CN115131111A - Product recommendation method, device, equipment and storage medium based on e-commerce platform - Google Patents
Product recommendation method, device, equipment and storage medium based on e-commerce platform Download PDFInfo
- Publication number
- CN115131111A CN115131111A CN202211051024.0A CN202211051024A CN115131111A CN 115131111 A CN115131111 A CN 115131111A CN 202211051024 A CN202211051024 A CN 202211051024A CN 115131111 A CN115131111 A CN 115131111A
- Authority
- CN
- China
- Prior art keywords
- product
- user
- commerce platform
- target
- account
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0631—Item recommendations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9536—Search customisation based on social or collaborative filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
- G06Q30/0202—Market predictions or forecasting for commercial activities
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Finance (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Strategic Management (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- Marketing (AREA)
- Economics (AREA)
- Data Mining & Analysis (AREA)
- General Business, Economics & Management (AREA)
- Entrepreneurship & Innovation (AREA)
- General Engineering & Computer Science (AREA)
- Game Theory and Decision Science (AREA)
- Human Computer Interaction (AREA)
- Multimedia (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a product recommendation method, a device, equipment and a storage medium based on an e-commerce platform, wherein the method comprises the following steps: identifying the user type according to the image information of the user to obtain a target user type, and determining a product recommendation range corresponding to the target user type; acquiring user communication data of a current user account, determining all friend accounts in the user communication data, and taking the current user account and the friend accounts as target accounts; traversing each target account, collecting browsing information of the traversed target account in a preset e-commerce platform, and constructing recommended product content according to the browsing information and the product recommendation range; inputting the recommended product content into a preset cyclic neural network model for model training to obtain a prediction probability vector; and sequentially fusing the prediction probability vectors corresponding to the target account numbers to obtain all prediction probabilities, and recommending the product corresponding to the maximum prediction probability in the prediction probabilities. The invention improves the accuracy of recommending products for the user.
Description
Technical Field
The invention relates to the technical field of data processing, in particular to a product recommendation method, a product recommendation device, product recommendation equipment and a computer readable storage medium based on an e-commerce platform.
Background
The electronic commerce refers to the commerce activity which takes the information network technology as a means and takes commodity exchange as a center; the method can also be understood as the transaction activities and related service activities performed in an electronic transaction mode on the Internet, an intranet and a value-added network, and is electronization, networking and informatization of each link of the traditional commercial activities; the business behaviors using the internet as a medium all belong to the category of electronic commerce. Electronic commerce generally refers to a novel business operation mode in which, in wide commercial and trade activities worldwide, in an internet environment open on the internet, buyers and sellers conduct various commercial and trade activities without conspiracy based on a client/server application mode, and consumer online shopping, online transactions and online electronic payments among merchants, and various commercial activities, transaction activities, financial activities, and related comprehensive service activities are realized.
Moreover, with the continuous development of electronic commerce, the scale of online shopping users is in a growing situation all the time, so that the recommendation technology in a shopping website is applied, and the brought effect is more and more obvious. For the exploration of the personalized recommendation method, each e-commerce platform is continuously increased in investment and deeply explored, and interested e-commerce is recommended for users to promote the formation. However, most of the existing recommendation methods recommend users based on browsing, purchasing and collecting behaviors of the users, and the recommendation effect is poor. Therefore, how to improve the accuracy of recommending products for users becomes a problem which needs to be solved urgently at present.
Disclosure of Invention
The invention mainly aims to provide a product recommendation method, a product recommendation device, product recommendation equipment and a computer readable storage medium based on an e-commerce platform, and aims to solve the technical problem of improving the accuracy of recommending products for users.
In order to achieve the purpose, the invention provides a product recommendation method based on an e-commerce platform, which comprises the following steps of:
identifying the user type according to the image information of the user to obtain a target user type, and determining a product recommendation range corresponding to the target user type;
acquiring user communication data of a current user account, determining all friend accounts in the user communication data, and taking the current user account and the friend accounts as target accounts;
traversing each target account number, collecting browsing information of the traversed target account number in a preset e-commerce platform, and constructing recommended product content according to the browsing information and the product recommendation range;
inputting the recommended product content into a preset recurrent neural network model for model training to obtain a prediction probability vector;
and sequentially fusing the prediction probability vectors corresponding to the target account numbers to obtain all prediction probabilities, and recommending the product corresponding to the maximum prediction probability in the prediction probabilities.
Optionally, the step of sequentially fusing the prediction probability vectors corresponding to the target account numbers to obtain all prediction probabilities includes:
taking the prediction probability vector corresponding to the current user account as a first probability vector, and taking the prediction probability vector corresponding to each friend account as a second probability vector;
determining a first weight occupied by the current user account in each target account, and determining a second weight occupied by the friend account in each target account;
calculating a first product of the first probability vector and the first weight, and calculating a second product between the second weight and the second probability vector, and taking a sum of the first product and the second product as a prediction probability.
Optionally, the step of collecting browsing information of the traversed target account in a preset e-commerce platform and constructing recommended product content according to the browsing information and the product recommendation range includes:
detecting whether the traversed target account number has browsing information in a preset e-commerce platform;
and if the browsing information exists, determining all product contents corresponding to the browsing information, and taking the product contents which accord with the product recommendation range in the product contents as recommended product contents.
Optionally, the step of determining all product contents corresponding to the browsing information includes:
extracting product words and emotional attribute information of the product words in the browsing information according to the character pattern attributes in the browsing information;
mining in a preset emotion knowledge graph according to the emotion attribute information to obtain emotion keywords;
searching all shops in the E-commerce platform according to the emotion keywords and the product words to obtain target shops;
and determining all product contents corresponding to the browsing information according to the product contents in the target shop.
Optionally, the step of searching all stores in the e-commerce platform according to the emotion keywords and the product words to obtain target stores includes:
determining the product type corresponding to the product word, searching all shops in the e-commerce platform according to the product type to obtain search shops, determining matched shops matched with the emotion keywords in the search shops, and taking the matched shops as target shops.
Optionally, the step of determining all product contents corresponding to the browsing information according to the product contents in the target store includes:
determining image video information corresponding to product content in the target shop, and extracting sound signals and image signals which belong to the same time step in the image video information;
carrying out yellow-wading storm detection on the sound signal and the image signal simultaneously;
and if the fact that the sound signal and the image signal do not relate to yellow storm is detected, taking the product content in the target shop as the product content corresponding to the browsing information.
Optionally, the step of identifying the user type according to the image information of the user includes:
acquiring image information with a user body image, and carrying out human body shape recognition on the image information;
and taking the type corresponding to the recognition result of the human body form recognition as the user type.
In addition, in order to achieve the above object, the present invention further provides a product recommendation device based on an e-commerce platform, including:
the determining module is used for identifying the user type according to the image information of the user to obtain a target user type and determining a product recommendation range corresponding to the target user type;
the system comprises an acquisition module, a processing module and a display module, wherein the acquisition module is used for acquiring user communication data of a current user account, determining all friend accounts in the user communication data, and taking the current user account and the friend accounts as target accounts;
the construction module is used for traversing each target account, collecting browsing information of the traversed target account in a preset e-commerce platform, and constructing recommended product content according to the browsing information and the product recommendation range;
the input module is used for inputting the recommended product content into a preset recurrent neural network model for model training to obtain a prediction probability vector;
and the recommending module is used for sequentially fusing the prediction probability vectors corresponding to the target account numbers to obtain all prediction probabilities and recommending products corresponding to the maximum prediction probability in the prediction probabilities.
In addition, in order to achieve the above object, the present invention further provides a product recommendation device based on an e-commerce platform, where the product recommendation device based on the e-commerce platform includes a memory, a processor, and a product recommendation program based on the e-commerce platform, which is stored in the memory and can be run on the processor, and when the product recommendation program based on the e-commerce platform is executed by the processor, the steps of the product recommendation method based on the e-commerce platform as described above are implemented.
In addition, in order to achieve the above object, the present invention further provides a storage medium, in which a product recommendation program based on an e-commerce platform is stored, and when being executed by a processor, the product recommendation program based on the e-commerce platform implements the steps of the product recommendation method based on the e-commerce platform.
The invention identifies the user type according to the image information of the user to determine the product recommendation range, thereby roughly deducing the products which are probably bought by the user from the macro, then according to the user communication data of the current user account, the friend account and the current user account are taken as the target accounts together, the recommended product content is constructed according to the browsing information of each target account in the e-commerce platform and the product recommendation range, then the recommended product content is input into a model for training to obtain the prediction probability vector, each prediction probability vector is fused to obtain the prediction probability, the product corresponding to the maximum probability is recommended, thereby further determining the recommended product content according to the browsing habits of friends of the user and the user per se, and the product corresponding to the maximum prediction probability is recommended, thereby ensuring that the e-commerce platform recommends the products to the user more accurately, the accuracy of recommending products for the user is improved.
Drawings
Fig. 1 is a schematic terminal structure diagram of a hardware operating environment according to an embodiment of the present invention;
fig. 2 is a schematic flowchart of a first embodiment of the product recommendation method based on an e-commerce platform according to the present invention;
FIG. 3 is a schematic diagram of the device units of the product recommendation device based on the E-commerce platform according to the present invention;
fig. 4 is a schematic diagram illustrating determination of a target account in the product recommendation method based on the e-commerce platform according to the present invention;
fig. 5 is a flowchart illustrating a second embodiment of the product recommendation method based on the e-commerce platform according to the present invention.
The objects, features and advantages of the present invention will be further explained with reference to the accompanying drawings.
Detailed Description
It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
As shown in fig. 1, fig. 1 is a schematic terminal structure diagram of a hardware operating environment according to an embodiment of the present invention.
The terminal of the embodiment of the invention is a product recommendation device based on an e-commerce platform.
As shown in fig. 1, the terminal may include: a processor 1001, such as a CPU, a network interface 1004, a user interface 1003, a memory 1005, a communication bus 1002. Wherein a communication bus 1002 is used to enable connective communication between these components. The user interface 1003 may include a Display screen (Display), an input unit such as a Keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface, a wireless interface. The network interface 1004 may optionally include a standard wired interface, a wireless interface (e.g., WI-FI interface). The memory 1005 may be a high-speed RAM memory or a non-volatile memory (e.g., a magnetic disk memory). The memory 1005 may alternatively be a storage device separate from the processor 1001.
Optionally, the terminal may further include a camera, a Radio Frequency (RF) circuit, a sensor, an audio circuit, a WiFi module, and the like. Such as light sensors, motion sensors, and other sensors, among others. Specifically, the light sensor may include an ambient light sensor that adjusts the brightness of the display screen according to the brightness of ambient light, and a proximity sensor that turns off the display screen and/or the backlight when the terminal device is moved to the ear. Of course, the terminal device may also be configured with other sensors such as a gyroscope, a barometer, a hygrometer, a thermometer, and an infrared sensor, which are not described herein again.
Those skilled in the art will appreciate that the terminal structure shown in fig. 1 is not intended to be limiting and may include more or fewer components than those shown, or some components may be combined, or a different arrangement of components.
As shown in fig. 1, a memory 1005, which is a kind of computer storage medium, may include therein an operating system, a network communication module, a user interface module, and an e-commerce platform-based product recommendation program.
In the terminal shown in fig. 1, the network interface 1004 is mainly used for connecting to a backend server and performing data communication with the backend server; the user interface 1003 is mainly used for connecting a client (user side) and performing data communication with the client; and the processor 1001 may be configured to invoke the e-commerce platform based product recommendation program stored in the memory 1005 and perform the following operations:
referring to fig. 2, the present invention provides a product recommendation method based on an e-commerce platform, and in a first embodiment of the product recommendation method based on the e-commerce platform, the product recommendation method based on the e-commerce platform includes the following steps:
step S10, identifying the user type according to the image information of the user to obtain the target user type, and determining the product recommendation range corresponding to the target user type;
due to various defects of the existing recommendation method, the accuracy of recommending products for users by the e-commerce platform is low. Therefore, in the embodiment, the mobile terminal or the display terminal and the e-commerce platform are included. The mobile terminal is used for registering and logging in the e-commerce platform after a user inputs personal information and sending the personal information to the e-commerce platform, and after the user logs in the e-commerce platform, the user can select commodities and then carry out transaction payment; the personal information comprises a name, a mobile phone number authenticated by a real name, user image information (including head portrait information, human body shape and the like), verification data and the like, and the verification data comprises preset living body action and secret protection problems and has a time-efficient check code. And the e-commerce platform can keep records of inquiry, purchase and the like of each user when logging in the e-commerce platform in real time. And then, when the user logs in the e-commerce platform, the commodity which the user is interested in is directly recommended to the user.
Therefore, in this embodiment, after the user logs in the e-commerce platform through the mobile terminal or the display terminal, the image information of the user is actively acquired through the camera function of the mobile terminal or the display terminal. Wherein the image information includes at least half-body photograph information, whole-body photograph information, or head photograph information of the user. And then, the e-commerce platform identifies the acquired image information, identifies the user type of the image information, such as determining the gender, age range and the like of the user, and determines a corresponding product recommendation range according to the identified target user type. Namely, a preset type product recommendation table is obtained, and a product recommendation range corresponding to the target user type in the product recommendation table is determined. For example, if the image information is recognized and the user is found to be a male person and the age range is an elderly person, it is possible to specify that the product recommendation range is an elderly product and a family product, and delete the product like a female product. That is, in the present embodiment, by identifying the image information, the recommended range of products that the user may purchase in a rough type is first screened out.
Step S20, obtaining user communication data of a current user account, determining all friend accounts in the user communication data, and taking the current user account and the friend accounts as target accounts;
in addition, because the user logs in the e-commerce platform for the first time or the number of times of logging in the e-commerce platform is too small, the product recommendation range is determined only through the image information, and the recommendation accuracy is low. At this time, the user communication data of the current user account may be acquired. The current user account is account data logged in by the user through the mobile terminal or the display terminal. The user communication data may be user communication detail data. The user communication detailed list data comprises information of friends in a user mobile phone address list, friends in an e-commerce platform, WeChat friends and the like. And constructing a user communication contact circle based on the user communication detail list data, calculating a user-user intimacy index through a communication contact intimacy index algorithm, finding a first-degree direct friend of the user based on intimacy, and mining a second-degree potential friend of the user by utilizing a FASTUNFLODING algorithm to form a friend community of the user. And taking all user accounts in the user friend community as friend accounts. For convenience of subsequent explanation, in this embodiment, the current user account and the friend account are used as the target account, that is, the target account may be the current user account or the friend account. For example, as shown in fig. 4, if the current user account is 1, and it is found through query that the friend accounts of the current user account are 2 and 3, and the friend accounts of the friend account 2 are 1, 4 and 5; the friend accounts 3 have friend accounts 1, 6, 7, and 8. Thus, accounts 1-8 can be considered together as the target account.
Step S30, traversing each target account, collecting browsing information of the traversed target account in a preset e-commerce platform, and constructing recommended product content according to the browsing information and the product recommendation range;
after all the target account numbers are acquired, the same operation mode can be adopted for each target account number, that is, each target account number can be traversed, whether the traversed target account number has a browsing record in the e-commerce platform in advance is determined, if not, the next target account number is actively jumped to, and the same detection process is carried out. However, if the traversed target account has a browsing record in the e-commerce platform, browsing information corresponding to the browsing record is acquired, wherein the browsing information includes image and text information of a product browsed by the target account, input comment information and the like. After all target account numbers are traversed, browsing information related to the product recommendation range, namely related browsing information, is determined in each browsing information, products corresponding to the related browsing information are used as recommended products, and the contents of the recommended products are used as the contents of the recommended products.
In another scenario, after the recommended product content is acquired, each target account may be screened according to the recommended product content to obtain a screened target account. And screening browsing information associated with the recommended product content in the target account. And then, sequencing the screening target account according to browsing information of the screening target account and the sequence of browsing time nodes to form short-term preference content information, and calculating the content grading of the user according to the behavior record (namely behavior content) of the screening target account for a longer period of time by integrating the behavior type, the behavior times and the content on-shelf time of the user to generate an N x M user content grading matrix R, wherein N represents the number of the users, M represents the number of the contents, and the content grading matrix R is used as the long-term preference information of the user. The short-term and long-term content preference information is also formed for each user in the user's friend community in the manner described above.
Generating long-term candidate recommended contents (namely long-term candidate recommended contents) of the user by combining the scoring matrix and utilizing a collaborative filtering recommendation method based on matrix decomposition, and then fusing the long-term candidate recommended contents and the short-term content preference information to form a final recommendation result, namely inputting the long-term candidate recommended contents and the short-term preference content information into a preset recurrent neural network model of an LSTM recurrent neural network structure for model training, and determining a product to be recommended based on the training result.
Step S40, inputting the recommended product content into a preset recurrent neural network model for model training to obtain a prediction probability vector;
in this embodiment, after the recommended product content is obtained, the recommended product content may be directly input into a recurrent neural network model set in advance for model training. The recurrent neural network model may be an LSTM (Long Short-Term Memory) model. However, when the long-term candidate recommended content and the short-term preference content information are determined according to the recommended product content, the long-term candidate recommended content and the short-term preference content information can be input into an LSTM recurrent neural network structure (namely, a preset recurrent neural network model) capable of fusing the interest preferences of the long-term candidate recommended content and the short-term content of the user, and recommendation results fusing the long-term interest and the short-term interest are respectively generated for the user and friends of the user. The improved LSTM structure for fusing long and short interests comprises an input layer, a hidden layer and an output layer. Wherein the input layer comprises a long-term candidate recommended content L _ TOPK { Ui } of the user Ui and a user Short-term content interest preference sequence Short _ Favor _ seq (Ui), and the long-term candidate recommended content and the Short-term content preference sequence are input as hidden layers at each time t; the hidden layer is composed of long and short time interesting memory cells, and the cell state at the time t is the cell state at the time t-1 (namely h) t-1 ) Time t long term interest information (i.e. x) t ) Time t short term interest information (i.e. |) t ) Jointly, the expression is as shown in formula 4. The output layer outputs an M-dimensional user-content probability vector according to the state of the hidden layer cells through a single-layer neural network using a SOFTMAX activation function
，
And the prediction preference probability of the user to the M contents at the time t is represented, and the vector value is between (0 and 1) and is calculated through a formula 5. Take the output of the last time T
As a user-content prediction preference probability vector (i.e., prediction probability vector).
and, for friends Uj (namely all other target account numbers) in the friend community f (Ui) of the user Ui, the prediction preference probability vector of the user Ui for the content is generated by fusing the long-term interest and the short-term interest with the communication social relationship recommendation according to the above method. And using the prediction preference probability vector as a prediction probability vector.
And step S50, sequentially fusing the prediction probability vectors corresponding to the target account numbers to obtain all prediction probabilities, and recommending products corresponding to the maximum prediction probability in the prediction probabilities.
In this embodiment, after each prediction probability vector is obtained through calculation, fusion may be performed according to target accounts of different users to obtain all prediction probabilities, that is, each prediction probability may be obtained through calculation according to the following formula
. Namely:
wherein
、
User-content prediction preference vectors produced for the users ui and uj based on the fused long-term interests are recommended,
and the intimacy index of the interaction circle of the user ui and the user uj is obtained.
Giving the number m of final recommendation results, and taking
And taking the top M products with the highest prediction probability as the content recommendation result of the user. Namely recommending the product corresponding to the maximum prediction probability in the prediction probabilities.
In the embodiment, the product recommendation range is determined by identifying the user type according to the image information of the user, so that products which are likely to be purchased by the user can be roughly inferred from the macro, then the friend account and the current user account are taken as target accounts together according to the user communication data of the current user account, recommended product contents are constructed according to the browsing information of each target account in the e-commerce platform and the product recommendation range, then the recommended product contents are input into a model for training to obtain a prediction probability vector, each prediction probability vector is fused to obtain a prediction probability, and a product corresponding to the maximum probability is recommended, so that the recommended product contents can be further determined according to the habits of the user and the browsing of the user per se, and the product corresponding to the maximum prediction probability is recommended, so that the products recommended to the user by the e-commerce platform are more accurate, the accuracy of recommending products for the user is improved.
Further, based on the first embodiment of the present invention, a second embodiment of the product recommendation method based on an e-commerce platform is provided, as shown in fig. 5, in this embodiment, step S50 of the above embodiment sequentially fuses prediction probability vectors corresponding to each target account number, and refines the step of obtaining all prediction probabilities, where the step includes:
step a, taking the prediction probability vector corresponding to the current user account as a first probability vector, and taking the prediction probability vector corresponding to each friend account as a second probability vector;
in this embodiment, after obtaining the prediction probability vectors corresponding to all target accounts, in order to calculate the prediction probabilities of all behavior contents, the prediction probability vector corresponding to the current user account may be used as a first probability vector, and the prediction probability vector corresponding to each friend account may be used as a second probability vector.
B, determining a first weight occupied by the current user account in each target account, and determining a second weight occupied by the friend account in each target account;
in this embodiment, a communication social network is constructed according to each target account, a current user account is a first-level node of the communication social network, an associated user account associated with the current user account is a second-level node, a user account associated with the associated user account is a third-level node, and different weights are assigned to nodes of different levels, for example, the weight of the first-level node is 60%, the weight of the second-level node is 30% (that is, the sum of the weights of all nodes in the second-level node is 10%), and the weight of the third-level node is 10% (that is, the sum of the weights of all nodes in the third-level node is 10%). Therefore, when determining the weight occupied by the current user account and the friend account in each target account, the determination can be performed according to the communication social network. The friend account comprises a related user account and an account related to the related user account.
And c, calculating a first product of the first probability vector and the first weight, calculating a second product of the second weight and the second probability vector, and taking the sum of the first product and the second product as a prediction probability.
In calculating the prediction probability for a certain recommended product content, a first product of the first probability vector and the first weight may be calculated. And calculating a second product between the second weight and the second probability vector, and when the second product is calculated, calculating each friend account number in the same way, namely the number of the second products is multiple, then adding all the second products and the first products, and taking the obtained sum as the prediction probability of the recommended product content. In this embodiment, the calculation may be performed in the same manner for all recommended product contents to obtain the prediction probability of each recommended product content.
In the embodiment, the accuracy and the effectiveness of the calculated prediction probability are guaranteed by calculating the first product between the first probability vector corresponding to the current account and the first weight, calculating the second product between the second probability vector corresponding to the friend account and the second weight, and taking the sum of the first product and the second product as the prediction probability.
Further, the method comprises the steps of collecting browsing information of the traversed target account in a preset e-commerce platform, and constructing recommended product content according to the browsing information and the product recommendation range, and comprises the following steps:
d, detecting whether the traversed target account number has browsing information in a preset e-commerce platform;
and e, if the browsing information exists, determining all product contents corresponding to the browsing information, and taking the product contents which accord with the product recommendation range in each product content as recommended product contents.
In this embodiment, after the target account is determined, the target account may be traversed, and then whether the target account of the variable has browsing information before the current time node of the e-commerce platform (such as the airbus, the naoba, etc.) is detected, that is, whether the target account uses the e-commerce platform is detected. And traversing the next target account when the browsing information does not exist. When the browsing information exists, the content of the browsing information needs to be extracted to obtain all product contents corresponding to the browsing information, determine which product contents are product contents conforming to the product recommendation range in each product content, and use the product contents conforming to the product recommendation range as recommended product contents.
In this embodiment, when the traversed target account has browsing information in the e-commerce platform, the product content corresponding to the product recommendation range in the product content corresponding to the browsing information is used as the recommended product content, so that the accuracy and effectiveness of the acquired recommended product content are guaranteed.
Specifically, the step of determining all product contents corresponding to the browsing information includes:
step f, extracting product words and emotional attribute information of the product words in the browsing information according to the character pattern attributes in the browsing information;
step g, mining in a preset emotion knowledge graph according to the emotion attribute information to obtain emotion keywords;
h, searching all shops in the E-commerce platform according to the emotion keywords and the product words to obtain target shops;
and i, determining all product contents corresponding to the browsing information according to the product contents in the target shop.
In this embodiment, when determining the product content corresponding to the browsing information, all web pages corresponding to the browsing information may be determined first, and the text and pattern attributes in the web pages are determined to perform entity extraction, and the extracted entities are used as product words. For example, if a pattern of a cloth shoe is displayed in the web page, the product word includes the cloth shoe. If the characters of the badminton under the brand A are displayed in the webpage, the product words comprise the badminton. And the emotional attribute information of the product word can comprise name, origin, using population, style, material, shape, style, pattern and the like. For example, the entity is shoes, and the emotional attribute information comprises cowhide, synthetic leather, mesh cloth, polyurethane and the like.
And after the product word and the corresponding emotion attribute information are obtained, mining can be performed in an emotion knowledge map set in advance to obtain more emotion keywords associated with the product word, for example, if the product word is a leather shoe and the emotion attribute information is C and B, the emotion keywords can be x brand leather shoes with emotion attribute information C and B. The emotion knowledge map is provided with a plurality of product words, emotion attribute information and emotion keywords corresponding to the product words. Including for example, a cloth shoe-canvas-old beijing cloth shoe.
After the emotion keywords are determined, searching can be carried out according to the emotion keywords and the product words, namely, first shops selling products corresponding to the product words are screened out from the e-commerce platform, and then shops matched with the emotion keywords are determined in the first shops and serve as target shops. And then taking the product contents of all the products in the target shop as the product contents corresponding to the browsing information.
In the embodiment, the product words and the emotion attribute information in the browsing information are extracted and mined in the emotion knowledge map to obtain the emotion keywords, searching is performed according to the emotion keywords and the product words to obtain the target store, and all product contents corresponding to the browsing information are determined according to the product contents of the target store, so that the accuracy and the effectiveness of determining the product contents corresponding to the browsing information are guaranteed.
Specifically, the step of searching all shops in the e-commerce platform according to the emotion keywords and the product words to obtain target shops comprises the following steps:
step j, determining the product type corresponding to the product word, searching all shops in the E-commerce platform according to the product type to obtain search shops, determining matched shops matched with the emotion keywords in the search shops, and taking the matched shops as target shops.
In this embodiment, when searching for a target store, the product type corresponding to the product word, such as shoes, balls, books, etc., may be determined. And then searching all shops in the e-commerce platform, determining which shops sell products with the product types consistent with the product types corresponding to the product words, and if so, taking the shop as a searching shop. And then, carrying out secondary search on each search shop, namely determining whether products matched with each emotion keyword exist in each product sold in the search shop, and if so, taking the search shop where the search shop is located as a target shop.
In the embodiment, the product type is determined, the search shop is determined according to the product type, and the target shop is determined according to the search shop and the emotion keywords, so that the accuracy and the effectiveness of the obtained target shop are guaranteed.
Further, the step of determining all product contents corresponding to the browsing information according to the product contents in the targeted shop includes:
step l, determining image video information corresponding to product content in the target shop, and extracting sound signals and image signals which belong to the same time step in the image video information;
step m, carrying out yellow-related storm detection on the sound signals and the image signals simultaneously;
and n, if the fact that the sound signals and the image signals do not relate to yellow storm is detected, taking the product content in the target shop as the product content corresponding to the browsing information.
In this embodiment, when determining the product content corresponding to the browsing information, it is necessary to first determine all the image video information corresponding to the product content in the target store. The image video information is a propaganda video or an image uploaded by a merchant of the target store aiming at a product of the merchant. And then, extracting the sound signals and the image signals from the propaganda videos in the image video information, namely extracting the sound signals and the image signals at the same time step, and respectively carrying out yellow-related explosion detection on the sound signals and the image signals. The detection method may be to convert the sound signal into a two-dimensional image, such as an MFCC (Mel-Frequency Cepstral Coefficients, Mel Frequency cepstrum) spectrogram, then perform a detection analysis on the two-dimensional image related to a yellow-related explosion, and perform a detection on the image signal related to a yellow-related explosion in a conventional manner, which is not described herein, and the image video information corresponding to the product content in the target store is not related to a yellow-related explosion, so that the product content in the target store is used as the product content corresponding to the browsing information.
In the embodiment, the yellow-related explosion detection is carried out on the product content in the target store, so that the product content subsequently recommended to the user is guaranteed to be healthy, normal and effective.
Further, the step of identifying the user type according to the image information of the user includes:
step x, acquiring image information with a user body image, and carrying out human body shape recognition on the image information;
and y, taking the type corresponding to the recognition result of the human body form recognition as the user type.
In this embodiment, when acquiring the image information of the user, it is necessary to acquire image information having a user body image including a user head portrait, a user body portrait, a user whole body portrait, and the like. And then, carrying out human body shape recognition on the image information, wherein the human body shape recognition comprises the steps of recognizing the gender of the user, recognizing the age group (such as the old, children and adults) to which the user belongs, recognizing whether the user is a disabled person or not and the like. And the recognition result of the human body form recognition is used as the user type, for example, the user is male, the old people and the disabled people.
In the embodiment, the human body form is identified for the image information, and the user type is determined according to the identification result, so that the accuracy and the effectiveness of the acquired user type are guaranteed.
In addition, referring to fig. 3, an embodiment of the present invention further provides a product recommendation device based on an e-commerce platform, including:
the determining module A10 is used for carrying out user type identification according to the image information of the user to obtain a target user type and determining a product recommendation range corresponding to the target user type;
an obtaining module a20, configured to obtain user communication data of a current user account, determine all friend accounts in the user communication data, and use the current user account and the friend accounts as target accounts;
the construction module A30 is used for traversing each target account, collecting browsing information of the traversed target account in a preset e-commerce platform, and constructing recommended product content according to the browsing information and the product recommendation range;
the input module A40 is used for inputting the recommended product content into a preset recurrent neural network model for model training to obtain a prediction probability vector;
and the recommending module A50 is used for sequentially fusing the prediction probability vectors corresponding to the target account numbers to obtain all the prediction probabilities and recommending the products corresponding to the maximum prediction probability in the prediction probabilities.
Optionally, the recommending module a50 is configured to:
taking the prediction probability vector corresponding to the current user account as a first probability vector, and taking the prediction probability vector corresponding to each friend account as a second probability vector;
determining a first weight occupied by the current user account in each target account, and determining a second weight occupied by the friend account in each target account;
calculating a first product of the first probability vector and the first weight, and calculating a second product between the second weight and the second probability vector, and taking a sum between the first product and the second product as a prediction probability.
Optionally, a module a30 is constructed for:
detecting whether the traversed target account number has browsing information in a preset e-commerce platform;
and if the browsing information exists, determining all product contents corresponding to the browsing information, and taking the product contents which accord with the product recommendation range in the product contents as recommended product contents.
Optionally, a module a30 is constructed for:
extracting product words and emotion attribute information of the product words in the browsing information according to the character pattern attributes in the browsing information;
mining in a preset emotion knowledge graph according to the emotion attribute information to obtain emotion keywords;
searching all shops in the E-commerce platform according to the emotion keywords and the product words to obtain target shops;
and determining all product contents corresponding to the browsing information according to the product contents in the target shop.
Optionally, a module a30 is constructed for:
determining the product type corresponding to the product word, searching all shops in the e-commerce platform according to the product type to obtain search shops, determining matched shops matched with the emotion keywords in the search shops, and taking the matched shops as target shops.
Optionally, a module a30 is constructed for:
determining image video information corresponding to product content in the target shop, and extracting sound signals and image signals which belong to the same time step in the image video information;
carrying out yellow-related storm detection on the sound signals and the image signals simultaneously;
and if the fact that the sound signal and the image signal do not relate to yellow storm is detected, taking the product content in the target shop as the product content corresponding to the browsing information.
Optionally, determining a module a10, configured to:
acquiring image information with a user body image, and carrying out human body shape recognition on the image information;
and taking the type corresponding to the recognition result of the human body form recognition as the user type.
The steps implemented by the functional modules of the product recommendation device based on the e-commerce platform can refer to the embodiments of the product recommendation method based on the e-commerce platform, and are not described herein again.
In addition, the invention also provides a product recommendation device based on the e-commerce platform, which comprises: a memory, a processor, and an e-commerce platform based product recommendation program stored on the memory; the processor is used for executing the product recommendation program based on the e-commerce platform so as to realize the steps of the product recommendation method based on the e-commerce platform.
The present invention also provides a storage medium, which may be a computer-readable storage medium, having one or more programs stored thereon, where the one or more programs are further executable by one or more processors for implementing the steps of the embodiments of the product recommendation method based on e-commerce platform.
The specific implementation manner of the computer-readable storage medium of the present invention is substantially the same as that of each embodiment of the product recommendation method based on the e-commerce platform, and is not described herein again.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or system. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or system that comprises the element.
The above-mentioned serial numbers of the embodiments of the present invention are merely for description and do not represent the merits of the embodiments.
Through the above description of the embodiments, those skilled in the art will clearly understand that the method of the above embodiments can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but in many cases, the former is a better implementation manner. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium (e.g., ROM/RAM, magnetic disk, optical disk) as described above and includes instructions for enabling a terminal device (e.g., a mobile phone, a computer, a server, an air conditioner, or a network device) to execute the method according to the embodiments of the present invention.
The above description is only a preferred embodiment of the present invention, and not intended to limit the scope of the present invention, and all modifications of equivalent structures and equivalent processes, which are made by using the contents of the present specification and the accompanying drawings, or directly or indirectly applied to other related technical fields, are included in the scope of the present invention.
Claims (10)
1. A product recommendation method based on an E-commerce platform is characterized by comprising the following steps:
identifying the user type according to the image information of the user to obtain a target user type, and determining a product recommendation range corresponding to the target user type;
acquiring user communication data of a current user account, determining all friend accounts in the user communication data, and taking the current user account and the friend accounts as target accounts;
traversing each target account, collecting browsing information of the traversed target account in a preset e-commerce platform, and constructing recommended product content according to the browsing information and the product recommendation range;
inputting the recommended product content into a preset recurrent neural network model for model training to obtain a prediction probability vector;
and sequentially fusing the prediction probability vectors corresponding to the target account numbers to obtain all prediction probabilities, and recommending the product corresponding to the maximum prediction probability in the prediction probabilities.
2. The e-commerce platform-based product recommendation method of claim 1, wherein the step of sequentially fusing the prediction probability vectors corresponding to the target account numbers to obtain all prediction probabilities comprises:
taking the prediction probability vector corresponding to the current user account as a first probability vector, and taking the prediction probability vector corresponding to each friend account as a second probability vector;
determining a first weight occupied by the current user account in each target account, and determining a second weight occupied by the friend account in each target account;
calculating a first product of the first probability vector and the first weight, and calculating a second product between the second weight and the second probability vector, and taking a sum of the first product and the second product as a prediction probability.
3. The e-commerce platform-based product recommendation method of claim 1, wherein the step of collecting browsing information of traversed target account numbers in a preset e-commerce platform and constructing recommended product contents according to the browsing information and the product recommendation range comprises:
detecting whether the traversed target account number has browsing information in a preset e-commerce platform;
and if the browsing information exists, determining all product contents corresponding to the browsing information, and taking the product contents which accord with the product recommendation range in the product contents as recommended product contents.
4. The e-commerce platform-based product recommendation method of claim 3, wherein the step of determining all product contents corresponding to the browsing information comprises:
extracting product words and emotional attribute information of the product words in the browsing information according to the character pattern attributes in the browsing information;
mining in a preset emotion knowledge graph according to the emotion attribute information to obtain emotion keywords;
searching all shops in the E-commerce platform according to the emotion keywords and the product words to obtain target shops;
and determining all product contents corresponding to the browsing information according to the product contents in the target shop.
5. The e-commerce platform-based product recommendation method of claim 4, wherein the step of searching all shops in the e-commerce platform according to the emotion keyword and the product word to obtain a target shop comprises:
determining the product type corresponding to the product word, searching all shops in the e-commerce platform according to the product type to obtain search shops, determining matched shops matched with the emotion keywords in the search shops, and taking the matched shops as target shops.
6. The e-commerce platform-based product recommendation method of claim 4, wherein the step of determining all product contents corresponding to the browsing information according to the product contents in the target store comprises:
determining image video information corresponding to product content in the target shop, and extracting sound signals and image signals which belong to the same time step in the image video information;
carrying out yellow-wading storm detection on the sound signal and the image signal simultaneously;
and if the fact that the sound signal and the image signal do not relate to yellow storm is detected, taking the product content in the target shop as the product content corresponding to the browsing information.
7. The e-commerce platform based product recommendation method of any one of claims 1-6, wherein the step of performing user type identification based on the image information of the user comprises:
acquiring image information with a user body image, and carrying out human body shape recognition on the image information;
and taking the type corresponding to the recognition result of the human body form recognition as the user type.
