CN116542738A - Information pushing method based on electronic commerce big data - Google Patents
Information pushing method based on electronic commerce big data Download PDFInfo
- Publication number
- CN116542738A CN116542738A CN202310503507.8A CN202310503507A CN116542738A CN 116542738 A CN116542738 A CN 116542738A CN 202310503507 A CN202310503507 A CN 202310503507A CN 116542738 A CN116542738 A CN 116542738A
- Authority
- CN
- China
- Prior art keywords
- user
- interest
- commerce
- neural network
- model
- 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.)
- Withdrawn
Links
- 238000000034 method Methods 0.000 title claims abstract description 32
- 238000013528 artificial neural network Methods 0.000 claims abstract description 77
- 239000013598 vector Substances 0.000 claims description 44
- 230000002068 genetic effect Effects 0.000 claims description 19
- 238000012549 training Methods 0.000 claims description 18
- 238000004422 calculation algorithm Methods 0.000 claims description 13
- 238000012216 screening Methods 0.000 claims description 12
- 238000004364 calculation method Methods 0.000 claims description 11
- 210000002569 neuron Anatomy 0.000 claims description 8
- 239000011159 matrix material Substances 0.000 claims description 7
- 238000005065 mining Methods 0.000 claims description 5
- 238000010276 construction Methods 0.000 claims description 3
- 238000010353 genetic engineering Methods 0.000 claims description 3
- 230000035772 mutation Effects 0.000 claims description 3
- 230000011218 segmentation Effects 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 238000004140 cleaning Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 230000004927 fusion Effects 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000013515 script Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
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
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/086—Learning methods using evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Business, Economics & Management (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Molecular Biology (AREA)
- Evolutionary Biology (AREA)
- General Health & Medical Sciences (AREA)
- Finance (AREA)
- Computational Linguistics (AREA)
- Accounting & Taxation (AREA)
- General Business, Economics & Management (AREA)
- Development Economics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Economics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- Physiology (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses an information pushing method based on electronic commerce big data, which comprises the following steps: acquiring data information of a target user; constructing a target user interest model, and calculating to obtain a first interest set and a second interest set; based on the BP neural network, the first interest set and the second interest set are used as input nodes to obtain a final pushing result; and pushing the pushing result to a target user. According to the invention, the interest model of the target user is built based on various data of the e-commerce platform, so that the real interest preference of the target user can be mined, the target user can be pushed accurately, the requirement of the target user can be met according to the user requirement, the searching difficulty of searching favorite commodities by the user is reduced, the shopping experience of the user in using the e-commerce platform and the experience of the user are improved, the use efficiency of the e-commerce platform is improved, the operation cost of the e-commerce platform is reduced, and the commodity traffic is increased.
Description
Technical Field
The invention relates to the technical field of electronic commerce, in particular to an information pushing method based on electronic commerce big data.
Background
Electronic commerce is a special business model which is developed along with the development of informatization, and has become an indispensable part of human production and life. The electronic commerce promotes the human life style to be greatly changed, brings convenience to the production and life of people, and can buy the required goods. The electronic commerce mainly refers to commercial trade exchange activities performed in a network environment, and a buyer and a seller do not need to face each other, but perform a novel commercial operation mode of commodity exchange activities through a browser or a related server APP; electronic commerce has four main modes: electronic commerce between businesses and consumers, electronic commerce between businesses, electronic commerce between consumers and electronic commerce between online commerce and the internet, etc., have been developed differently in various fields, but their existence must depend on various devices and network technologies; electronic commerce today has included not only single-finger online shopping but also online transactions, web marketing, electronic trading markets, etc., where the support of the internet, extranet, email, databases, electronic catalogs, and mobile phones is not available in this series of activities.
At present, when a user needs to make shopping, the user usually searches for an article which the user wants to purchase by using a search engine of the electronic commerce platform, then selects an article which meets the user's desire to purchase from the searched result, but in the using process, the user can not accurately recommend the article which the user needs according to the shopping requirement of the user, so that the shopping difficulty of the user can be increased, and when the user cannot purchase the article which the user wants to purchase on the electronic commerce platform, the user can abandon the purchase wish or purchase the article by turning to other electronic commerce platforms, and the shopping experience of the user can be further reduced.
For the problems in the related art, no effective solution has been proposed at present.
Disclosure of Invention
Aiming at the problems in the related art, the invention provides an information pushing method based on electronic commerce big data, so as to overcome the technical problems existing in the prior related art.
For this purpose, the invention adopts the following specific technical scheme:
an information pushing method based on electronic commerce big data comprises the following steps:
s1, acquiring data information of a target user;
s2, constructing a target user interest model, and calculating to obtain a first interest set and a second interest set;
s3, based on the BP neural network, the first interest set and the second interest set are used as input nodes to obtain a final pushing result;
and S4, pushing the pushing result to a target user.
Further, the step of obtaining the data information of the target user includes the following steps:
s11, collecting target user data and e-commerce product data based on an e-commerce website;
s12, collecting behavior data of a target user based on a preset log system;
further, the step of constructing the target user interest model and calculating to obtain the first interest set and the second interest set includes the following steps:
s21, constructing a target user interest model based on an improved vector space model representation method;
s22, calculating the similarity between the target user and other users based on the cosine similarity;
s23, screening out the first m users with highest similarity, and constructing a user nearest neighbor set;
s24, screening all the e-commerce products purchased by the user based on the nearest neighbor set of the user, and calculating the interest level value of the user on the e-commerce products;
s25, selecting the first j E-commerce products with the highest interest level value to construct a first interest set;
s26, selecting an e-commerce product with a higher score of a target user;
s27, calculating the similarity between the E-commerce products;
s28, screening out the first k E-commerce products with highest similarity, and constructing a nearest neighbor set of the E-commerce products;
s29, calculating the interest degree value of the user on each E-commerce product in the nearest neighbor set of the E-commerce products, and constructing a second interest set by the first j E-commerce products with the highest interest degree value.
Further, the constructing the target user interest model based on the improved vector space model representation method comprises the following steps:
s211, establishing an initialized user interest model according to registration information and interest labels of target users;
s212, mining keywords interested by a user through webpage text page content and constructing a text page vector space model;
s213, comparing the initialized user interest model with the text page vector space model, if the initialized user interest model and the text page vector space model contain the same interest words, updating the weight of the keywords in the initialized user interest model to be the weight of the keywords in the text page vector space model, if the initialized user interest model does not contain the interest words in the text page vector space model, adding the interest words in the text page vector space model into the initialized user interest model, and deleting the interest words lower than the average weight of the interest words in the text page vector space model in the initialized user interest model;
s214, initializing a user interest model by using the deleted interest word as a target user interest model.
Further, the calculation formula for calculating the similarity between the target user and other users based on the cosine similarity is as follows:
wherein, user i An interest model vector representing user i;
User j an interest model vector representing user j.
Further, the calculation formula for screening all the e-commerce products purchased by the user based on the user nearest neighbor set and calculating the interest level value of the user to the e-commerce products is as follows:
wherein S (a, p) represents the interest level value of the user a on the E-commerce product p;
representing average scores of all neighbor users on the E-commerce products p;
score b,p representing the score of the user b to the e-commerce product p;
representing the average score of user b for e-commerce product p;
sim (a, b) represents the similarity between user a and user b;
U e representing the set of nearest neighbors of the user consisting of the first m users.
Further, the obtaining a final push result based on the BP neural network by taking the first interest set and the second interest set as input nodes comprises the following steps:
s31, constructing and training a BP neural network, and obtaining a stable neural network;
s32, inputting the obtained first interest set and second interest set into the neural network as input nodes of the neural network;
s33, calculating based on the sim prediction function to obtain a final pushing result.
Further, the construction and training of the BP neural network and obtaining of the stable neural network comprise the following steps:
s311, acquiring sample data, and taking the sample data as training data;
s312, determining a BP neural network structure, and setting various parameters of the neural network; the parameters of the neural network comprise the number of neurons of an input layer, the number of nodes of an hidden layer and the number of neurons of an output layer
S313, initializing the BP neural network and obtaining an initial weight and a threshold value of the neural network;
s314, initializing a genetic algorithm and optimizing initial weights and thresholds of the neural network;
s315, setting a fitness function of a genetic algorithm;
s316, calculating a fitness value and performing genetic operation;
s317, judging whether the maximum iteration times are met, if yes, determining an optimal weight and a threshold, and if not, returning to the step S316;
and S318, taking the optimal individual in the genetic operation as an initial weight and a threshold value of the BP neural network, and training the BP neural network through the training data to obtain a stable BP neural network.
Further, the genetic manipulation may include selection, crossover and mutation, the selection may be by roulette, and the crossover may be by real crossover.
Further, the calculation formula for calculating the final pushing result based on the sim prediction function is as follows:
Y=sim(net,X)
where net represents the initial neural network;
x represents a K X N matrix input to the neural network, K represents the number of network inputs, and N represents the number of data samples;
y represents the output matrix q×n, Q represents the number of network outputs, and N represents the number of data samples.
The beneficial effects of the invention are as follows:
1. according to the invention, various data of the electronic commerce platform are used as a basis, and the target user interest model is constructed according to the browsing, collecting, evaluating and other actions of the target user, so that the real target user interest preference can be mined, the target user can be pushed accurately, the requirement of the target user can be met according to the user requirement, the searching difficulty of searching favorite goods of the user is reduced, the shopping experience of the user in using the electronic commerce platform and the user experience are improved, the use efficiency of the electronic commerce platform is improved, the operation cost of the electronic commerce platform can be reduced, and the commodity volume is increased.
2. According to the invention, through analyzing the similarity between the user and other users, the e-commerce products which are possibly interested by the target user can be identified, the interests of the user are analyzed and mined by means of the vector space interest model, the first interest set and the second interest set are obtained, and then the BP neural network is adopted for fusion, so that the analysis result is more accurate, and the e-commerce products which accord with the target user can be selected.
3. According to the invention, the BP neural network is optimized through the genetic algorithm, and the defect that the BP neural network is in a local minimum can be avoided by combining the global optimal characteristic of the genetic algorithm with the BP neural network, so that the service performance of the BP neural network can be improved, and further, the target user can be accurately pushed as an interesting product.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are needed in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flowchart of an information pushing method based on e-commerce big data according to an embodiment of the present invention.
Detailed Description
For the purpose of further illustrating the various embodiments, the present invention provides the accompanying drawings, which are a part of the disclosure of the present invention, and which are mainly used for illustrating the embodiments and for explaining the principles of the operation of the embodiments in conjunction with the description thereof, and with reference to these matters, it will be apparent to those skilled in the art to which the present invention pertains that other possible embodiments and advantages of the present invention may be practiced.
According to the embodiment of the invention, an information pushing method based on electronic commerce big data is provided.
The invention will be further described with reference to the accompanying drawings and the specific embodiments, as shown in fig. 1, the method for pushing information based on e-commerce big data according to the embodiment of the invention comprises the following steps:
s1, acquiring data information of a target user;
the step of acquiring the data information of the target user comprises the following steps:
s11, collecting target user data and e-commerce product data based on an e-commerce website;
specifically, the user data includes user registration information, interest tags of the users, and the like;
s12, collecting behavior data of a target user based on a preset log system;
specifically, the behavior data of the target user comprises a historical browsing record of the target user, a search commodity, praise scoring and the like;
specifically, the data of the website may be shared with the user data and the commodity data, and these data may be continuously updated by the website. The logs which can be extracted from the log system mainly analyze the evaluation of the commodity by the user, and the evaluation can be collected according to the data requirements of the recommendation engine.
