | 2020-06-29 | 2022-01-14 | 青岛海尔洗衣机有限公司 | Putting-through recommendation method and device, electronic equipment and storage medium |
| KR102364533B1 (en) * | 2021-10-05 | 2022-02-18 | 김의현 | Method for interior service based on artificial intelligence and apparatus for performing the method |
| KR20220039697A (en) * | 2020-12-28 | 2022-03-29 | 오드컨셉 주식회사 | Method, apparatus and computer program for style recommendation |
| KR20220111592A (en) * | 2021-02-02 | 2022-08-09 | 주식회사 패션에이드 | Fashion coordination style recommendation system and method by artificial intelligence |
| CN115115232A (en) * | 2022-07-01 | 2022-09-27 | 鑫洋互联网科技(广州)有限公司 | Business information analysis and pushing method embedded into campus life |
| CN115495643A (en) * | 2022-07-25 | 2022-12-20 | 张跃慧 | Intelligent wearing APP |
| CN115905593A (en) * | 2022-10-27 | 2023-04-04 | 浙江大学 | Method and system for recommending existing clothes to be worn and put on based on current season style |
| CN115935064A (en) * | 2022-12-05 | 2023-04-07 | 北京宾理信息科技有限公司 | Accessory recommendation method and control method for vehicle |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9639880B2 (en) * | 2009-12-17 | 2017-05-02 | Google Inc. | Photorealistic recommendation of clothing and apparel based on detected web browser input and content tag analysis |
| CN108230082A (en) * | 2017-06-16 | 2018-06-29 | 深圳市商汤科技有限公司 | The recommendation method and apparatus of collocation dress ornament, electronic equipment, storage medium |
| US11195221B2 (en) * | 2019-12-13 | 2021-12-07 | The Mada App, LLC | System rendering personalized outfit recommendations |
-
2023
- 2023-08-31 CN CN202311108051.1A patent/CN116823361B/en active Active
Patent Citations (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20090074720A (en) * | 2009-06-29 | 2009-07-07 | 최명렬 | Program recording medium |
| JP2018120527A (en) * | 2017-01-27 | 2018-08-02 | 株式会社リコー | Image processing apparatus, image processing method, and image processing system |
| CN107993131A (en) * | 2017-12-27 | 2018-05-04 | 广东欧珀移动通信有限公司 | Wear to take and recommend method, apparatus, server and storage medium |
| WO2019134560A1 (en) * | 2018-01-08 | 2019-07-11 | Oppo广东移动通信有限公司 | Method for constructing matching model, clothing recommendation method and device, medium, and terminal |
| CN110021061A (en) * | 2018-01-08 | 2019-07-16 | 广东欧珀移动通信有限公司 | Collocation model building method, dress ornament recommended method, device, medium and terminal |
| KR20190115510A (en) * | 2018-03-15 | 2019-10-14 | 주식회사 비주얼 | Method and electric apparatus for recommending jewelry product |
| CN108564439A (en) * | 2018-03-28 | 2018-09-21 | 深圳市零度智控科技有限公司 | Dress ornament recommends method and system, terminal and storage medium |
| CN109165997A (en) * | 2018-07-19 | 2019-01-08 | 阿里巴巴集团控股有限公司 | It does shopping under a kind of line the generation method and device of recommendation |
| CN109117779A (en) * | 2018-08-06 | 2019-01-01 | 百度在线网络技术（北京）有限公司 | One kind, which is worn, takes recommended method, device and electronic equipment |
| CN109146626A (en) * | 2018-08-14 | 2019-01-04 | 中山大学 | A kind of fashion clothing collocation recommended method based on user's dynamic interest analysis |
| CN110110181A (en) * | 2019-05-09 | 2019-08-09 | 湖南大学 | A kind of garment coordination recommended method based on user styles and scene preference |
| KR102178961B1 (en) * | 2020-04-21 | 2020-11-13 | 주식회사 스타일쉐어 | Artificial neural network apparatus and method for recommending fashion item using user clustering |
| CN113935790A (en) * | 2020-06-29 | 2022-01-14 | 青岛海尔洗衣机有限公司 | Putting-through recommendation method and device, electronic equipment and storage medium |
| CN112150239A (en) * | 2020-09-10 | 2020-12-29 | 浙江网安文化发展有限公司 | Wearing image information recommendation method and device |
