CN111861674B - Product recommendation method and system - Google Patents
Product recommendation method and system Download PDFInfo
- Publication number
- CN111861674B CN111861674B CN202010740021.2A CN202010740021A CN111861674B CN 111861674 B CN111861674 B CN 111861674B CN 202010740021 A CN202010740021 A CN 202010740021A CN 111861674 B CN111861674 B CN 111861674B
- Authority
- CN
- China
- Prior art keywords
- user
- products
- product
- type
- user type
- 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.)
- Active
Links
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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Engineering & Computer Science (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Finance (AREA)
- Accounting & Taxation (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- General Business, Economics & Management (AREA)
- Physiology (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Economics (AREA)
- Development Economics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a product recommendation method and a product recommendation system, which are characterized in that first behavior data of a user on a specific class of products are obtained; judging the user type indicated by the first line of data through each judgment rule in a preset knowledge base to obtain a judgment result of each judgment rule, and determining the user type of the user according to the judgment result of each judgment rule; if the user type of the user to which the first row of data belongs is the first user type of the time length spent in selecting the product, acquiring a purchased product list of the user to which the first row of data belongs; inputting the display sequence of the purchased product list, the first behavior data and the specific type products into a pre-trained first neural network model, determining the products which are most matched with the first behavior data from the specific type products through the first neural network model, recommending the most suitable products for the users who select the time spent on the products, reducing the number of the products displayed for the users and the time spent on selecting the products by the users, and improving the selection efficiency.
Description
Technical Field
The invention relates to the technical field of data processing, in particular to a product recommendation method and system.
Background
The arrival of the Internet age has led to changes in people's lifestyle and shopping habits, and more consumers begin to consume online through networks. The product provider distributes the products through the network, so that the difficulty of rapidly and accurately finding out the products meeting the requirements of users in mass products is increased while the products provided for the users are continuously increased. In order to conduct targeted product recommendation on different users, personalized product recommendation can be conducted on the basis of big data, and the personalized product recommendation is mainly based on the fact that related products are recommended for the users according to interest preferences of the users.
However, the above-mentioned product recommendation method has the following problems: the product recommendation method recommends a plurality of product data to users, and for a part of users with selection difficulties, the selection duration can be increased due to overload of the product data, and the selection efficiency can be reduced.
Disclosure of Invention
In view of the above, the embodiments of the present invention provide a product recommendation method and system, so as to achieve the purpose of targeted product recommendation for users with difficult selection.
In order to achieve the above object, an aspect of an embodiment of the present invention provides a product recommendation method, including:
acquiring first behavior data of a user on a specific class of products; the first row of data indicates a viewing condition of the user of at least one product of the particular category of products;
judging the user type indicated by the first line of data through each judging rule in a preset knowledge base to obtain a judging result of each judging rule, and determining the user type of the user according to the judging result of each judging rule, wherein the user type comprises a first user type and a second user type, and the time spent by the user selected by the first user type is longer than the time spent by the user selected by the second user type;
if the user type of the user to which the first line of data belongs is a first user type, acquiring a purchased product list of the user to which the first line of data belongs;
inputting the purchased product list, the first behavior data and the display sequence of the specific category of products into a pre-trained first neural network model, and determining the products which are most matched with the first behavior data from the specific category of products through the first neural network model.
Another aspect of an embodiment of the present invention provides a product recommendation system, including: the device comprises a first acquisition unit, a judgment unit, a second acquisition unit and a first determination unit;
the first acquisition unit is used for acquiring first behavior data of a user on a specific class of products; the first row of data indicates a viewing condition of the user of at least one product of the particular category of products;
the judging unit is used for judging the user type indicated by the first row of data through each judging rule in a preset knowledge base to obtain a judging result of each judging rule, and determining the user type of the user according to the judging result of each judging rule, wherein the user type comprises a first user type and a second user type, and the time spent by the user belonging to the first user type is longer than the time spent by the user belonging to the second user type;
the second obtaining unit is configured to obtain a purchased product list of the user to which the first data belong if the user type of the user to which the first data belong is the first user type;
the first determining unit is configured to input the purchased product list, the first behavior data, and the display sequence of the specific category of products into a pre-trained first neural network model, and determine, from the specific category of products, a product that is the best match with the first behavior data through the first neural network model.
According to the technical scheme, the first behavior data of the user on the specific category of products is obtained; judging the user type indicated by the first line of data through each judgment rule in a preset knowledge base to obtain a judgment result of each judgment rule, and determining the user type of the user according to the judgment result of each judgment rule; if the user type of the user to which the first row of data belongs is the first user type, acquiring a purchased product list of the user to which the first row of data belongs; inputting a display sequence of a purchased product list, first behavior data and specific type products into a pre-trained first neural network model, determining the products which are the most matched with the first behavior data from the specific type products through the first neural network model, recommending the most suitable products for users of the first user type, wherein the time spent by the users of the first user type in selecting the products is longer than the time spent by the users of the second user type in selecting the products, and in general, the users with difficulty in selecting the products can spend too much time, so that the products most suitable for the users with difficulty in selecting can be recommended for the users with difficulty in selecting, the quantity of the products displayed for the users with difficulty in selecting is reduced, the time spent by the users in selecting the products is shortened, and the selection efficiency is improved.
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 required to be used in the embodiments or the description of the prior art will be briefly described below, and it is obvious that the drawings in the following description are only embodiments of the present invention, and that other drawings can be obtained according to the provided drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a product recommendation method disclosed in an embodiment of the present invention;
FIG. 2 is a flowchart of a process for generating a judgment rule in a preset knowledge base;
FIG. 3 is a flow chart of a training process of a first neural network model;
FIG. 4 is a flowchart of a method for recommending products according to another embodiment of the present invention;
FIG. 5 is a block diagram of a product recommendation system correspondingly disclosed in an embodiment of the present invention;
fig. 6 is a block diagram of a product recommendation system according to another embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
In this application, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus 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 apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
As shown in fig. 1, a flowchart of a product recommendation method disclosed in an embodiment of the present invention includes the following steps:
s101, acquiring first behavior data of a user on a specific category of products.
In this embodiment, the first row of data indicates a user's view of at least one product in a particular category of products. Viewing conditions include, but are not limited to, time of viewing, number of views, and operations performed by the user on the product (e.g., purchasing, collecting, and/or joining a shopping cart) while viewing, etc. The first behavioral data may include overall viewing of the particular category of products and/or viewing of individual products in the particular category of products. In particular, the first behavioral data may include, but is not limited to, at least one of the following behavioral parameters: the number of clicks on a product of a particular category (including the number of clicks on a single product and/or the total number of clicks on a product of a category), the number of viewed products, the length of time the product is viewed (including the length of time the single product is viewed and/or the total length of time the product of a category) and the purchase of the product. The content of the first behavior data is not limited herein, and may be adjusted and defined according to scene needs.
In this embodiment, the specific type of product is used to indicate that the first behavior data is derived from the same type of product, and there may be multiple implementations of specifying the specific type and obtaining the first behavior data of the user on the specific type of product, which are not listed here, and the following description is given by taking two implementations as examples:
one implementation is as follows: if the product is displayed according to the category, for example, different categories are displayed on different pages. And responding to the operation of closing the product display page by the user. And acquiring the product category displayed by the closed product display page as a specific category. And acquiring the times of clicking products, the number of products to be checked, the time length for checking the products and the conditions of purchasing the products of the user on the products of the specific category as first behavior data of the user on the products of the specific category. Wherein the statistical time of the first behavior data may be from the completion of the loading of the presentation page for a particular category of product to the receipt of a designation by the user to close the presentation page.
Another implementation is:
firstly, the time length for checking the product in the behavior data of the user is acquired. Wherein the behavioural data comprises behavioural data of the user at least one category of product, the behavioural data corresponding to the category of product. That is to say, the first behavior data is stored in a collection form, the collection comprises a plurality of groups of first behavior data corresponding to product categories, the time length for checking the product is one parameter in the behavior data, and the time length for checking the product corresponds to the product categories.
Secondly, taking the product category with the product checking time length longer than the preset time length as a specific category.
Because users who view products for a longer period of time are more likely to be users with difficulty in selection, the category of products for which the period of time for viewing the products is longer than the preset period of time is taken as a specific category. The size of the preset time period is not limited herein, and may be set as needed. It should be noted that, in other implementations, the time period for checking the product may be replaced by other parameters in the behavior data, such as, for example, the number of checked products, which is not limited herein.
Finally, the times of clicking products, the number of products to be checked, the time length of checking the products and the conditions of purchasing the products of the user on the products of the specific category are obtained as first behavior data of the user on the products of the specific category.
Taking the example of storing a plurality of groups of first behavior data corresponding to the product category identification information in a collection form, the process of acquiring the first behavior characteristics can screen out the first behavior data of a specific category according to the category identification information of the specific category.
S102, judging the user type indicated by the first line of data through each judgment rule in a preset knowledge base, obtaining a judgment result of each judgment rule, and determining the user type of the user according to the judgment result of each judgment rule.
In this embodiment, the user types include a first user type and a second user type, and the time spent by the user belonging to the first user type selecting the product is longer than the time spent by the user belonging to the second user type selecting the product. For example, the first user type may be a user with difficulty in selection and the second user type may be a user without difficulty in selection.
The preset knowledge base may be a knowledge base in an expert system, where the preset knowledge base includes at least one judgment rule for judging the user type according to the first behavior feature. The judging rule includes judging conditions and conclusions, and the format of the judging rule can be as follows: IF parameter 1 satisfies condition 1, then user is the first user type. Wherein, the parameter 1 satisfies the condition 1 as a judging condition, and the user is the first user type as a conclusion. Parameter 1 is a parameter in the first behavioral characteristic, and condition 1 may be a definition of a parameter value of parameter 1. A specific example of a judgment rule is given for easy understanding: the average number of views of an IF individual product is greater than 10, the then user being the first user type. The judgment condition may be a condition in which parameter values of a plurality of parameters are defined, and the plurality of conditions are combined into one judgment condition by a logic operation, for example: IF product quantity is less than 5 and viewing time is longer than 30 minutes. In addition, in addition to the format of the above example in which the user type (the first user type or the second user type) is directly given, the conclusion in the judgment rule may be to give a probability value or score that the user is the user of the first user type. The format of the judgment rule is not particularly limited here.
The following describes a user type implementation process of determining the user according to the judgment rule by taking the probability value or the score of the user as the first user type as an example given by the judgment rule: and obtaining a judging result of each judging rule according to the first behavior characteristic, wherein the judging result indicates that the user type belongs to the probability value or the score of the first user type. And each judgment rule corresponds to a weight, each judgment result is multiplied by the weight and then summed to obtain total probability or score, and the user type of the user is determined according to the total probability or score value range. After the user type of the user is determined, the user type of the user can be stored, so that the use of subsequent operation is convenient.
In this embodiment, the judgment rules in the preset knowledge base may be obtained through a machine learning algorithm, which includes but is not limited to a genetic algorithm, and in this embodiment, in the process of obtaining the judgment rules, an implementation manner of obtaining the judgment rules is described by taking the genetic algorithm as an example.
Referring to fig. 2, a flowchart illustrating a process for generating a judgment rule in a preset knowledge base is shown, including the following steps:
s201, setting each initial judgment rule in a preset knowledge base.
Any initial judging rule is used for judging whether the user type is a first user type or a second user type and indicating the corresponding relation between the user type and the first behavior data; that is to say for indicating the content of the first behavior data on which the user type is determined. The judging condition of the initial judging rule can be set according to the behavior parameters in the first behavior characteristic, and the conclusion is the user type of the user. All combinations between the first row data and the user type can be exhausted, and as an initial judgment rule, the initial judgment rule can be a table, and each row records one piece of rule data: the first row is data-user type. The initial weight of the initial judgment rule may also be set, and the initial judgment rule may be recorded as: first behavior data 1-first user type-weight 1; first behavior data 2-first user type-weight 2; the first row is data 3-the second user type-weight 3. For the format description of the initial judgment rule, please refer to the format description of the judgment rule in step S102, the formats of the two rules are similar, and will not be described herein.
The initial judgment rule is a rule initial value of the judgment rule obtained by learning in a preset knowledge base. Furthermore, in order to make the set initial judgment rule more reasonable, reduce the workload of subsequent adjustment to the judgment rule, historical user data of the system can be collected, and the correlation between different characteristic parameters in the first behavior data and the user types can be obtained by carrying out statistics and induction on the historical user data. And setting an initial judgment rule according to the relevance of different characteristic parameters in the first behavior data of the induction statistics and the user type. The initial judgment rule obtained in this way is closer to the judgment rule in the preset knowledge base. The historical user data comprises a plurality of groups of data pairs of first behavior data and user types, the user types of the users are calibrated, the user types can be determined through investigation on the purchasing product behaviors of the users, and the user types obtained through investigation are calibrated. Summarizing historical user data, extracting statistical correlation between behavior data and user types, and obtaining initial judgment rules, wherein the weight corresponding to each initial judgment rule can be set in the process of obtaining the initial judgment rules.
S202, inputting each initial judgment rule and corresponding weight into a machine learning algorithm for judging the user type, and obtaining the user type corresponding to each initial judgment rule output by the machine learning algorithm.
S203, outputting the user type corresponding to each initial judgment rule according to the machine learning algorithm, and adjusting the initial judgment rule and the corresponding weight in the preset knowledge base to obtain the judgment rule of the preset knowledge base.
Steps S202 and S203 are processes of training initial judgment rules and weights to obtain the optimal solutions of the judgment rules and the corresponding weights. There are various implementations of machine learning algorithms, and thus there are various implementations of optimizing initial rules and weights by machine learning algorithms. The following describes a process of optimizing the initial judgment rule and the corresponding weight, taking a genetic algorithm as an example.
Sample data for training is prepared, the sample data being a plurality of sets of data pairs of the first behavior data and the user type, and the historical user data in step S201 may be used as the sample data. And establishing the prediction accuracy of the initial judgment rule and the corresponding weight as an evaluation function, inputting the initial judgment rule and the corresponding weight into the genetic variation model, and adjusting the initial judgment rule and/or the corresponding weight for a plurality of times through a genetic algorithm according to the result of the evaluation function until the evaluation function reaches the preset requirement, and outputting the optimal solution of the judgment rule and the corresponding weight by the genetic variation model to obtain the judgment rule of the preset knowledge base. Specifically, the method comprises the following steps:
the first step: the initial judgment rule and the corresponding weight are used as a population to be used as the primary solution of the problem.
And a second step of: and searching a proper coding scheme to code the individuals in the population, for example, an initial judgment rule and the corresponding weight can be selected to construct a knowledge base for predicting the user type.
And a third step of: the accuracy of predicting the user type by the knowledge base is used as the fitness of the individuals, the fitness of each individual in the population is calculated, and the calculated fitness provides basis for the subsequent individual selection.
Fourth step: parent and parent involved in reproduction are selected from the initial judgment rules and the corresponding weights according to the fitness (namely, the accuracy of predicting the user type by the knowledge base), wherein the selection principle is that the higher the fitness is, the more likely the individuals are selected, so that the individuals with low fitness are continuously eliminated.
Fifth step: genetic operation is carried out on the selected father and mother body, namely, the genes of the father and mother body are duplicated, and operators such as crossing, mutation and the like are adopted to generate offspring. On the basis of retaining excellent genes to a large extent, mutation increases the diversity of genes, thereby improving the probability of finding the optimal solution.
Sixth step: and judging whether to continue to execute the algorithm or find out the individual with the highest fitness in all the offspring to return as the optimal solution according to a certain criterion, and ending the program. The criterion for the judgment can be a set threshold value of the evaluation function or a specified iteration number.
In other implementations, outputting a user type corresponding to each initial judgment rule according to a machine learning algorithm, and adjusting the initial judgment rule and the corresponding weight in a preset knowledge base to obtain the judgment rule of the preset knowledge base, wherein the judgment rule comprises the following steps:
if the machine learning algorithm indicates that the initial judgment rule cannot identify the user type, deleting the initial judgment rule from the preset knowledge base; or the initial judgment rule corresponding weight is set to 0.
If the machine learning algorithm indicates that the user type identified by the initial judgment rule is changed from the first user type to the second user type or from the second user type to the first user type, modifying the judgment result corresponding to the initial judgment rule into the result output by the machine learning algorithm; or the corresponding weight of the initial judgment rule is set to be the opposite number.
If the machine learning mode indicates that the weight identified by the initial judgment rule changes, modifying the weight corresponding to the initial judgment rule into the weight output by the machine learning algorithm.
And S103, if the user type of the user to which the first line of data belongs is the first user type, acquiring a purchased product list of the user to which the first line of data belongs.
S104, inputting the display sequence of the purchased product list, the first behavior data and the specific category products into a pre-trained first neural network model, and determining the products which are most matched with the first behavior data from the specific category products through the first neural network model.
In this embodiment, the display order of the specific product may be the display order of the specific product in the page. Referring to fig. 3, which shows a flowchart of a training process of the first neural network model, the method includes the following steps:
s301, acquiring an initial neural network model.
The initial neural network model can be a general model for recommending products for users, and the input data of the initial neural network model comprises first behavior data of the users on a certain type of products, the display sequence of the products and corresponding purchased product lists of the users; the output data are respectively recommended product information, for example, the product information can be product identification.
S302, acquiring historical data of a user as training data of an initial neural network model.
The historical data of the user is the historical data of the user belonging to the first user type, wherein one group of training data comprises input data of an initial neural network model and output data obtained correspondingly; the input data are first behavior data of historical data on a certain type of products, the display sequence of the type of products and corresponding purchased product lists, and the output data are product information selected from the type of products in the historical data.
And S303, training the initial neural network model by using training data to obtain a first neural network model.
By training the initial neural network by using the historical data of the user determined to be the first user type, the initial neural network is enabled to be structurally adjusted according to the historical data of the user, which is equivalent to obtaining a neural network model which learns the selection mode of the user, and more suitable products can be recommended for the user.
According to the product recommendation method, first behavior data of a user on a specific category of products are obtained; judging the user type indicated by the first line of data through each judgment rule in a preset knowledge base to obtain a judgment result of each judgment rule, and determining the user type of the user according to the judgment result of each judgment rule; if the user type of the user to which the first row of data belongs is the first user type, acquiring a purchased product list of the user to which the first row of data belongs; inputting a display sequence of a purchased product list, first behavior data and specific type products into a pre-trained first neural network model, determining the products which are the most matched with the first behavior data from the specific type products through the first neural network model, recommending the most suitable products for users of the first user type, wherein the time spent by the users of the first user type in selecting the products is longer than the time spent by the users of the second user type in selecting the products, and in general, the users with difficulty in selecting the products can spend too much time, so that the products most suitable for the users with difficulty in selecting can be recommended for the users with difficulty in selecting, the quantity of the products displayed for the users with difficulty in selecting is reduced, the time spent by the users in selecting the products is shortened, and the selection efficiency is improved.
For further improvement, please refer to fig. 4, which shows a flowchart of another embodiment of the present invention for providing a product recommendation method, and compared with fig. 1, the embodiment further includes the following steps:
s105, highlighting the product which is matched with the first behavior data best in the specific type of product.
In this embodiment, the highlighting may be to show the most matched product in front of other products, or to mark the most matched product in a specific display manner, such as a bright color or a magic effect; or directly outputting a message to prompt the user to select the best matching product. In other embodiments, to exclude interference from other products to the user, only one of the best matching products may be displayed in the recommendation area.
And S106, if the user type of the user to which the first row of data belongs is a second user type, determining a plurality of products matched with the user information from the specific category of products through a second neural network model. The process is as follows:
if the user type of the user to which the first row of data belongs is the second user type, acquiring the user information of the user to which the first row of data belongs, inputting the user information and the list data representing the specific category of products into a pre-trained second neural network model, and determining a plurality of products matched with the user information from the specific category of products through the second neural network model. Wherein the second neural network model is different from the model structure of the first neural network model.
In other embodiments, in order to alleviate the calculation pressure, the product display of the user of the second user type may not be adjusted, that is, in step S106, if the user type of the user to which the first row of data belongs is the second user type, no processing is performed.
The embodiment provides a processing method when the user is the second user type, and further improves the scheme of the invention.
Based on the product recommendation method disclosed in the above embodiment of the present invention, the embodiment of the present invention also correspondingly discloses a product recommendation system, please refer to fig. 5, which shows a structure diagram of the product recommendation system, including: a first acquisition unit 101, a judgment unit 102, a second acquisition unit 103, and a first determination unit 104.
The first obtaining unit 101 is configured to obtain first behavior data of a user on a specific product category.
The first row of data indicates a user's view of at least one product in the particular category of products.
The judging unit 102 is configured to judge the user type indicated by the first line of data according to each judging rule in the preset knowledge base, obtain a judging result of each judging rule, and determine the user type of the user according to the judging result of each judging rule.
The user types comprise a first user type and a second user type, and the time spent by the user belonging to the first user type for selecting the product is longer than the time spent by the user belonging to the second user type for selecting the product.
The second obtaining unit 103 is configured to obtain a purchased product list of the user to whom the first data belong if the user type of the user to which the first data belong is the first user type.
The first determining unit 104 is configured to input the purchased product list, the first behavior data, and the display sequence of the specific category of products into a pre-trained first neural network model, and determine, from the specific category of products, a product that is the most matched with the first behavior data through the first neural network model.
The working process of each unit is referred to the related description of steps S101-S104 in the above method embodiment, and will not be repeated here.
The product recommendation system acquires first behavior data of a user on a specific category of products through a first acquisition unit 101; the judging unit 102 judges the user type indicated by the first line of data through each judging rule in the preset knowledge base to obtain a judging result of each judging rule, and determines the user type of the user according to the judging result of each judging rule; if the user type of the user to which the first data line belongs is the first user type, the second obtaining unit 103 obtains a purchased product list of the user to which the first data line belongs; the first determining unit 104 inputs the purchased product list, the first behavior data and the display sequence of the specific category products into a pre-trained first neural network model, determines the products which are the most matched with the first behavior data from the specific category products through the first neural network model, and realizes that the users of the first user type recommend the most suitable products, the time spent by the users of the first user type in selecting the products is longer than the time spent by the users of the second user type in selecting the products, and in general, the users with difficulty in selecting the products can spend too much time in selecting the products, so that the products which are the most suitable for the users can be recommended for the users with difficulty in selecting, the number of the products displayed for the users with difficulty in selecting is reduced, the time spent by the users in selecting the products is shortened, and the selection efficiency is improved.
Referring to fig. 6, a block diagram of a product recommendation system according to another embodiment of the present invention is shown, and compared with fig. 5, the product recommendation system further includes: a display unit 105 and a second determination unit 106.
And a display unit 105 for highlighting a product that most matches the first behavior data among the specific type of products.
And the second determining unit 106 is configured to, if the user type of the user to which the first row of data belongs is the second user type, obtain user information of the user to which the first row of data belongs, input the user information and inventory data representing the specific category of products into a pre-trained second neural network model, and determine a plurality of products matching the user information from the specific category of products through the second neural network model.
The working process of each unit is referred to the related description of steps S105-S106 in the above method embodiment, and will not be repeated here.
The embodiment provides a processing method when the user is the second user type, and further improves the scheme of the invention.
The embodiments in the present specification are described in a progressive or combined manner, and the same and similar parts of the embodiments are all referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for a system or system embodiment, since it is substantially similar to a method embodiment, the description is relatively simple, with reference to the description of the method embodiment being made in part. The systems and system embodiments described above are merely illustrative, wherein the elements illustrated as separate elements may or may not be physically separate, and the elements shown as elements may or may not be physical elements, may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present invention without undue burden.
Those of skill would further appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the various illustrative elements and steps are described above generally in terms of functionality in order to clearly illustrate the interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (10)
1. A method of product recommendation, comprising:
acquiring first behavior data of a user on a specific class of products; the first row of data indicates a viewing condition of the user of at least one product of the particular category of products;
judging the user type indicated by the first line of data through each judging rule in a preset knowledge base to obtain a judging result of each judging rule, and determining the user type of the user according to the judging result of each judging rule, wherein the user type comprises a first user type and a second user type, and the time spent by the user selected by the first user type is longer than the time spent by the user selected by the second user type;
if the user type of the user to which the first line of data belongs is a first user type, acquiring a purchased product list of the user to which the first line of data belongs;
inputting the purchased product list, the first behavior data and the display sequence of the specific category of products into a pre-trained first neural network model, and determining the products which are most matched with the first behavior data from the specific category of products through the first neural network model.
2. The product recommendation method according to claim 1, further comprising;
among the specific category of products, the product that most matches the first row of data is highlighted.
3. The product recommendation method according to claim 1, further comprising:
if the user type of the user to which the first row of data belongs is a second user type, acquiring user information of the user to which the first row of data belongs, inputting the user information and list data representing the specific category of products into a pre-trained second neural network model, and determining a plurality of products matched with the user information from the specific category of products through the second neural network model.
4. The product recommendation method of claim 1, wherein the obtaining first behavior data of the user on the specific category of products comprises:
responding to the operation of closing the product display page by the user;
acquiring the product category displayed by the closed product display page as a specific category;
acquiring the times of clicking products, the number of products to be checked, the time length of checking the products and the conditions of purchasing the products of a user on the specific type of products as first behavior data of the user on the specific type of products;
or alternatively, the process may be performed,
acquiring the time length of checking products in behavior data of a user;
taking the product category with the product checking time length longer than the preset time length as the specific category;
and acquiring the times of clicking products, the number of products to be checked, the time length for checking the products and the conditions of purchasing the products of the user on the specific type of products as first behavior data of the user on the specific type of products.
5. The product recommendation method according to claim 1, wherein the generating process of the judgment rule in the preset knowledge base includes:
setting each initial judgment rule in the preset knowledge base, wherein any initial judgment rule is used for judging whether the user type is a first user type or a second user type and indicating the corresponding relation between the user type and the first behavior data;
inputting each initial judgment rule and corresponding weight into a machine learning algorithm for judging the user type, and obtaining the user type corresponding to each initial judgment rule output by the machine learning algorithm;
and outputting a user type corresponding to each initial judgment rule according to the machine learning algorithm, and adjusting the initial judgment rules and the corresponding weights in the preset knowledge base to obtain the judgment rules of the preset knowledge base.
6. The product recommendation method according to claim 5, wherein the outputting, according to the machine learning algorithm, the user type and the weight corresponding to each initial judgment rule, and adjusting the initial judgment rule in the preset knowledge base to obtain the judgment rule of the preset knowledge base includes:
if the machine learning algorithm indicates that the initial judgment rule cannot identify the user type, deleting the initial judgment rule from the preset knowledge base;
if the machine learning algorithm indicates that the user type identified by the initial judgment rule is changed from the first user type to the second user type or from the second user type to the first user type, modifying the judgment result corresponding to the initial judgment rule into the result output by the machine learning algorithm;
and if the machine learning algorithm indicates that the weight identified by the initial judgment rule changes, modifying the weight corresponding to the initial judgment rule into the weight output by the machine learning algorithm.
7. A product recommendation system, comprising: the device comprises a first acquisition unit, a judgment unit, a second acquisition unit and a first determination unit;
the first acquisition unit is used for acquiring first behavior data of a user on a specific class of products; the first row of data indicates a viewing condition of the user of at least one product of the particular category of products;
the judging unit is used for judging the user type indicated by the first row of data through each judging rule in a preset knowledge base to obtain a judging result of each judging rule, and determining the user type of the user according to the judging result of each judging rule, wherein the user type comprises a first user type and a second user type, and the time spent by the user belonging to the first user type is longer than the time spent by the user belonging to the second user type;
the second obtaining unit is configured to obtain a purchased product list of the user to which the first data belong if the user type of the user to which the first data belong is the first user type;
the first determining unit is configured to input the purchased product list, the first behavior data, and the display sequence of the specific category of products into a pre-trained first neural network model, and determine, from the specific category of products, a product that is the best match with the first behavior data through the first neural network model.
8. The product recommendation system of claim 7, further comprising;
and the display unit is used for highlighting the product which is most matched with the first line of data in the specific category of products.
9. The product recommendation system of claim 7, further comprising:
and the second determining unit is used for acquiring the user information of the user to which the first row of data belongs if the user type of the user to which the first row of data belongs is a second user type, inputting the user information and the list data representing the specific category of products into a pre-trained second neural network model, and determining a plurality of products matched with the user information from the specific category of products through the second neural network model.
10. The product recommendation system of claim 7, wherein the first acquisition unit is specifically configured to:
responding to the operation of closing the product display page by the user; acquiring the product category displayed by the closed product display page as a specific category; acquiring the times of clicking products, the number of products to be checked, the time length of checking the products and the conditions of purchasing the products of a user on the specific type of products as first behavior data of the user on the specific type of products;
or alternatively, the process may be performed,
acquiring the time length of checking products in behavior data of a user; taking the product category with the product checking time length longer than the preset time length as the specific category; and acquiring the times of clicking products, the number of products to be checked, the time length for checking the products and the conditions of purchasing the products of the user on the specific type of products as first behavior data of the user on the specific type of products.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010740021.2A CN111861674B (en) | 2020-07-28 | 2020-07-28 | Product recommendation method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010740021.2A CN111861674B (en) | 2020-07-28 | 2020-07-28 | Product recommendation method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111861674A CN111861674A (en) | 2020-10-30 |
| CN111861674B true CN111861674B (en) | 2023-07-18 |
Family
ID=72948623
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010740021.2A Active CN111861674B (en) | 2020-07-28 | 2020-07-28 | Product recommendation method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111861674B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117056613A (en) * | 2023-10-12 | 2023-11-14 | 中质国优测评技术(北京)有限公司 | Evaluation optimization method and system based on user joint preference |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109242631A (en) * | 2018-09-17 | 2019-01-18 | 平安科技（深圳）有限公司 | Product intelligent recommended method, server and storage medium |
| CN111046297A (en) * | 2020-03-12 | 2020-04-21 | 深圳市成功快车科技有限公司 | Service intelligent matching recommendation method, device, equipment and storage medium based on machine learning algorithm |
-
2020
- 2020-07-28 CN CN202010740021.2A patent/CN111861674B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109242631A (en) * | 2018-09-17 | 2019-01-18 | 平安科技（深圳）有限公司 | Product intelligent recommended method, server and storage medium |
| CN111046297A (en) * | 2020-03-12 | 2020-04-21 | 深圳市成功快车科技有限公司 | Service intelligent matching recommendation method, device, equipment and storage medium based on machine learning algorithm |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111861674A (en) | 2020-10-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106485562B (en) | Commodity information recommendation method and system based on user historical behaviors | |
| US11449916B2 (en) | Information push method and apparatus, server, and storage medium | |
| JP6719727B2 (en) | Purchase behavior analysis device and program | |
| Chan et al. | An intelligent fuzzy regression approach for affective product design that captures nonlinearity and fuzziness | |
| CN110147925B (en) | Risk decision method, device, equipment and system | |
| US20020055890A1 (en) | Recommendation information providing method, recommendation information transmission system, recommendation information transmission apparatus and computer memory product | |
| JP2010204966A (en) | Sampling device, sampling method, sampling program, class distinction device and class distinction system | |
| CN107077687A (en) | Obtain the data relevant with consumer, the processing data and the output that the consumer's quotation being electronically generated is provided | |
| CN108595493A (en) | Method for pushing and device, storage medium, the electronic device of media content | |
| CN110874787A (en) | Recommendation model effect evaluation method and related device | |
| CN111695084A (en) | Model generation method, credit score generation method, device, equipment and storage medium | |
| CN111797320A (en) | Data processing method, device, equipment and storage medium | |
| CN111861674B (en) | Product recommendation method and system | |
| JP2013033376A (en) | Information processor, information processing method, and program | |
| CN115204985A (en) | Shopping behavior prediction method, device, equipment and storage medium | |
| CN110502639B (en) | Information recommendation method and device based on problem contribution degree and computer equipment | |
| CN109063120A (en) | A kind of collaborative filtering recommending method and device based on cluster | |
| CN111177564B (en) | Product recommendation method and device | |
| CN112365302B (en) | Product recommendation network training method, device, equipment and medium | |
| KR20200065754A (en) | Method for recommending book and service device supporting the same | |
| JP4129127B2 (en) | Recommended information transmission system | |
| CN108073702A (en) | Wood products recommend method, apparatus and computer readable storage medium | |
| JP2006012115A (en) | Recommendation information providing method, recommendation information transmission system, recommendation information transmission apparatus and storage medium | |
| CN112925982B (en) | User redirection method and device, storage medium and computer equipment | |
| KR102643108B1 (en) | Good exposure system based on keyword analyzing |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |