TWI852055B - Apparel size recommendation method and system - Google Patents

Abstract

An apparel size recommendation method and system are provided. The method includes the following steps. Basic body information of a user is obtained. Body circumference data of the user is determined according to the basic body information and common body model data. A plurality of first model parameters of a size recommendation model are determined according to a first apparel product, so as to apply the first model parameters corresponding to the first apparel product to the size recommendation model. A recommended size of the first apparel product is determined by using the size recommendation model according to the body circumference data of the user and apparel characteristic data of the first apparel product. The recommended size of the first apparel product is provided to the user.

Description

Clothing size recommendation method and system

The present invention relates to a clothing recommendation method, and in particular to a clothing size recommendation method and system.

With the booming development of e-commerce, online shopping has become an indispensable way of shopping in modern people's daily life. For digital channels, since consumers cannot actually see the physical product, how to accurately convey the product information to consumers is a very important part that affects the online shopping experience. In particular, for the online sales of clothing products, since consumers cannot actually try on the clothing they want to buy, they often encounter the dilemma of not knowing how to choose the size of the clothing.

Currently, clothing sellers provide consumers with various measurements of clothing products, such as the width or length of clothing products. They even provide models’ fitting reports for consumers to refer to. However, even if consumers can refer to the above information provided by the seller, they may still not be able to successfully select the appropriate clothing size due to various factors. The above factors may include the difference in body shape between the model and the consumer or the consumer’s lack of understanding of the detailed data of their own body shape. On the other hand, although there are services in the market that use big data analysis technology to recommend clothing sizes suitable for consumers, it is not easy to map consumers' body information to appropriate clothing sizes based on limited transaction data for clothing products with diverse styles.

In view of this, the present invention proposes a clothing size recommendation method and system, which can provide size recommendations that meet consumer needs for diversified clothing products.

The embodiment of the present invention provides a clothing size recommendation method, which includes the following steps. Obtaining basic human body information of a user. Determining the user's body circumference data based on the basic human body information and public body model data. Determining a first model parameter of a size recommendation model based on a first clothing product, so as to apply the first model parameter corresponding to the first clothing product to the size recommendation model. Determining the recommended size of the first clothing product based on the user's body circumference data and the clothing characteristic data of the first clothing product by using the size recommendation model. Providing the recommended size of the first clothing product to the user.

The embodiment of the present invention provides a clothing size recommendation system, which includes a storage device and a processor. The processor is coupled to the storage device and configured to perform the following steps. Obtain basic human body information of a user. Determine the user's body circumference data based on the basic human body information and public body model data. Determine a first model parameter of a size recommendation model based on a first clothing product, so as to apply the first model parameter corresponding to the first clothing product to the size recommendation model. Determine the recommended size of the first clothing product based on the user's body circumference data and the clothing characteristic data of the first clothing product by using the size recommendation model. Provide the recommended size of the first clothing product to the user.

Based on the above, in the embodiment of the present invention, the human body circumference data can be generated according to the basic human body information provided by the user, and the model parameters used by the size recommendation model can be changed according to different clothing products. Based on this, for clothing products with diverse styles, the recommended size suitable for the user can be determined through the size recommendation model based on the detailed human body circumference data and the exclusive model parameters, thereby optimizing the online shopping experience of clothing products.

110: Clothing size recommendation system

111: Storage device

112: Processor

120: User device

M1: Size recommended model

P_1~P_n: matching information

A_d: Clothing characteristic data

Body_info: human body measurement data

S210~S250, S402~S420, S502~S518, S602~S618, S802~S814: Steps

Figure 1 is a schematic diagram of a clothing size recommendation system according to an embodiment of the present invention.

Figure 2 is a flow chart of a clothing size recommendation method according to an embodiment of the present invention.

FIG3 is a schematic diagram of a size recommendation model according to an embodiment of the present invention.

Figure 4 is a flow chart of a clothing size recommendation method according to an embodiment of the present invention.

Figure 5 is a flow chart of a clothing size recommendation method according to an embodiment of the present invention.

Figure 6 is a flow chart of a clothing size recommendation method according to an embodiment of the present invention.

Figure 7 is a flow chart of a clothing size recommendation method according to an embodiment of the present invention.

Some embodiments of the present invention will be described in detail with reference to the accompanying drawings. The component symbols cited in the following description will be regarded as the same or similar components when the same component symbols appear in different drawings. These embodiments are only part of the present invention and do not disclose all possible implementations of the present invention. More precisely, these embodiments are only examples of methods and systems within the scope of the patent application of the present invention.