| CN112163930A (en) * | 2020-09-27 | 2021-01-01 | 深圳莱尔托特科技有限公司 | Matching recommendation method and matching recommendation system |
| KR20220039697A (en) * | 2020-12-28 | 2022-03-29 | 오드컨셉 주식회사 | Method, apparatus and computer program for style recommendation |
| KR20220111592A (en) * | 2021-02-02 | 2022-08-09 | 주식회사 패션에이드 | Fashion coordination style recommendation system and method by artificial intelligence |
| KR102364533B1 (en) * | 2021-10-05 | 2022-02-18 | 김의현 | Method for interior service based on artificial intelligence and apparatus for performing the method |
| CN115115232A (en) * | 2022-07-01 | 2022-09-27 | 鑫洋互联网科技(广州)有限公司 | Business information analysis and pushing method embedded into campus life |
| CN115495643A (en) * | 2022-07-25 | 2022-12-20 | 张跃慧 | Intelligent wearing APP |
| CN115905593A (en) * | 2022-10-27 | 2023-04-04 | 浙江大学 | Method and system for recommending existing clothes to be worn and put on based on current season style |
| CN115935064A (en) * | 2022-12-05 | 2023-04-07 | 北京宾理信息科技有限公司 | Accessory recommendation method and control method for vehicle |
Non-Patent Citations (2)
| Title |
|---|
| 基于图像理解的智能服饰搭配研究;吴晓军;中国优秀硕士学位论文全文数据库 工程科技Ⅰ辑(第02期);B025-20 * |
| 基于标签的商品推荐模型研究;涂海丽;唐晓波;;数据分析与知识发现(第09期);28-39 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116823361A (en) | 2023-09-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102609523B (en) | The collaborative filtering recommending method classified based on taxonomy of goods and user | |
| US7613687B2 (en) | Systems and methods for enhancing web-based searching | |
| US11062379B2 (en) | Automatic fashion outfit composition and recommendation system and method | |
| US20100313141A1 (en) | System and Method for Learning User Genres and Styles and for Matching Products to User Preferences | |
| CN109189904A (en) | Individuation search method and system | |
| CN102855268A (en) | Image ranking method and system based on attribute correlation | |
| EP2438509A1 (en) | System and method for learning user genres and styles and matching products to user preferences | |
| US11157988B2 (en) | System and method for fashion recommendations | |
| KR102284148B1 (en) | Method and System of recommending fashion based on vector based deep learning | |
| KR102109338B1 (en) | Customized clothing analysis and recommendation system | |
| CN111724235A (en) | Online commodity recommendation method based on user novelty | |
| CN113450178A (en) | Internet-based e-commerce accurate marketing system | |
| CN116385048B (en) | Intelligent marketing method and system for agricultural products | |
| CN115496566B (en) | Regional specialty recommendation method and system based on big data | |
| CN113159826A (en) | Garment fashion element prediction system and method based on deep learning | |
| CN115392947A (en) | Demand prediction method and device | |
| CN112307333A (en) | Intelligent vehicle purchasing recommendation method based on machine vision and weighted KNN | |
| Tahir et al. | E-commerce platform based on Machine Learning Recommendation System | |
| Boutemedjet et al. | Predictive approach for user long-term needs in content-based image suggestion | |
| CN116823361B (en) | Jewelry collocation detection and pushing method based on artificial intelligence | |
| CN116805023A (en) | Takeaway recommendation method based on large language model | |
| CN111951080A (en) | Method and system for integrating artificial intelligence into platform | |
| US11526925B2 (en) | System and method for fashion recommendations | |
| US11430043B2 (en) | System and method for fashion recommendations | |
| CN114971805A (en) | Electronic commerce platform commodity intelligent analysis recommendation system based on deep learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |