CN117057886A - Commodity recommendation method, commodity recommendation device, commodity recommendation equipment and storage medium - Google Patents
Commodity recommendation method, commodity recommendation device, commodity recommendation equipment and storage medium Download PDFInfo
- Publication number
- CN117057886A CN117057886A CN202311110267.1A CN202311110267A CN117057886A CN 117057886 A CN117057886 A CN 117057886A CN 202311110267 A CN202311110267 A CN 202311110267A CN 117057886 A CN117057886 A CN 117057886A
- Authority
- CN
- China
- Prior art keywords
- commodity
- data
- attribute
- user
- determining
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 50
- 230000006399 behavior Effects 0.000 claims abstract description 55
- 238000012549 training Methods 0.000 claims description 19
- 230000002452 interceptive effect Effects 0.000 claims description 13
- 230000006870 function Effects 0.000 claims description 12
- 238000006243 chemical reaction Methods 0.000 claims description 9
- 238000012545 processing Methods 0.000 claims description 8
- 238000005070 sampling Methods 0.000 claims description 8
- 238000012163 sequencing technique Methods 0.000 claims description 7
- 238000010276 construction Methods 0.000 claims description 5
- 230000009466 transformation Effects 0.000 claims description 5
- 230000008707 rearrangement Effects 0.000 claims description 3
- 238000013473 artificial intelligence Methods 0.000 abstract description 2
- 230000008569 process Effects 0.000 description 6
- 238000004891 communication Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 239000013598 vector Substances 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000003287 optical effect Effects 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
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Physics & Mathematics (AREA)
- Economics (AREA)
- General Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Development Economics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention belongs to the technical field of artificial intelligence, and discloses a commodity recommendation method, a commodity recommendation device, commodity recommendation equipment and a storage medium. The method comprises the following steps: determining user characteristic data and tag data based on the user history data; establishing a user diversity preference model based on the user characteristic data and the tag data; determining diversity preference data of a target user based on the user diversity preference model; determining a breaking-up parameter of a breaking-up stage based on the diversity preference data of the target user; and recommending the commodity to the target user based on the scattering parameters of the scattering stage. By the method, the user diversity preference model is built according to the user history behaviors, and the preference degree of the user for diversity is predicted, so that key parameters of the scattering strategy are automatically updated, the recommendation accuracy is improved, the modeling takes the user as a dimension, and the personalized requirements of the user can be met.
Description
Technical Field
The present invention relates to the field of artificial intelligence technologies, and in particular, to a method, an apparatus, a device, and a storage medium for recommending commodities.
Background
With the development of digital economy, more and more information is contacted by people, a recommendation system plays an important role in assisting people in finding interesting contents, and the improvement of user diversity is a key part in improving the recommendation accuracy of the recommendation system. However, the key parameters of the existing method for improving the recommendation diversity need to be manually specified, the accuracy is low, the recommendation accuracy is low, and as the user characteristics are not utilized, the diversity recommendation modes of all people are the same, and the method cannot be applied to the inconsistent situation of different users on diversity demands, so that the user experience is poor.
Disclosure of Invention
The invention mainly aims to provide a commodity recommending method, a commodity recommending device, commodity recommending equipment and a commodity recommending storage medium, and aims to solve the technical problems that key parameters used in a recommending diversity improving method in the prior art are specified manually and cannot be suitable for different users, so that recommending accuracy is low.
In order to achieve the above object, the present invention provides a commodity recommendation method, which includes the steps of:
determining user characteristic data and tag data based on the user history data;
establishing a user diversity preference model based on the user characteristic data and the tag data;
determining diversity preference data of a target user based on the user diversity preference model;
determining a breaking-up parameter of a breaking-up stage based on the diversity preference data of the target user;
and recommending the commodity to the target user based on the scattering parameters.
Optionally, the user history data includes user behavior data and user attribute data, the user feature data includes behavior feature data and attribute feature data, and the determining, based on the user history data, the user feature data and the tag data includes:
sequencing the user behavior data according to the time data corresponding to the user behavior data to obtain a sample commodity sequence;
sampling the sample commodity sequence based on a sliding window, and determining window commodities and label commodities corresponding to the sliding window;
converting commodity attribute data of the window commodity into an attribute sequence corresponding to commodity attributes;
determining the behavior characteristic data according to the attribute sequence, and determining the attribute characteristic data according to the user attribute data;
determining an attribute classification tag corresponding to the commodity attribute based on the attribute sequence of the window commodity and commodity attribute data of the tag commodity;
and determining the tag data based on the attribute classification tag corresponding to the commodity attribute.
Optionally, the determining, based on the attribute sequence of the window commodity and the commodity attribute data of the labeled commodity, an attribute classification label corresponding to the commodity attribute includes:
according to the commodity attribute, determining label attribute data corresponding to the commodity attribute in commodity attribute data of the label commodity, and determining a window attribute sequence corresponding to the commodity attribute in attribute sequences of the window commodity;
determining whether the same attribute data exists in a window attribute sequence corresponding to the commodity attribute according to the label attribute data corresponding to the commodity attribute;
when the same attribute data exist, determining that the attribute classification label of the commodity attribute is a negative sample label;
and when the same attribute data does not exist, determining the attribute classification label of the commodity attribute as a positive sample label.
