WO2023108993A1 - Product recommendation method, apparatus and device based on deep clustering algorithm, and medium - Google Patents

Abstract

Embodiments of the present application relate to the technical field of artificial intelligence, and provide a product recommendation method, apparatus and device based on a deep clustering algorithm, and a medium. The method comprises: obtaining recommendation data and historical operation data of products to be recommended; inputting the recommendation data and the historical operation data into a preset product matching model for matching processing to obtain a candidate product set; performing weight computation on the products to be recommended of the candidate product set according to a preset weight algorithm to obtain a weight value of each product to be recommended; selecting a target product from the candidate product set according to the weight values; performing clustering processing on the target product according to a preset product clustering model to obtain a standard product comprising a product category label; and recommending the standard product to users by using a preset product recommendation platform. According to the embodiments of the present application, the recommended product can meet the actual requirements of the users, and the accuracy of product recommendation and the recommendation efficiency are improved.

Description

Product recommendation method, device, equipment and medium based on deep clustering algorithm

This application claims the priority of the Chinese patent application with the application number 202111539013.2 submitted to the China Patent Office on December 15, 2021, and the title of the invention is "Product Recommendation Method, Device, Equipment and Medium Based on Deep Clustering Algorithm", all of which The contents are incorporated by reference in this application.

technical field

The present application relates to the technical field of artificial intelligence, in particular to a product recommendation method, device, equipment and medium based on a deep clustering algorithm.

Background technique

At present, with the development of society and economy, various product series will include one or more products. When users have needs for products, they usually learn about each series of products through the Internet, and cannot quickly obtain information related to their own situation. compliant product.

technical problem

The following are the technical problems of the prior art that the inventors are aware of:

In related technologies, most product recommendation methods often provide a product recommendation entry in the form of a format box, and recommend products for users through the simple information input by the user. Often, the recommended product is not the product or product type that the user is more concerned about. , affecting the accuracy of product recommendation. Therefore, how to make the recommended products more in line with the actual needs of users and improve the accuracy of product recommendation has become an urgent technical problem to be solved.

technical solution

In the first aspect, the embodiment of the present application provides a product recommendation method based on a deep clustering algorithm, including:

Obtain recommended data and historical operating data of products to be recommended;

Inputting the recommendation data and the historical operation data into a preset product matching model for matching processing to obtain a set of candidate products;

performing weight calculation on the products to be recommended in the set of candidate products according to a preset weight algorithm to obtain the weight value of each product to be recommended;

selecting a target product from the set of candidate products according to the weight value;

Perform clustering processing on the target product according to a preset product clustering model to obtain standard products including product category labels;

The standard product is recommended to the user by using a preset product recommendation platform.

In the second aspect, the embodiment of the present application provides a product recommendation device based on a deep clustering algorithm, the device comprising:

The data acquisition module is used to acquire recommended data and historical operating data of products to be recommended;

A product matching module, configured to input the recommended data and the historical operation data into a preset product matching model for matching processing to obtain a set of candidate products;

A weight calculation module, configured to perform weight calculation on the products to be recommended in the set of candidate products according to a preset weight algorithm, to obtain the weight value of each product to be recommended;

a target product determination module, configured to select a target product from the set of candidate products according to the weight value;

A clustering module, configured to perform clustering processing on the target product according to a preset product clustering model to obtain standard products including product category labels;

A product recommendation module, configured to recommend the standard product to the user by using a preset product recommendation platform.

In the third aspect, the embodiment of the present application provides a product recommendation device based on a deep clustering algorithm, the product recommendation device based on a deep clustering algorithm includes a memory, a processor, stored in the memory and can be used in the A program running on the processor and a data bus for realizing connection and communication between the processor and the memory, when the program is executed by the processor, a product recommendation method based on a deep clustering algorithm is implemented, Wherein, the product recommendation method includes:

Obtain recommended data and historical operating data of products to be recommended;

Inputting the recommendation data and the historical operation data into a preset product matching model for matching processing to obtain a set of candidate products;

performing weight calculation on the products to be recommended in the set of candidate products according to a preset weight algorithm to obtain the weight value of each product to be recommended;

selecting a target product from the set of candidate products according to the weight value;

Perform clustering processing on the target product according to a preset product clustering model to obtain standard products including product category labels;

The standard product is recommended to the user by using a preset product recommendation platform.

In a fourth aspect, an embodiment of the present application provides a storage medium, the storage medium is a computer-readable storage medium for computer-readable storage, the storage medium stores one or more programs, and the one or more This program can be executed by one or more processors to implement a product recommendation method based on a deep clustering algorithm, wherein the product recommendation method includes:

Obtain recommended data and historical operating data of products to be recommended;

Inputting the recommendation data and the historical operation data into a preset product matching model for matching processing to obtain a set of candidate products;

performing weight calculation on the products to be recommended in the set of candidate products according to a preset weight algorithm to obtain the weight value of each product to be recommended;

selecting a target product from the set of candidate products according to the weight value;

Perform clustering processing on the target product according to a preset product clustering model to obtain standard products including product category labels;

The standard product is recommended to the user by using a preset product recommendation platform.

Beneficial effect

The product recommendation method, device, equipment and medium based on the deep clustering algorithm proposed by this application obtains the recommended data and the historical operation data of the product to be recommended, and inputs the recommended data and historical operation data into the preset product matching model Perform matching processing to obtain a candidate product set. This method can conveniently screen out products that meet the recommended requirements to form a candidate product set. Then, weight calculation is performed on the products to be recommended in the set of candidate recommended products according to a preset weight algorithm to obtain the weight value of each product to be recommended. Select the target product from the candidate product set according to the weight value. In this way, the products to be recommended in the candidate recommended product set can be further filtered according to the weight value to obtain the target product. This method shortens the screening time of the target product and improves the relationship between the target product and the current recommendation demand. Matching reduces the difficulty of recommendation and saves time and cost. After the target product is obtained, the target product is clustered according to the preset product clustering model, and the standard product including the product category label is obtained, and then the standard product is recommended by the preset product recommendation platform. By clustering the target products, the target products can be clearly classified, so that the basic information of standard products can be reflected more reasonably, and it is convenient for users to choose products. This method can make the recommended products more in line with the actual needs of users, and improve the accuracy and efficiency of product recommendation.

Description of drawings

Fig. 1 is the flowchart of the product recommendation method based on deep clustering algorithm provided by the embodiment of the present application;

Fig. 2 is the flowchart of step S101 in Fig. 1;

Fig. 3 is the flowchart of step S102 in Fig. 1;

Fig. 4 is the flowchart of step S103 in Fig. 1;

Fig. 5 is the flowchart of step S104 in Fig. 1;

Fig. 6 is the flowchart of step S105 in Fig. 1;

Fig. 7 is the flowchart of step S106 in Fig. 1;

FIG. 8 is a flowchart of step S701 in FIG. 7;

FIG. 9 is a schematic structural diagram of a product recommendation device based on a deep clustering algorithm provided by an embodiment of the present application;

FIG. 10 is a schematic diagram of a hardware structure of a product recommendation device based on a deep clustering algorithm provided by an embodiment of the present application.

Embodiments of the present invention

In order to make the purpose, technical solution and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present application, not to limit the present application.

It should be noted that although the functional modules are divided in the schematic diagram of the device, and the logical sequence is shown in the flowchart, in some cases, it can be executed in a different order than the module division in the device or the flowchart in the flowchart. steps shown or described. The terms "first", "second" and the like in the specification and claims and the above drawings are used to distinguish similar objects, and not necessarily used to describe a specific sequence or sequence.

Unless otherwise defined, all technical and scientific terms used in this application have the same meaning as commonly understood by one of ordinary skill in the technical field to which this application belongs. The terms used in this application are only for the purpose of describing the embodiments of this application, and are not intended to limit this application.

First, analyze some nouns involved in this application:

Artificial Intelligence (AI): It is a new technical science that studies and develops theories, methods, technologies and application systems for simulating, extending and expanding human intelligence; artificial intelligence is a branch of computer science. Intelligence attempts to understand the essence of intelligence and produce a new intelligent machine that can respond in a manner similar to human intelligence. Research in this field includes robotics, language recognition, image recognition, natural language processing, and expert systems. Artificial intelligence can simulate the information process of human consciousness and thinking. Artificial intelligence is also a theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

Natural language processing (NLP): NLP uses computers to process, understand and use human languages (such as Chinese, English, etc.). NLP belongs to a branch of artificial intelligence and is an interdisciplinary subject between computer science and linguistics. Known as computational linguistics. Natural language processing includes syntax analysis, semantic analysis, text understanding, etc. Natural language processing is often used in technical fields such as machine translation, handwritten and printed character recognition, speech recognition and text-to-speech conversion, information retrieval, information extraction and filtering, text classification and clustering, public opinion analysis and opinion mining. It involves language processing Related data mining, machine learning, knowledge acquisition, knowledge engineering, artificial intelligence research and linguistics research related to language computing, etc.

Information Extraction (Information Extraction, NER): A text processing technology that extracts specified types of factual information such as entities, relationships, and events from natural language texts, and forms structured data output. Information extraction is a technique to extract specific information from text data. Text data is composed of some specific units, such as sentences, paragraphs, and chapters. Text information is composed of some small specific units, such as words, words, phrases, sentences, paragraphs, or combinations of these specific units. . Extracting noun phrases, personal names, and place names in text data is all text information extraction. Of course, the information extracted by text information extraction technology can be various types of information.

Maximum Entropy Markov Model (MEMM): It is used to calculate the conditional probability distribution of each hidden state sequence Y for a given observation sequence X. It is to establish a joint probability for transition probability and performance probability. The statistics are conditional probabilities, not co-occurrence probabilities. Since MEMM only performs local normalization, MEMM is easy to fall into local optimum.

Conditional random field algorithm (conditional random field algorithm, CRF): It is a mathematical algorithm; it combines the characteristics of the maximum entropy model and the hidden Markov model, and is an undirected graph model. It has achieved good results in sequence labeling tasks such as entity recognition. The conditional random field is a typical discriminant model, and its joint probability can be written as the multiplication of several potential functions, the most commonly used of which is the linear chain conditional random field. If x=(x1, x2,...xn) represents the observed input data sequence, y=(y1, y2,...yn) represents a state sequence, given an input sequence, the CRF model of the linear chain Define the joint conditional probability of the state sequence as p(y|x)=exp{}(2-14); Z(x)={}(2-15); where: Z is the probability normalization conditional on the observation sequence x Normalization factor; fj(yi-1, yi, x, i) is an arbitrary feature function.

Long Short-Term Memory (LSTM): It is a time cyclic neural network, which is specially designed to solve the long-term dependence problem of general RNN (cyclic neural network). All RNNs have a A chain form of repeated neural network modules. In standard RNNs, this repeated structural module has only a very simple structure, such as a tanh layer. LSTM is a type of neural network that contains LSTM blocks (blocks) or others. In literature or other materials, LSTM blocks may be described as intelligent network units because they can memorize values for an indefinite length of time. There is a The gate can determine whether the input is important enough to be remembered and whether it can be output.