FIG1 is a schematic diagram of a clothing size recommendation system according to an embodiment of the present invention. Referring to FIG1 , the clothing size recommendation system 110 includes a storage device 111 and a processor 112. The clothing size recommendation system 110 may be composed of one or more computer devices, such as a computer or a server device, etc., and the present invention is not limited thereto. It should be noted that when the clothing size recommendation system 110 is composed of multiple computer devices, these computer devices may be connected via a wired or wireless communication interface.

In some embodiments, the clothing size recommendation system 110 is connected to the user device 120 via the network to receive data provided by the user device 120 or transmit data to the user device 120 via the network. The user device 120 may be, for example, an electronic device such as a laptop, a tablet computer, a mobile phone, etc., and the present invention is not limited thereto. In some embodiments, the user device 120 is an electronic device controlled by a user (also regarded as a consumer) for online shopping.

The storage device 111 can be used to store images, instructions, program codes, software modules and other data, and can be, for example, any type of fixed or removable random access memory (RAM), read-only memory (ROM), flash memory, hard disk or other similar devices, integrated circuits and combinations thereof.

The processor 112 is coupled to the storage device 111, such as a central processing unit (CPU), an application processor (AP), or other programmable general-purpose or special-purpose microprocessor, digital signal processor (DSP), image signal processor (ISP), graphics processing unit (GPU) or other similar devices, integrated circuits and combinations thereof. The processor 112 can access and execute the software module recorded in the storage device 111 to implement the clothing size recommendation method in the embodiment of the present invention. The above-mentioned software module can be broadly interpreted as meaning instructions, instruction sets, codes, program codes, programs, applications, software packages, threads, procedures, functions, etc.

FIG2 is a flow chart of a clothing size recommendation method according to an embodiment of the present invention, and the method flow of FIG2 can be implemented by the clothing size recommendation system 110 of FIG1. Please refer to FIG1 and FIG2 at the same time. The following is a description of the steps of the clothing size recommendation method of this embodiment in conjunction with the components of the clothing size recommendation system 110 in FIG1.

In step S210, the processor 112 obtains the basic body information of the user. The user can provide the basic body information to the processor 112 of the clothing size recommendation system 110 through the user device 120. In more detail, when the user conducts online shopping, the user device 120 can connect to the online shopping platform or online store via the network. The user can input the basic body information of the user through the application program interface displayed by the user device 120, so that the processor 112 of the clothing size recommendation system 110 can receive the basic body information of the user via the network.

In some embodiments, the basic human body information of the user may include height, weight, gender or age, etc. In addition, in some embodiments, the basic human body information of the user may also include body shape features, etc. The above body shape features are, for example, "inverted triangle", "pear shape", "straight tube shape" or "apple shape", etc., but the present invention is not limited thereto. For example, the application program interface displayed by the user device 120 may present graphical options of body shape features for the user to choose. Alternatively, in some embodiments, by the user taking a photo using the user device 120 or selecting an image stored in the user device 120, the user device 120 may transmit the user's full body image or half body image to the processor 112 of the clothing size recommendation system 110. The processor 112 can analyze the user's basic body information through image recognition technology or machine learning technology.

In step S220, the processor 112 determines the user's body circumference data based on the basic body information and the public body model data. In some embodiments, a database recording the public body model data is stored in the storage device 111, and the processor 112 can convert the basic body information provided by the user into the user's body circumference data based on the public body model data. The user's body circumference data may include chest circumference, shoulder width, waist circumference, hip circumference, leg circumference, arm circumference, arm length and leg length, etc. In more detail, the public body model data may include a plurality of preset body models, and these preset body models have corresponding body circumference data. The processor 112 can map the basic human body information provided by the user to a preset human body model in the public human body model data, and use the human body circumference data corresponding to the preset human body model as the human body circumference data of the user. In addition, in some embodiments, if the human body circumference information is known to the user, the user can also provide the human body circumference information to the clothing size recommendation system 110 through the user device 120 by direct input. In addition, in some embodiments, the user device 120 can transmit the user's full-body image or half-body image to the processor 112 of the clothing size recommendation system 110 by taking a photo with the user device 120 or selecting an image stored in the user device 120. The processor 112 can analyze the user's human body circumference information through image recognition technology or machine learning technology.

In step S230, the processor 112 determines a first model parameter of the size recommendation model according to the first clothing product, so as to apply the first model parameter corresponding to the first clothing product to the size recommendation model. The size recommendation model may be a machine learning model, and the machine learning algorithm applied thereto may be a linear regression, a logistic regression, a LASSO regression, a ridge regression, a decision tree, a neural network, or a support vector machine, etc., but the present invention is not limited thereto. Alternatively, the size recommendation model may be a statistical model, such as a statistical model established based on the central limit theorem, the law of large numbers, or the maximum likelihood method, but the present invention is not limited thereto. The model type in the embodiment of the present invention can be set according to the actual application. When the transaction data of the first clothing product has been collected enough, the processor 112 can perform machine learning or statistical analysis based on the transaction data of the first clothing product to obtain the first model parameter of the size recommendation model. When the transaction data of the first clothing product has not been collected, the processor 112 can first determine the second clothing product similar to the first clothing product, and then use the second model parameter of the second clothing product as the first model parameter of the first clothing product. The model parameters of the size recommendation model may be the regression coefficient of the regression model or the mean, expected value and variance of the statistical model, or the threshold, feature split point, weight, boundary, etc. in the machine learning model, but the present invention is not limited to this.

It should be particularly noted that in the embodiment of the present invention, the processor 112 can read the model parameters of the size recommendation model from the storage device 111, and each clothing product will correspond to a set of exclusive model parameters. More specifically, in response to receiving the instruction issued by the user to select the first clothing product, the processor 112 can determine a corresponding set of first model parameters based on the first clothing product. Correspondingly, in response to receiving the instruction issued by the user to select the second clothing product, the processor 112 can determine a corresponding set of second model parameters based on the second clothing product. The first model parameters of the first clothing product may be different from the second model parameters of the second clothing product.

In step S240, the processor 112 determines the recommended size of the first clothing product according to the user's body circumference data and the clothing characteristic data of the first clothing product by using a size recommendation model. The clothing characteristic data of the first clothing product may include clothing detail size data of various preset clothing sizes (such as XS, S, M, L, XL, etc.) of the first clothing product. The clothing detail size data may include the flat size of the first clothing product, such as sleeve length, sleeve width, clothing width, clothing length, skirt length, skirt width, trouser length, trouser leg width, etc., but the present invention is not limited thereto.

In detail, FIG. 3 is a schematic diagram of a size recommendation model according to an embodiment of the present invention. Referring to FIG. 3 , the processor 112 may input the user's body circumference data Body_info and the clothing characteristic data A_d of the first clothing product into the size recommendation model M1 applying the model parameter M_P (i.e., the first model parameter). However, in other embodiments, the processor 112 may first calculate the proportional relationship or multiplication result between the body circumference data and the clothing detail size data, and then input the aforementioned proportional relationship or multiplication result into the size recommendation model applying the first model parameter. Therefore, the size recommendation model M1 may output n matching information P_1~P_n corresponding to n clothing sizes of the first clothing product, respectively, where n is an integer greater than 1. For example, the size recommendation model M1 can output the matching information P_1 of the clothing size "S" of the first clothing product. The matching information P_1~P_n output by the size recommendation model M1 is, for example, classification category, classification probability, statistical probability, or a value generated based on the aforementioned data. Then, the processor 112 determines the recommended size of the first clothing product based on the n matching information P_1~P_n of the n clothing sizes of the first clothing product. However, in other embodiments, the size recommendation model can also directly output the recommended size of the first clothing product.

Next, in step S250, the processor 112 provides the user with the recommended size of the first clothing product. The processor 112 may transmit the recommended size of the first clothing product to the user device 120 for display, so that the user can refer to the recommended size of the first clothing product when purchasing the first clothing product online. Based on this, since the clothing size recommendation system 110 of the embodiment of the present invention can apply a dedicated model parameter to each clothing product, it can provide size recommendations that meet consumer needs for diversified clothing products of various styles.

In some embodiments, the processor 112 may also determine descriptive information of the recommended size and provide the descriptive information about the recommended size to the user. The processor 112 may transmit the descriptive information of the recommended size to the user device 120 for display, so that the user can refer to the descriptive information of the recommended size when purchasing the first clothing product online, thereby helping the user to further understand the recommended degree and fit of the recommended size. The descriptive information of the recommended size may be presented to the user through a numerical value, text, or icon.

Furthermore, in some embodiments, the descriptive information of the recommended size may be a recommendation score or a probability value, and the descriptive information may be generated according to the classification probability or statistical probability output by the size recommendation model. In some embodiments, the descriptive information of the recommended size may be a text description, and the text description may be generated by the processor 112 comparing the recommended size with a reference size. If the recommended size is larger than the reference size, the processor 112 may determine the text description to be the first text description "comfortable". If the recommended size is the same as the reference size, the processor 112 may determine the text description to be the second text description "standard". If the recommended size is smaller than the reference size, the processor 112 may determine the text description to be the third text description "fitting".

In some embodiments, the processor 112 may search for a standard size recommendation table based on the user's basic body information or body circumference information to determine a reference size for the user wearing the first clothing product. Afterwards, the processor 112 compares the recommended size generated by the size recommendation model with the reference size to provide descriptive information of the recommended size. The above-mentioned standard size recommendation table may be provided by the clothing manufacturer of the first clothing product and recorded in the storage device 111. Alternatively, in some embodiments, the processor 112 may perform statistical analysis on the purchase sizes of other consumers with similar body shapes who purchased the first clothing product to determine a reference size corresponding to the user.

FIG. 4 is a flow chart of a clothing size recommendation method according to an embodiment of the present invention, and the method flow of FIG. 4 can be implemented by the clothing size recommendation system 110 of FIG. 1 .

In step S402, the processor 112 obtains the basic human body information of the user. In step S404, the processor 112 determines a plurality of human body circumference data associated with the user based on the basic human body information and the public human body model data.

In step S406, the processor 112 determines whether the transaction data of the first apparel product meets the preset conditions. For example, the processor 112 can determine whether the number of transactions in the transaction data of the first apparel product is greater than the preset number (for example, 0 times). If the step S406 is judged as yes, it means that the transaction data of the first apparel product is sufficient and the first model parameter has been generated based on the transaction data of the first apparel product. In step S410, the processor 112 can obtain the first model parameter from the database. On the other hand, if the step S406 is judged as no, it means that the transaction data of the first apparel product is insufficient and the first model parameter has not yet been generated based on the transaction data of the first apparel product. Therefore, in step S408, when the transaction data of the first clothing product does not meet the preset condition, the processor 112 can select a second clothing product based on the clothing characteristic data of the first clothing product to use the second model parameter corresponding to the second clothing product as the first model parameter.

In more detail, in addition to the detailed size data of each preset clothing size, the clothing characteristic data of the first clothing product may also include other clothing description data, such as product name, text description, pattern, purpose, season, material, design style and pattern information, etc. The processor 112 may compare the clothing characteristic data of the first clothing product with the clothing characteristic data of other clothing products to select a second clothing product similar to the first clothing product, and use the second model parameter generated based on the transaction data of the second clothing product as the first model parameter. In some embodiments, the processor 112 may perform a similarity comparison between the clothing characteristic data of the first clothing product and the clothing characteristic data of multiple other clothing products to select the second clothing product from the other clothing products. In the process of the processor 112 marking the clothing characteristic data of the first clothing product, the processor 112 may use related methods such as one-hot encoding, bag-of-words model, latent semantic analysis or word embedding, etc.

For example, the processor 112 can use the Sequence Matcher algorithm to calculate the similarity between the clothing characteristic data of the first clothing product and the clothing characteristic data of other clothing products, thereby selecting a second clothing product similar to the first clothing product, and the selection method can be implemented by a machine learning model, a voting mechanism, etc. For example, assuming that the first clothing product is a long-sleeved suit shirt, and the second clothing product searched by the processor 112 from the database will be another long-sleeved suit shirt of a similar style. In this way, for clothing products that have just started selling, even if there is not enough transaction data, the clothing size recommendation system 110 of the embodiment of the present invention can still refer to the model parameters of other similar clothing products to determine the recommended size.

Next, in step S412, the processor 112 applies the first model parameter corresponding to the first clothing product to the size recommendation model. In step S414, the processor 112 determines the recommended size of the first clothing product according to the user's body circumference data and the clothing characteristic data of the first clothing product by using the size recommendation model. In step S416, the processor 112 provides the recommended size of the first clothing product to the user.