Specifically, the target user interest model is built based on various data of the e-commerce platform and according to the browsing, collecting, evaluating and other actions of the target user, so that real interest preference of the target user can be mined, the target user can be pushed accurately, the requirement of the target user can be met according to the user requirement, the searching difficulty of searching favorite commodities of the user is reduced, the shopping experience of the user in using the e-commerce platform and the user experience are improved, the use efficiency of the e-commerce platform is improved, the operation cost of the e-commerce platform can be reduced, and the commodity volume is increased.
S2, constructing a target user interest model, and calculating to obtain a first interest set and a second interest set;
the method for constructing the target user interest model and calculating to obtain the first interest set and the second interest set comprises the following steps:
s21, constructing a target user interest model based on an improved vector space model representation method;
in particular, vector space models are the most commonly used user model representation, in which many systems model user models in such a way that the representation is represented by a vector of feature words and weights; the weight may be a real value or a boolean value, indicating whether and to what extent the user is interested. The vector space model-based representation method has the advantages that the importance degree of different concepts or project resources in the user model can be calculated explicitly, meanwhile, the vector can be operated, and the resource matching task in the subsequent stage is convenient.
The method for constructing the target user interest model based on the improved vector space model representation method comprises the following steps of:
s211, establishing an initialized user interest model according to registration information and interest labels of target users;
s212, mining keywords interested by a user through webpage text page content and constructing a text page vector space model;
specifically, the mining of the interest of the target user is the basis of establishing an interest model of the target user, and keywords of interest of the user can be obtained through content mining of the text page of the webpage, and the specific process is as follows:
(1) Analyzing and cleaning the webpage text content, wherein the webpage text contains a large number of CSS styles, such as HTML labels, dynamic scripts, multimedia contents and the like, extracting important text contents through DOM analysis technology, and removing redundant label texts;
(2) After analyzing and cleaning the text content of the webpage, the text is subjected to word segmentation, the document is decomposed into independent words through word segmentation technology, and the word segmentation technology comprises Chinese word segmentation and English word segmentation.
S213, comparing the initialized user interest model with the text page vector space model, if the initialized user interest model and the text page vector space model contain the same interest words, updating the weight of the keywords in the initialized user interest model to be the weight of the keywords in the text page vector space model, if the initialized user interest model does not contain the interest words in the text page vector space model, adding the interest words in the text page vector space model into the initialized user interest model, and deleting the interest words lower than the average weight of the interest words in the text page vector space model in the initialized user interest model;
s214, initializing a user interest model by using the deleted interest word as a target user interest model.
Specifically, the method and the device establish the initial user interest model according to the registration information and the interest tag of the target user, so that the interest of the target user is not better known than the user, the user initial interest model established according to the interest tag selected when the user registers the account can reflect the current interest of the user, and the interest of the target user can change along with the time, so that the interest model of the target user needs to be updated, and then the steps S211-S214 need to be repeated for updating the interest model of the target user, so that the interest of the target user can be accurately mined.
S22, calculating the similarity between the target user and other users based on the cosine similarity;
specifically, cosine similarity, also called cosine similarity, is evaluated by calculating the cosine value of the included angle of two vectors; drawing the vector into a vector space according to the coordinate value by cosine similarity; the most common application is to calculate text similarity; two texts are established according to words of the two texts, cosine values of the two vectors are calculated, and the similarity condition of the two texts in a statistical method can be known.
The calculation formula for calculating the similarity between the target user and other users based on the cosine similarity is as follows:
wherein, user i An interest model vector representing user i;
User j an interest model vector representing user j.