Bi-directional Long Short-Term Memory (Bi-LSTM): It is a combination of forward LSTM and backward LSTM. It is often used to model contextual information in natural language processing tasks. On the basis of LSTM, Bi-LSTM combines the information of the input sequence in both forward and backward directions. For the output at time t, the forward LSTM layer has the information of time t and the previous time in the input sequence, and the backward LSTM layer has the information of time t and the subsequent time in the input sequence. The output of the forward LSTM layer at time t is denoted as , and the output result of the backward LSTM layer at time t is denoted as , and the vectors output by the two LSTM layers can be processed by addition, average or connection.

BERT (Bidirectional Encoder Representations from Transformers): It is a language representation model (language representation model). BERT uses the Transformer Encoder block for connection, which is a typical two-way encoding model.

The product recommendation method, device, device, and medium based on the deep clustering algorithm provided in the embodiments of the present application are specifically described through the following embodiments. First, the product recommendation method in the embodiments of the present application is described.

The embodiments of the present application may acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results. .

Artificial intelligence basic technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation/interaction systems, and mechatronics. Artificial intelligence software technology mainly includes computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning/deep learning.

The product recommendation method provided in the embodiment of the present application relates to the technical field of artificial intelligence. The product recommendation method provided in the embodiment of the present application may be applied to a terminal, may also be applied to a server, and may also be software running on the terminal or the server. In some embodiments, the terminal can be a smart phone, a tablet computer, a notebook computer, a desktop computer, etc.; the server end can be configured as an independent physical server, or can be configured as a server cluster or a distributed system composed of multiple physical servers, or It can be configured as a cloud that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The server; the software may be an application for realizing the product recommendation method, but is not limited to the above forms.

The application can be used in numerous general purpose or special purpose computer system environments or configurations. Examples: personal computers, server computers, handheld or portable devices, tablet-type devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, including A distributed computing environment for any of the above systems or devices, etc. This application may be described in the general context of computer-executable instructions, such as program modules, being executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. The application may also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media including storage devices.

FIG. 1 is an optional flow chart of a product recommendation method provided by an embodiment of the present application. The method in FIG. 1 may include, but is not limited to, step S101 to step S106.

Step S101, obtaining recommended data and historical operation data of products to be recommended;

Step S102, inputting the recommendation data and the historical operation data into a preset product matching model for matching processing to obtain a set of candidate products;

Step S103, perform weight calculation on the products to be recommended in the set of candidate products according to a preset weight algorithm, and obtain the weight value of each product to be recommended;

Step S104, selecting a target product from the candidate product set according to the weight value;

Step S105, clustering the target product according to a preset product clustering model to obtain a standard product including a product category label;

Step S106, using a preset product recommendation platform to recommend the standard product to the user. Through the above steps S101 to S106, the products meeting the recommended requirements can be conveniently screened out to form a candidate product set. Then, weight calculation is performed on the products to be recommended in the set of candidate recommended products according to a preset weight algorithm to obtain the weight value of each product to be recommended. According to the weight value, the products to be recommended in the candidate recommended product set are filtered to obtain the target product. This method shortens the screening time of the target product and improves the matching between the target product and the current recommendation demand. Through the clustering process of target products, the target products can be clearly classified into products, so that the basic information of standard products can be reflected more reasonably, and it is convenient for users to choose products, so that the recommended products are more in line with the actual needs of users and improve Product recommendation accuracy and recommendation efficiency.

Referring to FIG. 2, in some embodiments, step S101 may include but not limited to include steps S201 to S202:

Step S201, obtaining preset target demand dimensions;

Step S202, crawling recommendation data and historical operation data corresponding to each target demand dimension by means of a web crawler.

In order to improve the accuracy of product recommendation, it is necessary to obtain recommendation data and historical operation data of multiple demand dimensions, that is, firstly, it is necessary to obtain the target demand dimension, which includes time dimension, product dimension, and so on. In different demand dimensions, by writing a web crawler and setting up data sources, crawl data in a targeted manner to obtain recommended data and historical operation data under each target demand dimension. Among them, the recommendation data includes the expected recommendation time, target customer group data, recommendation theme data, recommendation purpose data, etc., and the historical operation data includes the historical sales volume, sales area, attribute data, etc. of the recommended products; for example, the acquisition time dimension Recommendation time; obtain product data, target customer group data, recommended theme data, recommendation purpose data, etc. under the product dimension.

Further, in step S202, for the recommended data of some target demand dimensions, corresponding demand priority orders may also be set, so as to further improve recommendation accuracy. Specifically, specific recommendation data in the product demand dimension is selected, and the recommendation data may be target customer group data and recommendation purpose data in the product dimension. Furthermore, these specific recommendation data are sorted according to a preset priority order. For example, in the target customer group data, the order of priority is that the age of the recommended object is greater than the income of the recommended object, the income of the recommended object is greater than the health status of the recommended object, the health status of the recommended object is greater than the location of the recommended object, and the location of the recommended object The region is greater than the family status of the recommended object; in the recommendation purpose data, the priority order is that attracting customers is greater than paying conversion, paying conversion is greater than improving retention, improving retention is greater than increasing usage time, and so on.

In addition, in the embodiment of the present application, it is also necessary to write a web crawler and crawl data with a purpose according to the set data source to obtain the historical operation data of the product to be recommended. Among them, historical operation data includes historical recommendation scene data, historical user rating data, high-frequency user portrait data, user churn rate, and high-efficiency application period data, etc.

Referring to FIG. 3, in some embodiments, step S102 may include but not limited to include steps S301 to S303:

Step S301, matching the recommendation data and historical operation data to obtain the matching value of each product to be recommended;

Step S302, selecting candidate products according to the size relationship between the matching value and the preset matching threshold;

Step S303, incorporating multiple candidate products into the same set to obtain a set of candidate products.