After the user refers to the recommended size of the first clothing item and conducts a transaction, in step S418, the processor 112 collects transaction data of the first clothing item. The transaction data of the first clothing item may include actual purchase size, customer feedback information, or return and exchange information. In step S420, the processor 112 updates the first model parameter of the size recommendation model according to the transaction data of the first clothing item. That is, after completing a transaction associated with the first clothing item, the transaction data of the transaction can be used to update the first model parameter of the size recommendation model. With the accumulation of transaction data of the first clothing item, the processor 112 can continuously update the first model parameter of the size recommendation model, so that the first model parameter of the size recommendation model can converge to a stable state.

In some embodiments, the processor 112 determines whether a recommended size for a transaction is successfully recommended based on the transaction data of the first clothing product. Then, the processor 112 updates the first model parameter of the size recommendation model based on whether the recommended size for the transaction is successfully recommended using the transaction data of the first clothing product.

Specifically, the processor 112 may compare the actual purchase size with the recommended size to determine whether the recommended size is successfully recommended. When the actual purchase size is different from the recommended size, the processor 112 may determine that the recommended size determined based on the size recommendation model is not successfully recommended (i.e., the recommendation fails). Alternatively, the processor 112 may perform keyword detection or natural language processing on the customer feedback information provided by the user, and determine whether the recommended size is successfully recommended based on the keywords or semantics in the customer feedback information. Alternatively, when the processor 112 determines that the user returns the first clothing product corresponding to the recommended size, the processor 112 may determine that the recommended size determined based on the size recommendation model is not successfully recommended.

In some embodiments, if the recommended size of a transaction is determined to be successfully recommended, the processor 112 may update the first model parameter of the size recommendation model according to the recommended size of the transaction and the user's body circumference data. If the recommended size of a transaction is determined to be unsuccessful, the processor 112 may not update the first model parameter of the size recommendation model according to the transaction data of the transaction.

FIG5 is a flow chart of a clothing size recommendation method according to an embodiment of the present invention, and the method flow of FIG5 can be implemented by the clothing size recommendation system 110 of FIG1.

In step S502, the processor 112 obtains basic human body information of the user. In step S504, the processor 112 determines multiple human body circumference data associated with the user based on the basic human body information and public human body model data. In step S506, the processor 112 determines the first model parameter of the size recommendation model based on the first clothing product, so as to apply the first model parameter corresponding to the first clothing product to the size recommendation model.

In step S508, the processor 112 uses a size recommendation model to obtain matching information for each of the multiple clothing sizes of the first clothing item based on the body circumference data and the clothing characteristic data of the first clothing item. The processor 112 can input the body circumference data and the clothing detailed size data of each clothing size into the size recommendation model and correspondingly obtain matching information for each of the multiple clothing sizes. It should be particularly noted that in order to ensure that the recommended sizes have ideal wearing comfort, the processor 112 can first filter out some obviously inappropriate clothing sizes. In step S510, the processor 112 can filter out some clothing sizes from the clothing sizes based on the body circumference data and the clothing detailed size data of each clothing size. In step S512, the processor 112 determines the recommended size of the first clothing item based on the matching information of the respective sizes of some clothing items.

For example, the processor 112 can use the size recommendation model to obtain the matching information of multiple clothing sizes "XS, S, M, L, XL" of the first clothing product. In addition, the processor 112 can compare the user's arm circumference with the sleeve width of the clothing sizes "XS, S, M, L, XL" to filter out some clothing sizes "M, L, XL". Then, the processor 112 can select and determine the recommended size of the first clothing product based on the matching information of some clothing sizes "M, L, XL".

In step S514, the processor 112 provides the recommended size of the first clothing product to the user. In step S516, the processor 112 collects transaction data of the first clothing product. In step S518, the processor 112 updates the first model parameter of the size recommendation model according to the transaction data of the first clothing product.

However, in other embodiments, the processor 112 may first compare the human body circumference data with the clothing detail size data of each clothing size to filter out some clothing sizes, and then input the clothing detail size data of some clothing sizes into the size recommendation model to obtain the matching information of some clothing sizes. For example, the processor 112 may compare the user's leg circumference with the trouser leg width of the clothing sizes "XS, S, M, L, XL" to filter out some clothing sizes "M, L, XL". Then, the processor 112 may use the size recommendation model to obtain the matching information of some clothing sizes "M, L, XL", and select the recommended size of the first clothing product accordingly.

FIG. 6 is a flow chart of a clothing size recommendation method according to an embodiment of the present invention, and the method flow of FIG. 6 can be implemented by the clothing size recommendation system 110 of FIG. 1 .

In step S602, the processor 112 obtains basic human body information of the user. In step S604, the processor 112 determines multiple body circumference data associated with the user based on the basic human body information and the public human body model data. In step S606, the processor 112 determines the first model parameter of the size recommendation model based on the first clothing product to apply the first model parameter corresponding to the first clothing product to the size recommendation model. In step S608, the processor 112 uses the size recommendation model to obtain matching information of multiple clothing sizes of the first clothing product based on the body circumference data and the clothing characteristic data of the first clothing product.

In some embodiments, the processor 112 determines whether the matching information corresponding to each clothing size of the first clothing product meets the matching condition to determine the recommended size of the first clothing product. In some embodiments, the processor 112 can determine whether the matching information meets the matching condition by determining whether the matching information is greater than a threshold value. In some embodiments, the processor 112 can sort the n matching information of n clothing sizes to determine whether the matching information meets the matching condition according to the sorting ranking corresponding to each clothing size.

In some embodiments, the matching information of each clothing size may include a weighted calculation value of multiple matching probability values, and the processor 112 may perform a weighted calculation operation on the multiple matching probability values associated with each clothing size to obtain a weighted calculation value for each clothing size. The above multiple matching probability values may be generated based on different clothing detail size data of a certain clothing size and the corresponding human circumference data. For example, the processor 112 may generate a first matching probability value using a first size recommendation model based on the width of the clothing size "M" and the chest circumference of the user, and generate a second matching probability value using a second size recommendation model based on the sleeve width of the clothing size "M" and the arm circumference of the user. Then, the processor 112 performs a weighted calculation on the first matching probability value and the second matching probability value. Afterwards, the processor 112 may perform a sorting operation or a threshold value comparison operation on the weighted calculated values of each clothing size to determine whether the matching information meets the matching condition according to the sorting ranking or threshold value comparison result corresponding to each clothing size.

In step S610, when the matching information of at least two of the clothing sizes meets the matching conditions, the processor 112 selects a recommended size from at least two of the clothing sizes according to the clothing style indicator of the first clothing item. The clothing style indicator of the first clothing item can be set to indicate at least one clothing detail size. Different clothing items can correspond to different clothing style indicators. Specifically, assuming that the first clothing item is a "formal suit", in order to achieve a formal visual effect, the clothing style indicator of the first clothing item can be set to indicate "length of clothing", because suits that are too short or too long can easily have an informal feeling. Therefore, since the clothing style indicator of the first clothing item is set to indicate length of clothing, the processor 112 can determine the recommended size based on the length of at least two of the clothing sizes that meet the matching conditions. For example, the processor 112 may determine the final recommended size based on the user's height or upper body length and the length of at least two of the clothing sizes. On the other hand, in step S612, when the matching information of only one of the clothing sizes meets the matching conditions, the processor 112 uses only one of the clothing sizes as the recommended size.

In step S614, the processor 112 provides the recommended size of the first clothing product to the user. In step S616, the processor 112 collects transaction data of the first clothing product. In step S618, the processor 112 updates the first model parameter of the size recommendation model according to the transaction data of the first clothing product.

FIG. 7 is a flow chart of a clothing size recommendation method according to an embodiment of the present invention, and the method flow of FIG. 7 can be implemented by the clothing size recommendation system 110 of FIG. 1 .

In step S702, the processor 112 obtains basic human body information of the user. In step S704, the processor 112 determines multiple human body circumference data associated with the user based on the basic human body information and the public human body model data. In step S706, the processor 112 determines the first model parameter of the size recommendation model based on the first clothing product to apply the first model parameter corresponding to the first clothing product to the size recommendation model. In step S708, the processor 112 determines the recommended size of the first clothing product based on the user's human body circumference data and the clothing characteristic data of the first clothing product by using the size recommendation model. In step S710, the processor 112 provides the recommended size of the first clothing product to the user.