8. An e-commerce platform-based product recommendation device, comprising:
the determining module is used for identifying the user type according to the image information of the user to obtain a target user type and determining a product recommendation range corresponding to the target user type;
the system comprises an acquisition module, a processing module and a display module, wherein the acquisition module is used for acquiring user communication data of a current user account, determining all friend accounts in the user communication data, and taking the current user account and the friend accounts as target accounts;
the construction module is used for traversing each target account, collecting browsing information of the traversed target account in a preset e-commerce platform, and constructing recommended product content according to the browsing information and the product recommendation range;
the input module is used for inputting the recommended product content into a preset recurrent neural network model for model training to obtain a prediction probability vector;
and the recommending module is used for sequentially fusing the prediction probability vectors corresponding to the target account numbers to obtain all prediction probabilities and recommending products corresponding to the maximum prediction probability in the prediction probabilities.
9. An e-commerce platform-based product recommendation device, comprising: a memory, a processor, and an e-commerce platform based product recommendation program stored on the memory and executable on the processor, the e-commerce platform based product recommendation program when executed by the processor implementing the steps of the e-commerce platform based product recommendation method as recited in any one of claims 1 to 7.
10. A storage medium, wherein the storage medium stores an e-commerce platform based product recommendation program, and the e-commerce platform based product recommendation program, when executed by a processor, implements the steps of the e-commerce platform based product recommendation method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211051024.0A CN115131111A (en) | 2022-08-31 | 2022-08-31 | Product recommendation method, device, equipment and storage medium based on e-commerce platform |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211051024.0A CN115131111A (en) | 2022-08-31 | 2022-08-31 | Product recommendation method, device, equipment and storage medium based on e-commerce platform |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115131111A true CN115131111A (en) | 2022-09-30 |
Family
ID=83388116
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211051024.0A Pending CN115131111A (en) | 2022-08-31 | 2022-08-31 | Product recommendation method, device, equipment and storage medium based on e-commerce platform |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115131111A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107292650A (en) * | 2016-04-05 | 2017-10-24 | 长沙海商网络技术有限公司 | A kind of full platform network distribution processing system software |
| CN108596646A (en) * | 2018-03-15 | 2018-09-28 | 南京航空航天大学 | A kind of garment coordination recommendation method of fusion face character analysis |
| CN110458638A (en) * | 2019-06-26 | 2019-11-15 | 平安科技（深圳）有限公司 | A kind of Method of Commodity Recommendation and device |
| CN113486250A (en) * | 2021-07-28 | 2021-10-08 | 中移（杭州）信息技术有限公司 | Content recommendation method, device, equipment and computer readable storage medium |
-
2022
- 2022-08-31 CN CN202211051024.0A patent/CN115131111A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107292650A (en) * | 2016-04-05 | 2017-10-24 | 长沙海商网络技术有限公司 | A kind of full platform network distribution processing system software |
| CN108596646A (en) * | 2018-03-15 | 2018-09-28 | 南京航空航天大学 | A kind of garment coordination recommendation method of fusion face character analysis |
| CN110458638A (en) * | 2019-06-26 | 2019-11-15 | 平安科技（深圳）有限公司 | A kind of Method of Commodity Recommendation and device |
| CN113486250A (en) * | 2021-07-28 | 2021-10-08 | 中移（杭州）信息技术有限公司 | Content recommendation method, device, equipment and computer readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11775844B2 (en) | Visual aspect localization presentation | |
| CN109509054B (en) | Commodity recommendation method under mass data, electronic device and storage medium | |
| JP6867479B2 (en) | Visual search based on image analysis and prediction | |
| KR101852581B1 (en) | Image evaluation | |
| CN111353091A (en) | Information processing method and device, electronic equipment and readable storage medium | |
| CN108230009B (en) | User preference prediction method and device and electronic equipment | |
| KR20160019445A (en) | Incorporating user usage of consumable content into recommendations | |
| CN108140212A (en) | For determining the system and method for nodes for research | |
| CN109584006B (en) | Cross-platform commodity matching method based on deep matching model | |
| KR102246109B1 (en) | Product recommendation system and method based on consumer preference | |
| KR102321987B1 (en) | Petition System based on Petition History and Petitioner Profile Analysis | |
| CN108305181B (en) | Social influence determination method and device, information delivery method and device, equipment and storage medium | |
| CN113191838B (en) | Shopping recommendation method and system based on heterogeneous graph neural network | |
| CN112100513A (en) | Knowledge graph-based recommendation method, device, equipment and computer readable medium | |
| CN113722583A (en) | Recommendation method, recommendation model training method and related products | |
| CN111292168B (en) | Data processing method, device and equipment | |
| CN113761347A (en) | Commodity recommendation method, commodity recommendation device, storage medium and commodity recommendation system | |
| Wiradinata et al. | Online Measuring Feature for Batik Size Prediction using Mobile Device: A Potential Application for a Novelty Technology | |
| Tsyrelchuk et al. | Optimization of business processes via Big Data | |
| KR101754124B1 (en) | The restaurant recommending system and the recommending method thereof | |
| CN108712683B (en) | Data transmission method, bullet screen information generation method and device | |
| CN115131111A (en) | Product recommendation method, device, equipment and storage medium based on e-commerce platform | |
| JP2002288216A (en) | Sensitivity item search system and method therefor | |
| KR20180092053A (en) | Intelligent shopping managing system | |
| CN112488355A (en) | Method and device for predicting user rating based on graph neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20220930 |
|
| RJ01 | Rejection of invention patent application after publication |