Method and device for determining similar stores

Abstract

The present invention discloses a method and apparatus for determining similar stores, relating to the field of artificial intelligence technology. The method comprises: generating a sequence of store identifiers to be processed based on user behavioral data on an e-commerce platform; performing word embedding processing on the sequence of store identifiers using a word embedding model to obtain embedding vectors for the stores; calculating the similarity between stores based on the embedding vectors, and then determining similar stores based on the similarity between the stores. These steps can accurately quantify the similarity between stores, reduce the difficulty of calculating similar stores, and improve the accuracy of the similarity calculation results.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a method and device for determining similar stores. Background Art

[0002] With the continuous development of e-commerce, more and more companies and individuals are choosing to enter e-commerce platforms, also known as "opening online stores." Identifying similar stores to a target store is crucial in practical business scenarios such as product recommendations. Existing technologies often rely on platforms to compare the categories, labels, locations, and descriptions of the products sold by the stores to find similar stores. Other methods extract keywords from store names and search keyword libraries to find similar stores.

[0003] In the process of realizing the present invention, the inventors of the present invention found that the existing methods for determining similar stores have the following problems: First, most of the existing technologies are qualitative descriptions of whether stores are similar, which makes it difficult to quantify and compare the similarity between stores more accurately; Second, the existing methods for finding similar stores are limited by factors such as store labels, keyword settings, and extraction, which makes it difficult to find similar stores and has low calculation accuracy. For example, whether the store label setting represents the characteristics of the store, whether it can distinguish different stores and has good representativeness for the same type of stores, will greatly affect the accuracy of the calculation results of similar stores. For example, how to extract keywords from store names, and how to deal with the situation where the store names of some stores contain little or no clear information, these factors increase the difficulty of calculating similar stores. Summary of the Invention

[0004] In view of this, the present invention provides a method and device for determining similar stores, which can accurately quantify the similarity between stores, reduce the difficulty of calculating similar stores, and improve the accuracy of the calculation results of similar stores.

[0005] To achieve the above object, according to a first aspect of the present invention, a method for determining similar stores is provided.

[0006] The method for determining similar stores of the present invention includes: generating a store identification sequence to be processed based on user behavior data on an e-commerce platform; performing word embedding processing on the store identification sequence to be processed through a word embedding model to obtain an embedding vector of the store; calculating the similarity between stores based on the embedding vector of the store, and then determining similar stores based on the similarity between the stores.

[0007] Optionally, the user behavior data on the e-commerce platform includes: browsing behavior records of multiple users on the e-commerce platform, and / or purchasing behavior records of multiple users on the e-commerce platform, and / or add-to-cart behavior records of multiple users on the e-commerce platform.

[0008] Optionally, generating a store identification sequence to be processed based on user behavior data on the e-commerce platform includes: segmenting the browsing behavior records of the multiple users on the e-commerce platform according to a set time window, and taking the sequence of store identification sequences obtained by segmentation in which the number of stores is greater than or equal to a preset threshold as the store identification sequence to be processed.

[0009] Optionally, the method further includes: before segmenting the browsing behavior records of the multiple users on the e-commerce platform according to the set time window, removing the browsing behavior records of abnormal users from the browsing behavior records of the multiple users on the e-commerce platform.

[0010] Optionally, the word embedding processing of the store identification sequence to be processed by the word embedding model to obtain the store's embedding vector includes: saving the store identification sequence to be processed as a two-dimensional array; loading a pre-trained word embedding model, and inputting the two-dimensional array into the pre-trained word embedding model to obtain the store's embedding vector.

[0011] Optionally, the word embedding model is a Word2vec model.

[0012] Optionally, calculating the similarity between stores based on the embedding vectors of the stores includes: combining the embedding vectors of the stores in pairs to obtain embedding vector pairs of stores, and calculating the cosine similarity of the embedding vector pairs of stores using a cosine similarity calculation formula.

[0013] To achieve the above object, according to a second aspect of the present invention, a device for determining similar stores is provided.

[0014] The device for determining similar stores of the present invention includes: a generation module for generating a store identification sequence to be processed based on user behavior data on an e-commerce platform; a word embedding processing module for performing word embedding processing on the store identification sequence to be processed through a word embedding model to obtain an embedding vector of the store; and a determination module for calculating the similarity between stores based on the embedding vectors of the stores, and then determining similar stores based on the similarity between the stores.

[0015] To achieve the above objective, according to a third aspect of the present invention, an electronic device is provided.

[0016] The electronic device of the present invention includes: one or more processors; and a storage device for storing one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors implement the method of determining similar stores of the present invention.

[0017] To achieve the above objective, according to a fourth aspect of the present invention, a computer-readable medium is provided.

[0018] The computer-readable medium of the present invention stores a computer program, which, when executed by a processor, implements the method for determining similar stores of the present invention.

[0019] One embodiment of the above invention has the following advantages or beneficial effects: by generating a store identification sequence to be processed based on the user's behavioral data on the e-commerce platform, performing word embedding processing on the store identification sequence to be processed through a word embedding model to obtain the store's embedding vector, calculating the similarity between stores based on the store's embedding vector, and then determining similar stores based on the similarity between the stores, the similarity between stores can be accurately quantified, the difficulty of calculating similar stores can be reduced, and the accuracy of the calculation results of similar stores can be improved.

[0020] The further effects of the above-mentioned non-conventional optional manner will be described below in conjunction with specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The accompanying drawings are provided for a better understanding of the present invention and are not intended to limit the present invention.

[0022] Figure 1 is an exemplary system architecture diagram in which embodiments of the present invention may be applied;

[0023] Figure 2 This is a schematic diagram of the main flow of the method for determining similar stores according to the first embodiment of the present invention;

[0024] Figure 3 is a schematic diagram of the main flow of a method for determining similar stores according to a second embodiment of the present invention;

[0025] Figure 4 It is a schematic diagram of generating a store ID sequence based on user browsing behavior data;

[0026] Figure 5 is a schematic diagram of main modules of an apparatus for determining similar stores according to a third embodiment of the present invention;

[0027] Figure 6 It is a schematic diagram of the structure of a computer system suitable for implementing an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0028] The following description of exemplary embodiments of the present invention is made in conjunction with the accompanying drawings, in which various details of the embodiments of the present invention are included to facilitate understanding. These details should be considered as merely exemplary. Therefore, it should be appreciated by those skilled in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present invention. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0029] It should be pointed out that the various embodiments of the present invention and the technical features in the embodiments can be combined with each other without affecting the implementation of the present invention.

[0030] Figure 1 An exemplary system architecture 100 is shown to which the method for determining similar stores or the apparatus for determining similar stores according to an embodiment of the present invention can be applied.

[0031] like Figure 1 As shown, system architecture 100 may include terminal devices 101, 102, 103, a network 104, and a server 105. Network 104 is a medium for providing communication links between terminal devices 101, 102, 103 and server 105. Network 104 may include various connection types, such as wired or wireless communication links or fiber optic cables.

[0032] Users can use terminal devices 101, 102, and 103 to interact with server 105 via network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, and 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0033] The terminal devices 101 , 102 , and 103 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, and desktop computers.

[0034] The server 105 may be a server that provides various services, such as a background management server that supports shopping applications browsed by users using the terminal devices 101, 102, and 103. For example, the background management server may process data processing requests sent by the terminal devices over the network and feed back the processing results to the terminal devices.

[0035] It should be noted that the method for determining similar stores provided in the embodiment of the present invention is generally executed by the server 105 . Accordingly, the device for determining similar stores is generally set in the server 105 .

[0036] It should be understood that Figure 1The number of terminal devices, networks and servers in the embodiment is merely illustrative. Any number of terminal devices, networks and servers may be provided as required.

[0037] Figure 2 FIG. 1 is a schematic diagram of the main flow of the method for determining similar stores according to the first embodiment of the present invention. Figure 2 As shown, the method for determining similar stores in an embodiment of the present invention includes:

[0038] Step S201: Generate a store identification sequence to be processed based on the user's behavior data on the e-commerce platform.

[0039] In one optional example, the user behavior data on the e-commerce platform includes browsing behavior records of multiple users on the e-commerce platform. For example, all user browsing behavior logs on the e-commerce platform for N days prior to a selected date may be obtained, where the value of N is related to the periodic update time of the store similarity, such as one week, one month, or one quarter.

[0040] In another optional example, the user behavior data on the e-commerce platform includes: purchase behavior records of multiple users on the e-commerce platform. For example, all user purchase behavior logs on the e-commerce platform N days before the selected date can be obtained.