In step S712, the processor 112 collects the transaction records of the user and counts the number of failed size recommendations for the user. In detail, each time a user conducts a transaction to purchase a certain clothing item, the processor 112 can determine whether the recommended size of the transaction is a failed recommendation based on the transaction data. How to determine whether a recommended size is a failed recommendation or a failed recommendation has been described in the aforementioned embodiment and will not be repeated here. Each time the processor 112 determines that the recommended size of a transaction of a certain user is a failed recommendation, the processor 112 can accumulate the number of failed size recommendations for the user. When the number of size recommendation failures for a certain user is greater than a preset value (e.g., 5 times), it means that the body shape of the user is significantly different from that of the general public, and therefore the user's body circumference data cannot be accurately estimated through the general public body model data. Therefore, in step S714, in response to the user's size recommendation failure number being greater than the preset value, the processor 112 uses the user's transaction record to update the user's body circumference data. In other words, when the number of size recommendation failures for the user is greater than the preset value, the processor 112 derives and updates the user's body circumference data based on the user's transaction record. After updating the user's body circumference data, the processor 112 uses a size recommendation model to determine the corresponding recommended size based on the updated body circumference data and the clothing characteristic data of a certain clothing product.

In some embodiments, the user's transaction record includes actual purchase sizes or return and exchange information for multiple transactions. For example, assuming that the actual purchase sizes of multiple transactions are larger than the recommended sizes, the processor 112 can adjust and update the user's body circumference data based on the difference between the actual purchase sizes and the recommended sizes.

In summary, in the embodiment of the present invention, human body circumference data can be generated based on basic human body information provided by the user, and the model parameters used by the size recommendation model can vary according to different clothing products. Therefore, for clothing products with diverse styles, the recommended size suitable for the user can be determined through the size recommendation model based on detailed human body circumference data and exclusive model parameters, thereby optimizing the online shopping experience of clothing products. In addition, when the transaction data of clothing products is insufficient, the model parameters of other similar clothing products can be used to make size recommendations to minimize the obvious error between the recommended size and the user's expectations. Furthermore, in the embodiment of the present invention, when the recommended size does not meet the user's expectations for many times, the user's body circumference data can be updated according to the transaction data, so that the size recommendation model can be applicable to consumers of various body shapes.

Although the present invention has been disclosed as above by the embodiments, it is not intended to limit the present invention. Anyone with ordinary knowledge in the relevant technical field can make some changes and modifications without departing from the spirit and scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the scope defined by the attached patent application.

S210~S250: Steps

Claims (18)