In order to improve the matching accuracy, before performing step S301, it is necessary to label the recommended data and the historical operation data respectively to obtain the labeled recommendation data and labeled operation data, wherein the labeled recommended data has a first labeled field, and labeled operational data There is a second labeling field on it, and the specific content of the first labeling field and the second labeling field can be determined according to preset keywords and the like, without limitation.

In step S301, the product matching model may be an ESIM model, and the product matching model includes multiple convolutional layers and pooling layers. The product matching model can perform convolution processing and pooling processing on the labeling operation data and labeling recommendation data respectively, extract the first labeling field in the labeling operation data and the second labeling field on the labeling recommendation data, and combine the first labeling field and the second labeling field The two marked fields are compared to confirm the consistency between the first marked field and the second marked field. If the first label field is consistent with the second label field, the comparison value is recorded as 1, and if the first label field is inconsistent with the second label field, the comparison value is recorded as 0. Through this comparison and marking method, traverse all the first label fields and the second label fields to obtain multiple comparison values, and sum all the comparison values to obtain the matching value of each product to be recommended.

Further, since in step S202, the recommended data on some target demand dimensions is set with a demand priority order, so when matching the recommended data with the historical operation data, it is possible to label the No. The first label field is compared with the second label field on the label recommendation data one by one to improve matching efficiency.

Furthermore, step S302 and step S303 are executed to compare the matching value with the preset matching threshold. If the matching value is greater than or equal to the preset matching threshold, it indicates that the product to be recommended has a high correlation with the current recommendation demand. Therefore, Take this product to be recommended as a candidate product. If the matching value is less than the preset matching threshold, it indicates that the product to be recommended is not highly relevant to the current recommendation demand, and the product to be recommended will not be considered. Furthermore, the selected multiple candidate products are statistically summarized to obtain a set of candidate products. For example, if the matching value of the product to be recommended is greater than 3, it indicates that the product to be recommended matches the current recommendation demand in at least three demand dimensions, and the correlation is high. The products are included in the same set to form a set of candidate products, so as to further screen the products to be recommended in the set of candidate products, so that the recommended products are more in line with the actual needs of users, and the accuracy of product recommendation is improved.

Referring to FIG. 4, in some embodiments, step S103 may include but not limited to include steps S401 to S402:

Step S401, obtaining the priority weight, matching value weight and product basic score of the product to be recommended;

Step S402, according to the preset weighting algorithm, perform weighted calculation on the priority weight, matching value weight and product basic score to obtain the weight value of each product to be recommended.

In order to improve the accuracy of recommendation, in step S401, score calculation is performed on the product to be recommended in terms of product characteristics, historical operation data performance of the product, and matching degree of recommendation requirements. Among them, there are preset product basic points in terms of product characteristics, floating points in terms of product historical operation data performance, and matching weighted points in terms of matching degree of recommended requirements. Among them, matching weighting is divided into priorities The average weight of the weight and the matching value weight.

In step S402, according to a preset weighting algorithm, the weight value of each product to be marketed is calculated by product basic score×floating score×matching weighted score. For example, set the product basic score of a series of shoulder and neck health products to 10 points. Among them, the historical operation data of a product to be recommended performs better, with a floating score of 0.83, a priority weight of 1, and a matching value weight of 1.1. The matching weight is divided into the average weight of the priority weight and the matching value weight, that is, the matching weight is divided into 1.05; the weight calculation of the product to be recommended is: weight value = basic score 10 × floating score 0.83 × matching weighted score weight 1.05 = 8.7 .

Referring to FIG. 5, in some embodiments, step S104 may also include but not limited to include steps S501 to S502:

Step S501, sort the products to be recommended in descending order according to the weight value, and obtain the sequence of products to be recommended;

In step S502, the products to be recommended in the sequence of products to be recommended are screened according to preset screening conditions to obtain target products.

In order to improve the recommendation efficiency, in step S501, compare the weight values of all products to be marketed in the candidate product set, sort the products to be recommended in descending order according to the weight value from large to small, and obtain a sequence of products to be recommended.

Furthermore, step S502 is executed, and the preset screening conditions may include screening quantity, weight threshold and so on. The products to be recommended in the sequence of products to be recommended are screened according to the screening data, the weight threshold, and the like. For example, if the number of required products under a certain recommendation data is 3, then the number of products to be screened under the current filtering conditions is 3, select the products to be recommended with the top three weight values in the sequence of products to be recommended, and combine these three products to be recommended product as the target product. In addition, the weight threshold can also be set, and the target product can be selected by combining the weight threshold and the quantity of demand. For example, the number of demanded products under a certain recommended data is 10, and the weight value of the product sequence to be recommended ranks in the top ten. For the products to be recommended, compare the weight value of the ten products to be recommended with the weight threshold. If the weight value is less than the weight threshold, the corresponding product to be recommended will be eliminated, and only the 10 products to be recommended with a weight value greater than or equal to The weight threshold, the products to be recommended whose weight value is greater than or equal to the weight threshold are taken as target products.

Through the above steps S401 to S402 and steps S501 to S502, the weight calculation of the products to be recommended in the candidate recommended product set can be performed according to the preset weight algorithm, and the weight value of each product to be recommended can be obtained, and according to the weight value from Select the target product from the candidate product set. Filtering the products to be recommended in the candidate recommended product set by weight values shortens the screening time of target products, improves the matching between target products and current recommendation requirements, reduces the difficulty of recommendation, and saves time and cost.

Referring to FIG. 6, step S105 in some embodiments may include but not limited to steps S601 to S602:

Step S601, inputting the preset product category label into the product clustering model to obtain the model label of the product clustering model;

In step S602, the target product is clustered according to the K-means clustering algorithm and the model label to obtain a standard product.