[0041] In another optional example, the user behavior data on the e-commerce platform includes: records of multiple users' add-to-cart (i.e., adding to cart) behavior on the e-commerce platform. For example, a log of all users' add-to-cart behavior on the e-commerce platform N days prior to a selected date may be obtained. Furthermore, the user behavior data on the e-commerce platform may also include any two or a combination of three of multiple users' browsing behavior records on the e-commerce platform, multiple users' purchasing behavior records on the e-commerce platform, and multiple users' add-to-cart behavior records on the e-commerce platform.

[0042] In this step, the store identification sequence can be obtained by cleaning the user's behavior data on the e-commerce platform.

[0043] Step S202: performing word embedding processing on the store identification sequence to be processed through a word embedding model to obtain an embedding vector of the store.

[0044] In an optional example, the word embedding model is a Word2vec model. The Word2vec model was first used to process NLP (Natural Language Processing) problems. It is a process of mapping sparse word vectors in one-hot form into an n-dimensional dense vector using a one-layer neural network. It is an unsupervised learning method. The Word2vec model is generally divided into two models: Skip-Gram and CBOW (Continuous Bag-of-Words). The Skip-Gram model inputs the word vector of a specific word and outputs the context word vector corresponding to the word vector of this specific word. The CBOW model is just the opposite. It inputs the word vector corresponding to the context of a specific word and outputs the word vector of this specific word. Both the Skip-Gram and CBOW models contain three layers, namely the input layer, the projection layer and the output layer.

[0045] Furthermore, in the above optional example, the Word2vec model can be pre-trained, and then the store identification sequence to be processed is input into the trained Word2vec model to obtain the store's embedding vector. In addition, the word embedding model can also use other neural network models without affecting the implementation of the present invention.

[0046] Step S203: Calculate the similarity between stores based on the embedding vectors of the stores, and then determine similar stores based on the similarity between the stores.

[0047] In this step, the similarity between stores can be measured by calculating the similarity between the embedding vector of one store and the embedding vectors of other stores. For example, when calculating the similarity of the embedding vectors of stores, the following similarity calculation methods can be used: Euclidean distance, cosine distance, Pearson coefficient, Hamming distance, or Manhattan distance.

[0048] In an optional example, after determining the similarity between two stores, the similarity between the two stores may be compared with a preset threshold. If the similarity is greater than or equal to the preset threshold, the two stores are considered similar stores; otherwise, the two stores are considered dissimilar.

[0049] In another optional example, after determining the similarity between the target store and other stores, the other stores may be sorted in descending order of similarity, and the top M stores may be used as similar stores to the target store. The value of M may be 1, 2, 3, or other values.

[0050] In an embodiment of the present invention, a store identification sequence to be processed is generated based on the user's behavioral data on the e-commerce platform, the store identification sequence to be processed is word embedded through a word embedding model to obtain an embedding vector of the store, the similarity between stores is calculated based on the embedding vector of the store, and then similar stores are determined based on the similarity between the stores. This method is different from the existing method of qualitatively determining similar stores. It determines the embedding vector of the store based on artificial intelligence technology, and calculates the similarity of the store based on the embedding vector of the store, thereby achieving accurate quantification of the similarity between stores, which not only improves the processing efficiency of finding similar stores, but also improves the accuracy of finding similar stores. Moreover, compared with the existing method of calculating similar stores based on tags or keywords, this method obtains a store identification sequence by processing the user's behavioral data on the e-commerce platform, and subsequently obtains the embedding vector of the store based on the store identification sequence, thereby reducing the difficulty of calculating similar stores and improving the accuracy of the similar store calculation results.

[0051] Figure 3 FIG. 1 is a schematic diagram of the main flow of the method for determining similar stores according to the second embodiment of the present invention. Figure 3 As shown, the method for determining similar stores in an embodiment of the present invention includes:

[0052] Step S301: Obtain browsing behavior records of multiple users on the e-commerce platform.

[0053] For example, in this step, the full set of user browsing behavior logs on the e-commerce platform N days before the selected date can be obtained, where the value of N is related to the time when the store similarity is updated periodically, for example, N is one week, one month, or one quarter.

[0054] Step S302: segmenting the browsing behavior records of the multiple users on the e-commerce platform according to the set time window, and taking the store identification sequences obtained by segmentation in which the number of stores is greater than or equal to a preset threshold as the store identification sequences to be processed.

[0055] In an embodiment of the present invention, it is considered that a period of closely continuous browsing behavior of a user is related, while two browsing behaviors separated by a long time have no correlation or very weak correlation. In order to improve the accuracy of subsequent calculation of similar stores, the user's browsing behavior records on the e-commerce platform can be segmented according to the set time window to obtain a store identification sequence. The value of the time window can be set to 1 hour, 2 hours, or other reasonable values. In addition, considering that the small number of stores included in each time window will affect the accuracy of subsequent calculation of similar stores, and even make it impossible to calculate the relationship between stores, after the store identification sequence is obtained by segmentation, a sequence with a number of stores greater than or equal to a preset threshold is screened out as the store identification sequence to be processed.

[0056] In an optional example, the method of an embodiment of the present invention further includes: before step S302, removing the browsing behavior records of abnormal users from the browsing behavior records of multiple users on the e-commerce platform. Specifically, in this optional example, users whose browsing times within a period of time are greater than a preset threshold can be regarded as abnormal users or overly active users, such as users whose browsing times within a month are greater than 1,000 times as overly active users. By removing the store browsing behavior records of overly active users from the store browsing behavior records of multiple users, the adverse effects of the abnormal browsing behavior of these overly active users on the subsequent calculation of the similarity between stores can be reduced, thereby improving the reliability of the calculated similarity between stores.

[0057] Step S303: Save the store identification sequence to be processed as a two-dimensional array.

[0058] After obtaining the store ID sequence to be processed, it can be saved as a two-dimensional array for subsequent processing using a word embedding model. Each row of the two-dimensional array represents the ID sequence of stores that a user browsed sequentially within a time window.

[0059] Step S304: Loading a pre-trained word embedding model, inputting the two-dimensional array into the pre-trained word embedding model to obtain an embedding vector for the store.

[0060] In an embodiment of the present invention, the word embedding model specifically adopts the Word2vec model. The Word2vec model was first used to process NLP (Natural Language Processing) problems. It is a process of mapping sparse word vectors in one-hot form into an n-dimensional dense vector using a one-layer neural network. It is an unsupervised learning method. The Word2vec model is generally divided into two models: Skip-Gram and CBOW (Continuous Bag-of-Words). The Skip-Gram model inputs the word vector of a specific word and outputs the context word vector corresponding to the word vector of this specific word. The CBOW model is just the opposite. It inputs the word vector corresponding to the context of a specific word and outputs the word vector of this specific word. Both the Skip-Gram and CBOW models contain three layers, namely the input layer, the projection layer and the output layer.

[0061] In this step, the two-dimensional array obtained in step S303 is input into the pre-trained word embedding model to obtain the embedding vector of the store.

[0062] Furthermore, the method of the embodiment of the present invention may further include: training the Word2vec model to obtain a trained word embedding model. Specifically, if a fully trained Word2vec model is already available, the previously trained model can be used as a pre-trained model when the model is next updated. Based on this model, new store logo sequence data can be used to train the model, which can, to a certain extent, mitigate the impact of data fluctuations on the model.

[0063] In specific implementation, parameters need to be set when training the Word2vec model. In an optional example, the main parameter settings when training the Word2vec model are as follows: Set the high-frequency word random sampling threshold to 1e -4 , set the model algorithm to Skip-Gram, set the trained vector dimension to 32, set the maximum distance between the current word and the predicted word in a sequence during training to 5, set min_count (minimum word frequency threshold, words with a frequency less than min_count will be discarded) to 5, set workers (control the number of parallel training) to 4, set the number of iterations to 5, and set the number of negatively sampled noise words to 10. In this embodiment of the present invention, by setting reasonable parameters, the obtained embedding vector can better represent the corresponding store, thereby improving the accuracy of the subsequent calculation of similar stores.

[0064] Step S305: Calculate the similarity between stores based on the embedding vectors of the stores.

[0065] In this step, the store embedding vectors are paired to obtain pairs of store embedding vectors. The cosine similarity of these pairs of store embedding vectors is calculated using the cosine similarity calculation formula. The cosine similarity value ranges from -1 to 1. A larger cosine similarity value indicates that the embedding vectors of the two stores are more similar, while a smaller cosine similarity value indicates that the embedding vectors of the two stores are less similar.

[0066] Specifically, the store's embedding vector can be substituted into the following cosine similarity calculation formula:

[0067]

[0068] Among them, A and B represent the embedding vectors of any two different stores, A, B∈X and A≠B, X represents the set of all store embedding vectors; n represents the dimension of the store embedding vector; A i and B i Represents the value of the i-th dimension of the corresponding store embedding vector.

[0069] Step S306: Determine similar stores based on the similarity between the stores.

[0070] Exemplarily, in this step, after determining the similarity between any two stores, the similarity between the two stores can be compared with a preset threshold. If the similarity is greater than or equal to the preset threshold, the two stores are considered to be similar stores; otherwise, the two stores are considered to be dissimilar.

[0071] In an embodiment of the present invention, the above steps achieve precise quantification of the similarity between stores. This method differs from existing methods for qualitatively determining similar stores. It uses artificial intelligence technology to determine the store's embedding vector and calculates the store's similarity based on the store's embedding vector, achieving precise quantification of the similarity between stores. This not only improves the efficiency of finding similar stores, but also improves the accuracy of finding similar stores. Moreover, compared with existing methods for calculating similar stores based on tags or keywords, this method processes user behavior data on the e-commerce platform to obtain a store identification sequence, and then processes the store's embedding vector based on the store identification sequence, reducing the difficulty of calculating similar stores and improving the accuracy of the similar store calculation results. Furthermore, the calculation of store similarity can be used as basic calculation data for other business processes, including but not limited to product recommendation, search association, and assisting store owners in specifying business strategies. For example, in a product recommendation scenario, after a user enters the e-commerce platform, the method of the embodiment of the present invention can first determine similar stores to the target store that the user has recently browsed, and then actively recommend products from these similar stores to the user. Alternatively, when a user searches for related stores, the method of the embodiment of the present invention can determine similar stores, and then return the search results of similar stores to the user for reference and selection.

[0072] Figure 4 This is a schematic diagram of generating a store ID sequence based on user browsing behavior data. Figure 4 The process of generating a store identification sequence based on user browsing behavior data is described as follows. Figure 4 As shown, assuming that there are browsing behavior data of three users, user 1, user 2, and user 3, A to F represent different stores respectively. The horizontal arrows in the figure represent the time axis, and the dotted lines represent the division of the time window. If the threshold of the number of stores included in each time window is set to 2, then four store identification sequences are obtained through the division in step S302, namely: A BC, EAD, FB, and CBAE.

[0073] Figure 5 FIG. 1 is a schematic diagram of the main modules of the apparatus for determining similar stores according to the third embodiment of the present invention. Figure 5 As shown, the apparatus 500 for determining similar stores according to an embodiment of the present invention includes: a generation module 501 , a word embedding processing module 502 , and a determination module 503 .

[0074] The generation module 501 is used to generate a store identification sequence to be processed based on the user's behavior data on the e-commerce platform.

[0075] In one optional example, the user behavior data on the e-commerce platform includes browsing behavior records of multiple users on the e-commerce platform. For example, all user browsing behavior logs on the e-commerce platform for N days prior to a selected date may be obtained, where the value of N is related to the periodic update time of the store similarity, such as one week, one month, or one quarter.

[0076] In another optional example, the user behavior data on the e-commerce platform includes: purchase behavior records of multiple users on the e-commerce platform. For example, all user purchase behavior logs on the e-commerce platform N days before the selected date can be obtained.

[0077] In another optional example, the user behavior data on the e-commerce platform includes: records of multiple users' add-to-cart (i.e., adding to cart) behavior on the e-commerce platform. For example, a log of all users' add-to-cart behavior on the e-commerce platform N days prior to a selected date may be obtained. Furthermore, the user behavior data on the e-commerce platform may also include any two or a combination of three of multiple users' browsing behavior records on the e-commerce platform, multiple users' purchasing behavior records on the e-commerce platform, and multiple users' add-to-cart behavior records on the e-commerce platform.

[0078] The word embedding processing module 502 is used to perform word embedding processing on the store identification sequence to be processed through a word embedding model to obtain an embedding vector of the store.