A clothing size recommendation method, suitable for being executed by a processor, comprises: obtaining basic human body information of a user; determining the user's body circumference data according to the basic human body information and public body model data; determining a first model parameter of a size recommendation model according to a first clothing product, so as to apply the first model parameter corresponding to the first clothing product to the size recommendation model, wherein the size recommendation model comprises a machine learning model or a statistical model; determining the recommended size of the first clothing product according to the user's body circumference data and clothing characteristic data of the first clothing product by using the size recommendation model; and providing the recommended size of the first clothing product. The method further comprises: collecting transaction data of the first clothing item, wherein the transaction data of the first clothing item is used to determine whether the recommended size is a failed recommendation or a successful recommendation; and updating the first model parameter of the size recommendation model according to the transaction data of the first clothing item, wherein the method further comprises: responding to receiving an instruction to select the first clothing item, determining the corresponding first model parameter according to the first clothing item; and responding to receiving an instruction to select a second clothing item, determining the corresponding second model parameter according to the second clothing item, wherein the first model parameter of the first clothing item is different from the second model parameter of the second clothing item. The clothing size recommendation method as described in claim 1, wherein the step of determining the first model parameter of the size recommendation model according to the first clothing item so as to apply the first model parameter corresponding to the first clothing item to the size recommendation model comprises: when the transaction data of the first clothing item does not meet a preset condition, selecting the second clothing item based on the clothing characteristic data of the first clothing item so as to use the second model parameter corresponding to the second clothing item as the first model parameter. The clothing size recommendation method as described in claim 2, wherein the step of selecting the second clothing item based on the clothing characteristic data of the first clothing item includes: performing a similarity comparison between the clothing characteristic data of the first clothing item and the clothing characteristic data of multiple other clothing items to select the second clothing item from the other clothing items. The clothing size recommendation method as described in claim 1, wherein the step of updating the first model parameter of the size recommendation model according to the transaction data of the first clothing product includes: judging whether the recommended size of a transaction is successfully recommended according to the transaction data of the first clothing product; and updating the first model parameter of the size recommendation model according to whether the recommended size of the transaction is successfully recommended using the transaction data of the first clothing product. The clothing size recommendation method as described in claim 1, wherein the step of determining the recommended size of the first clothing item based on the body circumference data of the user and the clothing characteristic data of the first clothing item by using the size recommendation model comprises: obtaining matching information of each of the multiple clothing sizes of the first clothing item using the size recommendation model based on the body circumference data and the clothing characteristic data of the first clothing item; and determining the recommended size of the first clothing item based on the matching information of each of the clothing sizes. The clothing size recommendation method as described in claim 5, wherein the step of determining the recommended size of the first clothing item according to the matching information of each clothing size includes: filtering out some of the clothing sizes from the clothing sizes according to the human body circumference data and the clothing detail size data of each clothing size; and determining the recommended size of the first clothing item according to the matching information of each of the some clothing sizes. As described in claim 5, the step of determining the recommended size of the first clothing item based on the matching information of each clothing size includes: when the matching information of at least two of the clothing sizes meets the matching condition, selecting the recommended size from at least two of the clothing sizes based on the clothing style index of the first clothing item. The clothing size recommendation method as described in claim 1 further comprises: collecting the transaction records of the user and counting the number of failed size recommendations of the user; and in response to the number of failed size recommendations of the user being greater than a preset value, using the transaction records of the user to update the body circumference data of the user. The clothing size recommendation method as described in claim 1 further includes: determining the descriptive information of the recommended size; and providing the descriptive information of the recommended size to the user. A clothing size recommendation system includes: a storage device; and a processor coupled to the storage device and configured to: obtain basic human body information of a user; determine the user's body circumference data based on the basic human body information and public body model data; determine a first model parameter of a size recommendation model based on a first clothing product, so as to apply the first model parameter corresponding to the first clothing product to the size recommendation model, wherein the size recommendation model includes a machine learning model or a statistical model; determine the recommended size of the first clothing product based on the user's body circumference data and the clothing characteristic data of the first clothing product by using the size recommendation model; and provide the first clothing product The processor is further configured to: collect transaction data of the first clothing item, wherein the transaction data of the first clothing item is used to determine whether the recommended size is a failed recommendation or a successful recommendation; and update the first model parameter of the size recommendation model according to the transaction data of the first clothing item, wherein the processor is further configured to: respond to receiving an instruction to select the first clothing item, determine the corresponding first model parameter according to the first clothing item; and respond to receiving an instruction to select a second clothing item, determine the corresponding second model parameter according to the second clothing item, wherein the first model parameter of the first clothing item is different from the second model parameter of the second clothing item. The clothing size recommendation system as described in claim 10, wherein the processor is further configured to: when the transaction data of the first clothing product does not meet the preset condition, select the second clothing product based on the clothing characteristic data of the first clothing product, and use the second model parameter corresponding to the second clothing product as the first model parameter. The clothing size recommendation system as described in claim 11, wherein the processor is further configured to: Compare the clothing characteristic data of the first clothing item with the clothing characteristic data of multiple other clothing items for similarity, so as to select the second clothing item from the other clothing items. The clothing size recommendation system as described in claim 10, wherein the processor is further configured to: determine whether the recommended size of a transaction is successfully recommended based on the transaction data of the first clothing product; and update the first model parameter of the size recommendation model using the transaction data of the first clothing product based on whether the recommended size of the transaction is successfully recommended. The clothing size recommendation system as described in claim 10, wherein the processor is further configured to: obtain the matching information of each of the multiple clothing sizes of the first clothing product using the size recommendation model according to the human body circumference data and the clothing characteristic data of the first clothing product; and determine the recommended size of the first clothing product according to the matching information of each clothing size. The clothing size recommendation system as described in claim 14, wherein the processor is further configured to: filter out some of the clothing sizes from the clothing sizes based on the human body circumference data and the clothing detail size data of each of the clothing sizes; and determine the recommended size of the first clothing item based on the matching information of each of the some of the clothing sizes. The clothing size recommendation system as described in claim 14, wherein the processor is further configured to: when the matching information of at least two of the clothing sizes meets the matching condition, select the recommended size from at least two of the clothing sizes according to the clothing style index of the first clothing item. The clothing size recommendation system as described in claim 10, wherein the processor is further configured to: collect the transaction records of the user and count the number of failed size recommendations of the user; and in response to the number of failed size recommendations of the user being greater than a preset value, update the body circumference data of the user using the transaction records of the user. A clothing size recommendation system as described in claim 10, wherein the processor is further configured to: determine descriptive information of the recommended size; and provide the descriptive information of the recommended size to the user.

Images