S23, screening out the first m users with highest similarity, and constructing a user nearest neighbor set;
s24, screening all the e-commerce products purchased by the user based on the nearest neighbor set of the user, and calculating the interest level value of the user on the e-commerce products;
the calculation formula for screening all the e-commerce products purchased by the user based on the user nearest neighbor set and calculating the interest level value of the user to the e-commerce products is as follows:
wherein S (a, p) represents the interest level value of the user a on the E-commerce product p;
representing average scores of all neighbor users on the E-commerce products p;
score b,p representing the score of the user b to the e-commerce product p;
representing the average score of user b for e-commerce product p;
sim (a, b) represents the similarity between user a and user b;
U e represents the user nearest neighbor set of the first m users, and e=1, 2, …, m.
S25, selecting the first j E-commerce products with the highest interest level value to construct a first interest set;
s26, selecting an e-commerce product with a higher score of a target user;
s27, calculating the similarity between the E-commerce products;
specifically, the similarity between the e-commerce products is calculated by adopting a cosine similarity principle.
S28, screening out the first k E-commerce products with highest similarity, and constructing a nearest neighbor set of the E-commerce products;
s29, calculating the interest degree value of the user on each E-commerce product in the nearest neighbor set of the E-commerce products, and constructing a second interest set by the first j E-commerce products with the highest interest degree value;
specifically, a calculation formula for calculating the interestingness value of the user for each e-commerce product in the nearest neighbor set of e-commerce products is as follows:
wherein S (a, p) represents the nearest neighbor set R of the user a to the E-commerce product k Interest value of each E-commerce product;
representing the nearest neighbor set R of the user a to the E-commerce product g Average scoring of the e-commerce products;
score a,h representing the score of the user a to the e-commerce product h;
sim (p, h) represents the similarity between the e-commerce product p and the e-commerce product h;
representing the average score of user b for e-commerce product p;
R g the first k e-commerce products are represented to constitute the e-commerce product nearest neighbor set, and g=1, 2, …, k.
S3, based on the BP neural network, the first interest set and the second interest set are used as input nodes to obtain a final pushing result;
the method for obtaining the final push result by taking the first interest set and the second interest set as input nodes based on the BP neural network comprises the following steps of:
s31, constructing and training a BP neural network, and obtaining a stable neural network;
the construction and training of the BP neural network and obtaining of the stable neural network comprise the following steps:
s311, acquiring sample data, and taking the sample data as training data;
specifically, the sample data includes interest e-commerce products of the user, scores of the e-commerce products by the user, and the like;
s312, determining a BP neural network structure, and setting various parameters of the neural network; each parameter of the neural network comprises the number of neurons of an input layer, the number of nodes of an hidden layer and the number of neurons of an output layer;
specifically, the BP neural network is a feedback neural network widely applied in the field of neural networks. The BP neural network reduces the difference between actual output and expected output through continuous training, and achieves the optimal state of the BP neural network through continuous indexes such as correction weight and the like, so as to predict the unknown condition;
the BP neural network structure comprises an input layer, a hidden layer and an output layer, wherein the input layer and the output layer are fixedly arranged as one layer, the hidden layer is designed flexibly, the number of the hidden layers is determined according to the actual problem, and in the normal case, the neural network usually comprising one hidden layer can solve most of the prediction problems; the neural network with the single hidden layer can analyze and calculate the nonlinear function with random complexity, and the process is efficient and simple.
S313, initializing the BP neural network and obtaining an initial weight and a threshold value of the neural network;
specifically, the initialization of the BP neural network is usually generated by adopting a newff function, and the format is usually as follows: net=newff (PR, [ S ] 1 ,S 2 …S n ],{TF 1 ,TF 2 …TF n },BTF);
Wherein PR represents an R2 matrix, representing the range between the minimum value and the maximum value of each dimension input in the R dimension input vector;
[S 1 ,S 2 …S n ]representing the number of neurons in each layer;
{TF 1 ,TF 2 …TF n -representing transfer functions employed by the neurons of each layer;
BTF represents a training function;
the newff function automatically initializes weights and thresholds of each layer of the network when initializing the BP network.
S314, initializing a genetic algorithm and optimizing initial weights and thresholds of the neural network;
specifically, the working principle of the genetic algorithm is that input data is firstly encoded, then crossover and mutation operation is selected through a certain probability until an individual with the largest fitness is selected as a target value to be output, and then the operation is stopped.
S315, setting a fitness function of a genetic algorithm;
specifically, the fitness function is a standard for measuring the individual capacity in the population, and in general, the objective function is selected as the fitness function of a genetic algorithm, and the inverse of the square error is adopted as the fitness function in the genetic algorithm.
S316, calculating a fitness value and performing genetic operation;
wherein the genetic manipulation comprises selection, crossing and variation, the selection is made by roulette, and the crossing is made by real crossing.
S317, judging whether the maximum iteration times are met, if yes, determining an optimal weight and a threshold, and if not, returning to the step S316;
specifically, the maximum iteration times are set to be 50, 100, 150 and 200 according to experience, and the optimal maximum iteration times value is selected through training;
and S318, taking the optimal individual in the genetic operation as an initial weight and a threshold value of the BP neural network, and training the BP neural network through the training data to obtain a stable BP neural network.
S32, inputting the obtained first interest set and second interest set into the neural network as input nodes of the neural network;
s33, calculating based on a sim prediction function to obtain a final pushing result;
the calculation formula for calculating the final pushing result based on the sim prediction function is as follows:
Y=sim(net,X)
where net represents the initial neural network;
x represents a K X N matrix input to the neural network, K represents the number of network inputs, and N represents the number of data samples;
y represents the output matrix q×n, Q represents the number of network outputs, and N represents the number of data samples.
And S4, pushing the pushing result to a target user.
In summary, by means of the technical scheme, the invention builds the interest model of the target user based on various data of the e-commerce platform and according to the browsing, collecting, evaluating and other actions of the target user, so that the real interest preference of the target user can be mined, the target user can be pushed accurately, the requirement of the target user can be met according to the user requirement, the searching difficulty of searching favorite goods of the user is reduced, the shopping experience of the user in using the e-commerce platform and the experience of the user are improved, the use efficiency of the e-commerce platform is improved, the operation cost of the e-commerce platform can be reduced, and the commodity traffic is increased; according to the invention, through analyzing the similarity between the user and other users, the e-commerce products which are possibly interested by the target user can be identified, the interests of the user are analyzed and mined by means of the vector space interest model to obtain the first interest set and the second interest set, and then the BP neural network is adopted for fusion, so that the analysis result is more accurate, and the e-commerce products which accord with the target user can be selected; according to the invention, the BP neural network is optimized through the genetic algorithm, and the defect that the BP neural network is in a local minimum can be avoided by combining the global optimal characteristic of the genetic algorithm with the BP neural network, so that the service performance of the BP neural network can be improved, and further, the target user can be accurately pushed as an interesting product.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, alternatives, and improvements that fall within the spirit and scope of the invention.
Claims (9)
1. An information pushing method based on electronic commerce big data is characterized by comprising the following steps:
s1, acquiring data information of a target user;
s2, constructing a target user interest model, and calculating to obtain a first interest set and a second interest set;
s3, based on the BP neural network, the first interest set and the second interest set are used as input nodes to obtain a final pushing result;
s4, pushing the pushing result to a target user;
the construction of the target user interest model and the calculation of the first interest set and the second interest set comprise the following steps:
s21, constructing a target user interest model based on an improved vector space model representation method;
s22, calculating the similarity between the target user and other users based on the cosine similarity;
s23, screening out the first m users with highest similarity, and constructing a user nearest neighbor set;
s24, screening all the e-commerce products purchased by the user based on the nearest neighbor set of the user, and calculating the interest level value of the user on the e-commerce products;
s25, selecting the first j E-commerce products with the highest interest level value to construct a first interest set;
s26, selecting an e-commerce product with a higher score of a target user;
s27, calculating the similarity between the E-commerce products;
s28, screening out the first k E-commerce products with highest similarity, and constructing a nearest neighbor set of the E-commerce products;
s29, calculating the interest degree value of the user on each E-commerce product in the nearest neighbor set of the E-commerce products, and constructing a second interest set by the first j E-commerce products with the highest interest degree value.
2. The method for pushing information based on e-commerce big data according to claim 1, wherein the step of obtaining the data information of the target user comprises the steps of:
s11, collecting target user data and e-commerce product data based on an e-commerce website;
s12, collecting behavior data of the target user based on a preset log system.
3. The information pushing method based on e-commerce big data according to claim 1, wherein the constructing the target user interest model based on the improved vector space model representation method comprises the following steps:
s211, establishing an initialized user interest model according to registration information and interest labels of target users;
s212, mining keywords interested by a user through webpage text page content and constructing a text page vector space model;
s213, comparing the initialized user interest model with the text page vector space model, if the initialized user interest model and the text page vector space model contain the same interest words, updating the weight of the keywords in the initialized user interest model to be the weight of the keywords in the text page vector space model, if the initialized user interest model does not contain the interest words in the text page vector space model, adding the interest words in the text page vector space model into the initialized user interest model, and deleting the interest words lower than the average weight of the interest words in the text page vector space model in the initialized user interest model;
s214, initializing a user interest model by using the deleted interest word as a target user interest model.
4. The information pushing method based on e-commerce big data according to claim 1, wherein the calculation formula for calculating the similarity between the target user and other users based on cosine similarity is as follows:
wherein, user i An interest model vector representing user i;
User j an interest model vector representing user j.
5. The method for pushing information based on e-commerce big data according to claim 1, wherein the calculation formula for screening all e-commerce products purchased by the user based on the user nearest neighbor set and calculating the interest level value of the user to the e-commerce products is as follows:
wherein S (a, p) represents the interest level value of the user a on the E-commerce product p;
representing average scores of all neighbor users on the E-commerce products p;
score b,p representing the score of the user b to the e-commerce product p;
representing the average score of user b for e-commerce product p;
sim (a, b) represents the similarity between user a and user b;
U e representing the set of nearest neighbors of the user consisting of the first m users.
6. The method for pushing information based on e-commerce big data according to claim 1, wherein the step of obtaining a final pushing result based on the BP neural network and using the first interest set and the second interest set as input nodes comprises the following steps:
s31, constructing and training a BP neural network, and obtaining a stable neural network;
s32, inputting the obtained first interest set and second interest set into the neural network as input nodes of the neural network;
s33, calculating based on the sim prediction function to obtain a final pushing result.
7. The information pushing method based on e-commerce big data according to claim 6, wherein the constructing and training the BP neural network and obtaining the stable neural network comprises the steps of:
s311, acquiring sample data, and taking the sample data as training data;
s312, determining a BP neural network structure, and setting various parameters of the neural network; the parameters of the neural network comprise the number of neurons of an input layer, the number of nodes of an hidden layer and the number of neurons of an output layer
S313, initializing the BP neural network and obtaining an initial weight and a threshold value of the neural network;
s314, initializing a genetic algorithm and optimizing initial weights and thresholds of the neural network;
s315, setting a fitness function of a genetic algorithm;
s316, calculating a fitness value and performing genetic operation;
s317, judging whether the maximum iteration times are met, if yes, determining an optimal weight and a threshold, and if not, returning to the step S316;
and S318, taking the optimal individual in the genetic operation as an initial weight and a threshold value of the BP neural network, and training the BP neural network through the training data to obtain a stable BP neural network.
8. The method of claim 7, wherein the genetic manipulation includes selecting, crossing and mutation, the selecting employs roulette, and the crossing employs real crossing.
9. The information pushing method based on e-commerce big data according to claim 6, wherein the calculation formula for calculating the final pushing result based on the sim prediction function is as follows:
Y=sim(net,X)
where net represents the initial neural network;
x represents a K X N matrix input to the neural network, K represents the number of network inputs, and N represents the number of data samples;
y represents the output matrix q×n, Q represents the number of network outputs, and N represents the number of data samples.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310503507.8A CN116542738A (en) | 2023-05-05 | 2023-05-05 | Information pushing method based on electronic commerce big data |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310503507.8A CN116542738A (en) | 2023-05-05 | 2023-05-05 | Information pushing method based on electronic commerce big data |
Publications (1)
Publication Number | Publication Date |
---|---|
CN116542738A true CN116542738A (en) | 2023-08-04 |
Family
ID=87457224
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310503507.8A Withdrawn CN116542738A (en) | 2023-05-05 | 2023-05-05 | Information pushing method based on electronic commerce big data |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116542738A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116883061A (en) * | 2023-09-08 | 2023-10-13 | 北京中奥通宇科技股份有限公司 | Adjustable intelligent line selection system for real-time analysis of data |
CN117668368A (en) * | 2023-12-18 | 2024-03-08 | 重庆机电职业技术大学 | E-commerce data pushing method and system based on big data |
-
2023
- 2023-05-05 CN CN202310503507.8A patent/CN116542738A/en not_active Withdrawn
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116883061A (en) * | 2023-09-08 | 2023-10-13 | 北京中奥通宇科技股份有限公司 | Adjustable intelligent line selection system for real-time analysis of data |
CN116883061B (en) * | 2023-09-08 | 2023-12-01 | 北京中奥通宇科技股份有限公司 | Adjustable intelligent line selection system for real-time analysis of data |
CN117668368A (en) * | 2023-12-18 | 2024-03-08 | 重庆机电职业技术大学 | E-commerce data pushing method and system based on big data |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111222332B (en) | Commodity recommendation method combining attention network and user emotion | |
Pan et al. | Study on convolutional neural network and its application in data mining and sales forecasting for E-commerce | |
CN110717098B (en) | Meta-path-based context-aware user modeling method and sequence recommendation method | |
CN111898031B (en) | Method and device for obtaining user portrait | |
CN116542738A (en) | Information pushing method based on electronic commerce big data | |
CN112991017A (en) | Accurate recommendation method for label system based on user comment analysis | |
CN114372573B (en) | User portrait information recognition method and device, computer equipment and storage medium | |
CN109584006B (en) | Cross-platform commodity matching method based on deep matching model | |
CN113268656A (en) | User recommendation method and device, electronic equipment and computer storage medium | |
CN111737578A (en) | Recommendation method and system | |
CN113946754A (en) | User portrait based rights and interests recommendation method, device, equipment and storage medium | |
CN114861050A (en) | Feature fusion recommendation method and system based on neural network | |
CN112632377A (en) | Recommendation method based on user comment emotion analysis and matrix decomposition | |
CN112055038B (en) | Method for generating click rate estimation model and method for predicting click probability | |
CN110851694A (en) | Personalized recommendation system based on user memory network and tree structure depth model | |
CN117574915A (en) | Public data platform based on multiparty data sources and data analysis method thereof | |
CN113761910A (en) | Comment text fine-grained emotion analysis method integrating emotional characteristics | |
CN116911949A (en) | Article recommendation method based on boundary rank loss and neighborhood perception graph neural network | |
CN115618875A (en) | Public opinion scoring method, system and storage medium based on named entity recognition | |
CN114298118B (en) | Data processing method based on deep learning, related equipment and storage medium | |
Agarwal et al. | Binarized spiking neural networks optimized with Nomadic People Optimization-based sentiment analysis for social product recommendation | |
Zafar Ali Khan et al. | Hybrid collaborative fusion based product recommendation exploiting sentiments from implicit and explicit reviews | |
CN116957740B (en) | Agricultural product recommendation system based on word characteristics | |
KR20200029647A (en) | Generalization method for curated e-Commerce system by user personalization | |
AU2021106572A4 (en) | A recommendation system and method for e-commerce using machine learning |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
WW01 | Invention patent application withdrawn after publication |
Application publication date: 20230804 |
|
WW01 | Invention patent application withdrawn after publication |