[0079] In an optional example, the word embedding model in the word embedding processing module 502 is a Word2vec model. The Word2vec model was first used to process NLP (Natural Language Processing) problems. It is a process of mapping sparse word vectors in the form of one-hot to an n-dimensional dense vector using a one-layer neural network. It is an unsupervised learning method. The Word2vec model is generally divided into two models: Skip-Gram and CBOW (Continuous Bag-of-Words). The Skip-Gram model inputs the word vector of a specific word and outputs the context word vector corresponding to the word vector of this specific word. The CBOW model is just the opposite. It inputs the word vector corresponding to the context of a specific word and outputs the word vector of this specific word. Both the Skip-Gram and CBOW models contain three layers, namely, the input layer, the projection layer, and the output layer.

[0080] Furthermore, in the above optional example, the Word2vec model can be pre-trained, and then the store identification sequence to be processed is input into the trained Word2vec model to obtain the store's embedding vector. In addition, the word embedding model can also use other neural network models without affecting the implementation of the present invention.

[0081] The determination module 503 is configured to calculate the similarity between stores based on the embedding vectors of the stores, and then determine similar stores based on the similarity between the stores.

[0082] Specifically, the determination module 503 measures the similarity between stores by calculating the similarity between the embedding vector of one store and the embedding vectors of other stores. For example, when calculating the similarity between the embedding vectors of stores, the determination module 503 may use the following similarity calculation methods: Euclidean distance, cosine similarity, Pearson coefficient, Hamming distance, or Manhattan distance.

[0083] In an optional example, after determining the similarity between two stores, the determination module 503 may compare the similarity between the two stores with a preset threshold. If the similarity is greater than or equal to the preset threshold, the two stores are considered to be similar stores; otherwise, the two stores are considered to be dissimilar.

[0084] In another optional example, after determining the similarity between the target store and other stores, the determination module 503 may sort the other stores in descending order of similarity and select the top M stores as similar stores to the target store. The value of M may be 1, 2, 3, or other values.

[0085] Unlike the existing solutions for qualitatively determining similar stores, the embodiments of the present invention determine the embedding vectors of stores based on artificial intelligence technology, and calculate the similarity of stores based on the embedding vectors of the stores, thereby achieving accurate quantification of the similarity between stores. This not only improves the processing efficiency of finding similar stores, but also improves the accuracy of finding similar stores. Moreover, compared with the existing solutions for calculating similar stores based on tags or keywords, the embodiments of the present invention obtain a store identification sequence by processing the user's behavior data on the e-commerce platform, and subsequently obtains the store's embedding vector based on the store identification sequence, thereby reducing the difficulty of calculating similar stores and improving the accuracy of the calculation results of similar stores.

[0086] Reference below Figure 6, which shows a schematic structural diagram of a computer system 600 of an electronic device suitable for implementing an embodiment of the present invention. Figure 6 The computer system shown is only an example and should not limit the functions and scope of use of the embodiments of the present invention.

[0087] like Figure 6 As shown, the computer system 600 includes a central processing unit (CPU) 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage unit 608 into a random access memory (RAM) 603. Various programs and data required for the operation of the system 600 are also stored in the RAM 603. The CPU 601, ROM 602, and RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0088] The following components are connected to the I / O interface 605: an input section 606 including a keyboard, a mouse, and the like; an output section 607 including devices such as a cathode ray tube (CRT), a liquid crystal display (LCD), and a speaker; a storage section 608 including a hard disk; and a communication section 609 including a network interface card such as a LAN card or a modem. The communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to the I / O interface 605 as needed. A removable medium 611, such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory, is installed in the drive 610 as needed, so that computer programs read therefrom can be installed into the storage section 608 as needed.

[0089] In particular, according to the embodiments disclosed in the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, the embodiments disclosed in the present invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program comprising a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 609, and / or installed from a removable medium 611. When the computer program is executed by the central processing unit (CPU) 601, the above-mentioned functions defined in the system of the present invention are executed.

[0090] It should be noted that the computer-readable medium described in the present invention can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media can include, but are not limited to, an electrical connection having one or more conductors, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In the present invention, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. This propagated data signal can take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device. Program code embodied on a computer-readable medium may be transmitted using any suitable medium, including but not limited to wireless, wireline, optical fiber cable, RF, or any suitable combination thereof.

[0091] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present invention. In this regard, each box in the flowchart or block diagram can represent a module, program segment, or a part of code, and the above-mentioned module, program segment, or a part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of boxes in the block diagram or flowchart, can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0092] The modules described in the embodiments of the present invention may be implemented in software or hardware. The modules described may also be provided in a processor. For example, they may be described as: a processor including a generation module, a word embedding processing module, and a determination module. The names of these modules do not, in some cases, limit the modules themselves. For example, the generation module may also be described as a "module for generating a sequence of store identifications to be processed."

[0093] As another aspect, the present invention further provides a computer-readable medium, which may be included in the device described in the above embodiment; or may exist independently and not be assembled into the device. The above computer-readable medium carries one or more programs, and when the above one or more programs are executed by a device, the device performs the following process: generating a store identification sequence to be processed based on user behavior data on the e-commerce platform; performing word embedding processing on the store identification sequence to be processed through a word embedding model to obtain an embedding vector of the store; calculating the similarity between stores based on the embedding vector of the store, and then determining similar stores based on the similarity between the stores.

[0094] According to the technical solution of the embodiment of the present invention, the similarity between stores can be accurately quantified, the difficulty of calculating similar stores can be reduced, and the accuracy of the calculation results of similar stores can be improved.

[0095] The above specific embodiments do not limit the scope of protection of the present invention. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may occur depending on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention are intended to be included within the scope of protection of the present invention.

Claims

1. A method for determining similar stores, characterized in that: The method comprises: Generate a store identification sequence to be processed based on the user's behavior data on the e-commerce platform; Performing word embedding processing on the store identification sequence to be processed by a word embedding model to obtain an embedding vector of the store; Calculating the similarity between stores based on the embedding vectors of the stores, and then determining similar stores based on the similarity between the stores; The method of generating a store identification sequence to be processed based on the user's behavior data on the e-commerce platform includes: Obtain browsing behavior records of multiple users on the e-commerce platform N days before a selected date, where the value of N is related to the time at which store similarity is periodically updated; segment the browsing behavior records of the multiple users on the e-commerce platform according to a set time window, and use the store identification sequences obtained by segmentation in which the number of stores is greater than or equal to a preset threshold as the store identification sequence to be processed.

2. The method according to claim 1, characterized in that The user behavior data on the e-commerce platform includes: browsing behavior records of multiple users on the e-commerce platform, purchasing behavior records of multiple users on the e-commerce platform, and add-to-cart behavior records of multiple users on the e-commerce platform.

3. The method according to claim 1, characterized in that The method further comprises: Before segmenting the browsing behavior records of the multiple users on the e-commerce platform according to the set time window, the browsing behavior records of abnormal users are removed from the browsing behavior records of the multiple users on the e-commerce platform.

4. The method according to claim 1, wherein The performing word embedding processing on the store identification sequence to be processed by the word embedding model to obtain the store embedding vector includes: The store identification sequence to be processed is saved as a two-dimensional array; a pre-trained word embedding model is loaded, and the two-dimensional array is input into the pre-trained word embedding model to obtain an embedding vector of the store.

5. The method according to claim 3, characterized in that The word embedding model is the Word2vec model.

6. The method according to claim 1, characterized in that Calculating the similarity between stores based on the embedding vectors of the stores includes: The embedding vectors of the stores are combined in pairs to obtain embedding vector pairs of the stores, and the cosine similarity of the embedding vector pairs of the stores is calculated using the cosine similarity calculation formula.

7. A device for determining similar stores, characterized in that: The device comprises: A generation module, used to generate a store identification sequence to be processed based on the user's behavior data on the e-commerce platform; A word embedding processing module, configured to perform word embedding processing on the store identification sequence to be processed through a word embedding model to obtain an embedding vector of the store; a determination module, configured to calculate the similarity between stores based on the embedding vectors of the stores, and then determine similar stores based on the similarity between the stores; The generation module is also used to: obtain the browsing behavior records of multiple users on the e-commerce platform N days before the selected date, where the value of N is related to the time when the store similarity is updated periodically; segment the browsing behavior records of the multiple users on the e-commerce platform according to the set time window, and use the sequences in the segmented store identification sequences whose number of stores is greater than or equal to a preset threshold as the store identification sequence to be processed.

8. An electronic device, characterized in that: include: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 6.

9. A computer-readable medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images