In step S601, the preset product category tags include interactive entertainment content difficulty, UI change difficulty, interactive entertainment duration, etc., wherein the interactive entertainment content difficulty level, interactive entertainment duration can be set according to actual conditions, for example, interactive entertainment Content difficulty levels include easy, medium, and difficult; interactive entertainment duration (ie game duration) includes less than 30 seconds, 30 seconds to 120 seconds, more than 120 seconds, etc., but is not limited thereto. By inputting product category labels into the product clustering model, the model labels of the product clustering model are attached with preset product categories, so that the product clustering model can cluster the target products according to these preset product categories , to improve the clustering accuracy.

Furthermore, step S602 is executed to extract features of the target product according to the K-means clustering algorithm to obtain feature data of each target product, wherein the feature data includes feature coordinate values. According to the characteristic coordinate value, mark the position of each target product on the preset product cluster map. Furthermore, the Euclidean distance from each target product to multiple reference seed points on the cluster feature map is obtained, and according to the Euclidean distance from each target product to multiple reference seed points on the cluster feature map, the minimum Euclidean distance The reference seed point corresponding to the Euclidean distance is used as the target position of the target product, that is, the target product is moved to the reference seed point corresponding to the minimum Euclidean distance, thereby obtaining multiple product clusters. The product clusters with model labels (that is, the preset product category labels including the difficulty of interactive entertainment content, the difficulty of UI changes, the duration of interactive entertainment, etc.) are used as target product clusters, and the target products in these target product clusters are standard products. Through the clustering process of the target product, the product classification of the target product can be clearly carried out, so that the basic information of the standard product can be reflected more reasonably, so that the recommended product is more in line with the actual needs of users, and the accuracy of product recommendation is improved. performance and recommendation efficiency.

Referring to FIG. 7, in some embodiments, step S106 may include but not limited to include steps S701 to S703:

Step S701, extracting entity features from historical operating data of standard products to obtain target operating data;

Step S702, visualizing the target operation data to generate a product recommendation report;

Step S703, uploading the product recommendation report to the product recommendation platform to recommend standard products to users.

In order to improve the efficiency of data acquisition, in step S701, the target text data in the historical operation data of standard products is extracted, and the entity features in the target text data are identified using the preset lexical analysis model, and then the entity features are classified and characterized. Extract to obtain the target operation data. This method can reduce the total amount of data, making it easier to extract the required target operation data.

Furthermore, step S702 is executed to perform multi-dimensional analysis on the target operation data in the form of charts, etc., and extract keyword segments in the target operation data, wherein the multi-dimensional analysis on the target operation data includes drilling down, scrolling up, and rotating the target operation data , slicing, linkage processing, etc. Furthermore, the extracted keyword segments are combined to generate a product recommendation report. The product recommendation report includes product charts and basic data such as the product name and product weight value of each standard product, which can more clearly reflect the basic information of standard products.

Finally, step S703 is executed, and the generated product recommendation report is uploaded to the product recommendation platform, and standard products are recommended through the product recommendation platform. It should be noted that standard products can also be recommended through various recommendation channels such as mobile APP application market and various social platforms, so that the recommendation forms are diversified, so that users can more easily pay attention to the currently recommended products, which is convenient Users make product selections to improve recommendation efficiency.

Referring to FIG. 8, in some embodiments, step S701 may include but not limited to include steps S801 to S804:

Step S801, extracting target text data in historical operation data;

Step S802, using a preset lexical analysis model to identify entity features in the target text data;

Step S803, using a pre-trained sequence classifier to classify entity features;

Step S804, performing feature extraction on the entity features after classification processing to obtain target operation data.

In step S801, the unstructured data in the historical operation data is converted into unified structured data, and the required target text data is extracted from the structured data, wherein the target text data is natural language text.

Furthermore, step S802 is executed, using a preset lexical analysis model to identify entity features in the target text data. For example, a product recommendation data thesaurus is pre-built, and the product recommendation data thesaurus may include proper nouns, terms, non-proprietary names, etc. related to various product recommendations. Through this product recommendation data lexicon, the preset lexical analysis model can list specific product recommendation names, for example, recommendation objects, recommendation scenarios, and so on. Input the target text data into the preset lexical analysis model, and identify the entity features in the target text data through the specific product recommendation corpus and preset part-of-speech categories contained in the preset lexical analysis model. The entity features can be Including the above-mentioned entity vocabulary related to proper nouns, terms, non-proper names, modifiers, and time information related to product recommendations.

In order to extract entity features more accurately, a pre-trained sequence classifier can also be used to mark entity features, so that these entity features can carry preset labels to improve classification efficiency.

When performing step S803, the pre-trained sequence classifier can be a maximum entropy Markov model (MEMM model) or a model based on conditional random field algorithm (CRF) or a model based on bidirectional long short-term memory algorithm (bi-LSTM) . For example, a sequence classifier can be constructed based on the bi-LSTM algorithm. In the model based on the bi-LSTM algorithm, the input word wi and character embedding are passed through left-to-right long-short-term memory and right-to-left long-short-term memory, so that the output is The connected locations generate a single output layer. The sequence classifier can pass the input entity features directly to the softmax classifier through this output layer, and create a probability distribution on the preset part-of-speech category label through the softmax classifier, so as to mark and classify the entity parameters according to the probability distribution. Finally, step S804 is executed to perform feature extraction on the entity features after classification processing to obtain the required target operation data.

In addition, in order to achieve data storage, the BERT encoder can also be used to convert the target operating data from text form to encoded form through the preset encoding function, so as to realize the storage of target operating data. This method can realize the feature extraction of historical operation data, reduce the total amount of data, and make it more convenient to extract the required target operation data.

In this embodiment of the present application, by obtaining the recommended data and the historical operation data of the products to be recommended, the recommended data and historical operation data are input into the preset product matching model for matching processing to obtain a set of candidate products. The recommended products are screened out to form a set of candidate products. Then, weight calculation is performed on the products to be recommended in the set of candidate recommended products according to a preset weight algorithm to obtain the weight value of each product to be recommended. Select the target product from the candidate product set according to the weight value. In this way, the products to be recommended in the candidate recommended product set can be further filtered according to the weight value to obtain the target product. This method shortens the screening time of the target product and improves the relationship between the target product and the current recommendation demand. Matching reduces the difficulty of recommendation and saves time and cost. After the target product is obtained, the target product is clustered according to the preset product clustering model, and the standard product including the product category label is obtained, and then the standard product is recommended by the preset product recommendation platform. By clustering the target products, the target products can be clearly classified, so that the basic information of standard products can be reflected more reasonably, and it is convenient for users to choose products. This method can make the recommended products more in line with the actual needs of users, and improve the accuracy and efficiency of product recommendation.

Please refer to FIG. 9 , the embodiment of the present application also provides a product recommendation device, which can implement the above product recommendation method, and the device includes:

A data acquisition module 901, configured to acquire recommended data and historical operating data of products to be recommended;

A product matching module 902, configured to input the recommended data and the historical operation data into a preset product matching model for matching processing to obtain a set of candidate products;

A weight calculation module 903, configured to perform weight calculation on the products to be recommended in the candidate product set according to a preset weight algorithm, to obtain the weight value of each product to be recommended;

A target product determination module 904, configured to select a target product from the set of candidate products according to the weight value;

A clustering module 905, configured to perform clustering processing on the target product according to a preset product clustering model to obtain standard products including product category labels;

The product recommendation module 906 is configured to recommend the standard product to the user using a preset product recommendation platform.

The specific implementation manner of the product recommendation device is basically the same as the specific embodiment of the above product recommendation method, and will not be repeated here.

The embodiment of the present application also provides a product recommendation device based on a deep clustering algorithm. The product recommendation device based on a deep clustering algorithm includes: a memory, a processor, a program stored in the memory and operable on the processor, and a user In order to implement the data bus connecting and communicating between the processor and the memory, when the program is executed by the processor, a product recommendation method based on a deep clustering algorithm is implemented, wherein the product recommendation method includes: obtaining recommended data and products to be recommended the historical operation data; input the recommendation data and the historical operation data into the preset product matching model for matching processing, and obtain the candidate product set; according to the preset weight algorithm, the candidate product set to be recommended Perform weight calculation on the product to obtain the weight value of each product to be recommended; select the target product from the candidate product set according to the weight value; perform clustering processing on the target product according to the preset product clustering model, A standard product including a product category label is obtained; and the standard product is recommended to the user by using a preset product recommendation platform. The recommended device for this product can be any smart terminal including tablet PCs and vehicle-mounted computers.

Please refer to FIG. 10. FIG. 10 illustrates the hardware structure of a product recommendation device based on a deep clustering algorithm in another embodiment. The product recommendation device includes:

The processor 1001 may be implemented by a general-purpose CPU (Central Processing Unit, central processing unit), a microprocessor, an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), or one or more integrated circuits, and is used to execute related programs to realize The technical solutions provided by the embodiments of the present application;

The memory 1002 may be implemented in the form of a read-only memory (ReadOnlyMemory, ROM), a static storage device, a dynamic storage device, or a random access memory (RandomAccessMemory, RAM). The memory 1002 can store operating systems and other application programs. When implementing the technical solutions provided by the embodiments of this specification through software or firmware, the relevant program codes are stored in the memory 1002 and called by the processor 1001 to execute the implementation of the present application. Examples of product recommendation methods;

Input/output interface 1003, used to realize information input and output;

The communication interface 1004 is used to realize the communication interaction between the device and other devices, and the communication can be realized through a wired method (such as USB, network cable, etc.), or can be realized through a wireless method (such as a mobile network, WIFI, Bluetooth, etc.); and

A bus 1005, which transmits information between various components of the device (such as a processor 1001, a memory 1002, an input/output interface 1003, and a communication interface 1004);

The processor 1001 , the memory 1002 , the input/output interface 1003 and the communication interface 1004 are connected to each other within the device through the bus 1005 .

The embodiment of the present application also provides a storage medium, the storage medium is a computer-readable storage medium for computer-readable storage, the storage medium stores one or more programs, and one or more programs can be processed by one or more to implement a product recommendation method based on a deep clustering algorithm, wherein the product recommendation method includes: obtaining recommended data and historical operating data of the product to be recommended; inputting the recommended data and the historical operating data Perform matching processing in the preset product matching model to obtain a set of candidate products; perform weight calculation on the products to be recommended in the set of candidate products according to the preset weight algorithm, and obtain the weight value of each product to be recommended; The weight value is used to select the target product from the set of candidate products; the target product is clustered according to the preset product clustering model to obtain a standard product containing the product category label; the preset product recommendation platform is used to The standard product is recommended to the user. In addition, the computer-readable storage medium may be non-volatile or volatile.

As a non-transitory computer-readable storage medium, memory can be used to store non-transitory software programs and non-transitory computer-executable programs. In addition, the memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one magnetic disk storage device, flash memory device, or other non-transitory solid-state storage devices. In some embodiments, the memory optionally includes memory located remotely from the processor, and these remote memories may be connected to the processor via a network. Examples of the aforementioned networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

The embodiments described in the embodiments of the present application are to illustrate the technical solutions of the embodiments of the present application more clearly, and do not constitute a limitation to the technical solutions provided by the embodiments of the present application. Those skilled in the art know that with the evolution of technology and new For the emergence of application scenarios, the technical solutions provided by the embodiments of the present application are also applicable to similar technical problems.

Those skilled in the art can understand that the technical solutions shown in Figures 1-8 do not constitute a limitation to the embodiments of the present application, and may include more or fewer steps than those shown in the illustrations, or combine certain steps, or be different A step of.

The device embodiments described above are only illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place, or may be distributed to multiple network units. Part or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

Those of ordinary skill in the art can understand that all or some of the steps in the methods disclosed above, the functional modules/units in the system, and the device can be implemented as software, firmware, hardware, and an appropriate combination thereof.

The terms "first", "second", "third", "fourth", etc. (if any) in the description of the present application and the above drawings are used to distinguish similar objects and not necessarily to describe specific sequence or sequence. It is to be understood that the data so used are interchangeable under appropriate circumstances such that the embodiments of the application described herein can be practiced in sequences other than those illustrated or described herein. Furthermore, the terms "comprising" and "having", as well as any variations thereof, are intended to cover a non-exclusive inclusion, for example, a process, method, system, product or device comprising a sequence of steps or elements is not necessarily limited to the expressly listed instead, may include other steps or elements not explicitly listed or inherent to the process, method, product or apparatus.

It should be understood that in this application, "at least one (item)" means one or more, and "multiple" means two or more. "And/or" is used to describe the association relationship of associated objects, indicating that there can be three types of relationships, for example, "A and/or B" can mean: only A exists, only B exists, and A and B exist at the same time , where A and B can be singular or plural. The character "/" generally indicates that the contextual objects are an "or" relationship. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one item (piece) of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c ", where a, b, c can be single or multiple.

In the several embodiments provided in this application, it should be understood that the disclosed devices and methods may be implemented in other ways. For example, the device embodiments described above are only illustrative. For example, the division of the above units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or can be Integrate into another system, or some features may be ignored, or not implemented. In another point, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces, and the indirect coupling or communication connection of devices or units may be in electrical, mechanical or other forms.

The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, each unit may exist separately physically, or two or more units may be integrated into one unit. The above-mentioned integrated units can be implemented in the form of hardware or in the form of software functional units.

If the integrated unit is realized in the form of a software function unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or part of the contribution to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium , including multiple instructions to make a computer device (which may be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the method in each embodiment of the present application. The aforementioned storage media include: U disk, mobile hard disk, read-only memory (Read-Only Memory, referred to as ROM), random access memory (Random Access Memory, referred to as RAM), magnetic disk or optical disc, etc., which can store programs. medium.

The preferred embodiments of the embodiments of the present application have been described above with reference to the accompanying drawings, which does not limit the scope of rights of the embodiments of the present application. Any modifications, equivalent replacements and improvements made by those skilled in the art without departing from the scope and essence of the embodiments of the present application shall fall within the scope of rights of the embodiments of the present application.

Claims (20)

1. A product recommendation method based on a deep clustering algorithm, wherein the method includes:

   Obtain recommended data and historical operating data of products to be recommended;

   Inputting the recommendation data and the historical operation data into a preset product matching model for matching processing to obtain a set of candidate products;

   performing weight calculation on the products to be recommended in the set of candidate products according to a preset weight algorithm to obtain the weight value of each product to be recommended;

   selecting a target product from the set of candidate products according to the weight value;

   Perform clustering processing on the target product according to a preset product clustering model to obtain standard products including product category labels;

   The standard product is recommended to the user by using a preset product recommendation platform.
2. The product recommendation method according to claim 1, wherein the step of acquiring recommendation data and historical operation data of the product to be recommended comprises:

   Obtain the preset target demand dimension;

   Crawling the recommendation data and the historical operation data corresponding to each target demand dimension by means of a web crawler.
3. The product recommendation method according to claim 1, wherein the step of inputting the recommendation data and the historical operation data into a preset product matching model for matching processing to obtain a set of candidate products includes:

   Perform matching processing on the recommended data and the historical operation data to obtain the matching value of each product to be recommended;

   Selecting candidate products according to the size relationship between the matching value and a preset matching threshold;

   A plurality of the candidate products are included in the same set to obtain the set of candidate products.
4. The product recommendation method according to claim 1, wherein the step of performing weight calculation on the products to be recommended in the set of candidate products according to a preset weight algorithm to obtain the weight value of each product to be recommended includes:

   Acquiring the priority weight, matching value weight and product basic score of the product to be recommended;

   According to a preset weighting algorithm, weighted calculation is performed on the priority weight, the matching value weight and the product basic score to obtain the weight value of each product to be recommended.
5. The product recommendation method according to claim 1, wherein the step of selecting a target product from the candidate product set according to the weight value comprises:

   sorting the products to be recommended in descending order according to the weight value to obtain a sequence of products to be recommended;

   The products to be recommended in the sequence of products to be recommended are screened according to preset screening conditions to obtain target products.
6. The product recommendation method according to claim 1, wherein the step of performing clustering processing on the target product according to a preset product clustering model to obtain a standard product containing a product category label includes:

   inputting preset product category labels into the product clustering model to obtain model labels of the product clustering model;

   The target product is clustered according to the K-means clustering algorithm and the model label to obtain the standard product.
7. The product recommendation method according to any one of claims 1 to 6, wherein the step of recommending the standard product to the user using a preset product recommendation platform includes:

   Extracting entity features from the historical operating data of the standard product to obtain target operating data;

   Visualize the target operation data and generate a product recommendation report;

   uploading the product recommendation report to the product recommendation platform to recommend the standard product to the user.
8. A product recommendation device based on a deep clustering algorithm, wherein the device includes:

   The data acquisition module is used to acquire recommended data and historical operation data of products to be recommended;

   A product matching module, configured to input the recommended data and the historical operation data into a preset product matching model for matching processing to obtain a set of candidate products;

   The weight calculation module is used to carry out weight calculation to the products to be recommended in the set of candidate products according to a preset weight algorithm to obtain the weight value of each product to be recommended;

   a target product determination module, configured to select a target product from the set of candidate products according to the weight value;

   A clustering module, configured to perform clustering processing on the target product according to a preset product clustering model to obtain standard products including product category labels;

   A product recommendation module, configured to recommend the standard product to the user by using a preset product recommendation platform.
9. A product recommendation device based on a deep clustering algorithm, wherein the product recommendation device based on a deep clustering algorithm includes a memory, a processor, a program stored in the memory and operable on the processor, and a user In order to realize the data bus connecting and communicating between the processor and the memory, when the program is executed by the processor, a product recommendation method based on a deep clustering algorithm is implemented, wherein the product recommendation method includes :

   Obtain recommended data and historical operating data of products to be recommended;

   Inputting the recommendation data and the historical operation data into a preset product matching model for matching processing to obtain a set of candidate products;

   performing weight calculation on the products to be recommended in the set of candidate products according to a preset weight algorithm to obtain the weight value of each product to be recommended;

   selecting a target product from the set of candidate products according to the weight value;

   Perform clustering processing on the target product according to a preset product clustering model to obtain standard products including product category labels;

   The standard product is recommended to the user by using a preset product recommendation platform.
10. The product recommendation device according to claim 9, wherein the step of acquiring recommendation data and historical operation data of the product to be recommended comprises:

    Obtain the preset target demand dimension;

    Crawling the recommendation data and the historical operation data corresponding to each target demand dimension by means of a web crawler.
11. The product recommendation device according to claim 9, wherein the step of inputting the recommendation data and the historical operation data into a preset product matching model for matching processing to obtain a set of candidate products includes:

    Perform matching processing on the recommended data and the historical operation data to obtain the matching value of each product to be recommended;

    Selecting candidate products according to the size relationship between the matching value and a preset matching threshold;

    A plurality of the candidate products are included in the same set to obtain the set of candidate products.
12. The product recommendation device according to claim 9, wherein the step of performing weight calculation on the products to be recommended in the set of candidate products according to a preset weight algorithm to obtain the weight value of each product to be recommended includes:

    Acquiring the priority weight, matching value weight and product basic score of the product to be recommended;

    According to a preset weighting algorithm, weighted calculation is performed on the priority weight, the matching value weight and the product basic score to obtain the weight value of each product to be recommended.
13. The product recommendation device according to claim 9, wherein the step of selecting a target product from the set of candidate products according to the weight value comprises:

    sorting the products to be recommended in descending order according to the weight value to obtain a sequence of products to be recommended;

    The products to be recommended in the sequence of products to be recommended are screened according to preset screening conditions to obtain target products.
14. The product recommendation device according to claim 9, wherein the step of performing clustering processing on the target product according to a preset product clustering model to obtain a standard product including a product category label includes:

    inputting preset product category labels into the product clustering model to obtain model labels of the product clustering model;

    The target product is clustered according to the K-means clustering algorithm and the model label to obtain the standard product.
15. A storage medium, the storage medium is a computer-readable storage medium for computer-readable storage, wherein the storage medium stores one or more programs, and the one or more programs can be used by one or more The processor executes to implement a product recommendation method based on a deep clustering algorithm, wherein the product recommendation method includes:

    Obtain recommended data and historical operating data of products to be recommended;

    Inputting the recommendation data and the historical operation data into a preset product matching model for matching processing to obtain a set of candidate products;

    performing weight calculation on the products to be recommended in the set of candidate products according to a preset weight algorithm to obtain the weight value of each product to be recommended;

    selecting a target product from the set of candidate products according to the weight value;

    Perform clustering processing on the target product according to a preset product clustering model to obtain standard products including product category labels;

    The standard product is recommended to the user by using a preset product recommendation platform.
16. The storage medium according to claim 15, wherein the step of acquiring recommendation data and historical operation data of the product to be recommended comprises:

    Obtain the preset target demand dimension;

    Crawling the recommendation data and the historical operation data corresponding to each target demand dimension by means of a web crawler.
17. The storage medium according to claim 15, wherein the step of inputting the recommended data and the historical operation data into a preset product matching model for matching processing to obtain a set of candidate products includes:

    Perform matching processing on the recommended data and the historical operation data to obtain the matching value of each product to be recommended;

    Selecting candidate products according to the size relationship between the matching value and a preset matching threshold;

    A plurality of the candidate products are included in the same set to obtain the set of candidate products.
18. The storage medium according to claim 15, wherein the step of performing weight calculation on the products to be recommended in the set of candidate products according to a preset weight algorithm to obtain the weight value of each product to be recommended includes:

    Acquiring the priority weight, matching value weight and product basic score of the product to be recommended;

    According to a preset weighting algorithm, weighted calculation is performed on the priority weight, the matching value weight and the product basic score to obtain the weight value of each product to be recommended.
19. The storage medium according to claim 15, wherein the step of selecting a target product from the set of candidate products according to the weight value comprises:

    sorting the products to be recommended in descending order according to the weight value to obtain a sequence of products to be recommended;

    The products to be recommended in the sequence of products to be recommended are screened according to preset screening conditions to obtain target products.
20. The storage medium according to claim 15, wherein the step of clustering the target product according to a preset product clustering model to obtain a standard product containing a product category label includes:

    inputting preset product category labels into the product clustering model to obtain model labels of the product clustering model;

    The target product is clustered according to the K-means clustering algorithm and the model label to obtain the standard product.

Images