Optionally, the determining diversity preference data of the target user based on the user diversity preference model includes:
acquiring an interactive commodity sequence of the target user;
determining target commodities in the interactive commodity sequence according to the selected quantity, and converting commodity attribute data of the target commodities into a target attribute sequence;
and inputting the target attribute sequence and the user attribute data of the target user into the user diversity preference model to obtain diversity preference data of the target user.
Optionally, the breaking-up parameters include at least a commodity breaking-up score, the breaking-up stage includes a merging multi-way recall result stage and a post-fine rearrangement stage, and the determining the breaking-up parameters of the breaking-up stage based on the diversity preference data of the target user includes:
determining the number of the commodities with the same attribute corresponding to the target commodity according to the target attribute sequence and the target commodity sequence;
acquiring commodity sorting scores of the target commodity in the scattering stage;
acquiring the corresponding relation among diversity preference data, the number of commodities with the same attribute, commodity sorting scores and commodity scattering scores;
and determining the commodity scattering score of the target commodity in the scattering stage according to the diversity preference data, the commodity quantity with the same attribute, the commodity sequencing score and the corresponding relation.
Optionally, the recommending the commodity to the target user based on the scattering parameter includes:
according to the scattering parameters, the target commodities are subjected to descending order sorting in the scattering stage to obtain a scattering commodity sequence;
determining a recommended commodity sequence of the target user based on the scattered commodity sequence;
and recommending the commodity to the target user based on the commodity recommending sequence.
Optionally, the user diversity preference model includes an input layer, an intermediate transformation layer, and a loss function layer, and the establishing the user diversity preference model based on the user feature data and the tag data includes:
inputting the user characteristic data into the input layer for data processing to obtain behavior characteristics and attribute characteristics;
the behavior characteristics and the attribute characteristics are spliced and then input into the intermediate conversion layer to carry out nonlinear conversion, so that training characteristic data are obtained;
and inputting the training feature data and the tag data into the loss function layer for model training to obtain the user diversity preference model.
In addition, in order to achieve the above object, the present invention also provides a commodity recommendation device, including:
the sample construction module is used for determining user characteristic data and label data based on the user history data;
the model building module is used for building a user diversity preference model based on the user characteristic data and the tag data;
the recommendation application module is used for determining diversity preference data of the target user based on the user diversity preference model;
the recommendation application module is further used for determining scattering parameters of a scattering stage based on the diversity preference data of the target user;
and the recommendation application module is also used for recommending the commodity to the target user based on the scattering parameters.
In addition, in order to achieve the above object, the present invention also proposes a commodity recommendation apparatus including: a memory, a processor, and a merchandise recommendation program stored on the memory and executable on the processor, the merchandise recommendation program configured to implement the steps of the merchandise recommendation method as described above.
In addition, in order to achieve the above object, the present invention also proposes a storage medium having stored thereon a commodity recommendation program which, when executed by a processor, implements the steps of the commodity recommendation method as described above.
In the method, user characteristic data and tag data are determined based on user history data, a user diversity preference model is established based on the user characteristic data and the tag data, diversity preference data of a target user are determined based on the user diversity preference model, scattering parameters of a scattering stage are determined based on the diversity preference data of the target user, and commodity recommendation is performed on the target user based on the scattering parameters of the scattering stage. Compared with the mode that the manual specified parameters are recommended, the method and the device have lower precision, the user diversity preference model is built according to the user history behaviors, and the preference degree of the user for diversity is predicted, so that the key parameters of the scattering strategy are automatically updated, the recommendation accuracy can be improved, the modeling takes the user as a dimension, the personalized requirements of the user can be met, in addition, the automatic updating can be performed according to the period, the dynamic change of the user preference can be captured, and the recommendation accuracy is further improved.
Drawings
FIG. 1 is a schematic diagram of a commodity recommendation device in a hardware operating environment according to an embodiment of the present invention;
FIG. 2 is a schematic flow chart of a first embodiment of the merchandise recommendation method of the present invention;
FIG. 3 is a flowchart of a second embodiment of the merchandise recommendation method of the present invention;
FIG. 4 is a flowchart of a third embodiment of a merchandise recommendation method according to the present invention;
fig. 5 is a block diagram showing a first embodiment of the commodity recommendation apparatus according to the present invention.
The achievement of the objects, functional features and advantages of the present invention will be further described with reference to the accompanying drawings, in conjunction with the embodiments.
Detailed Description
It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
Referring to fig. 1, fig. 1 is a schematic structural diagram of a commodity recommendation device in a hardware operation environment according to an embodiment of the present invention.
As shown in fig. 1, the commodity recommendation apparatus may include: a processor 1001, such as a central processing unit (Central Processing Unit, CPU), a communication bus 1002, a user interface 1003, a network interface 1004, a memory 1005. Wherein the communication bus 1002 is used to enable connected communication between these components. The user interface 1003 may include a Display, an input unit such as a Keyboard (Keyboard), and the optional user interface 1003 may further include a standard wired interface, a wireless interface. The network interface 1004 may optionally include a standard wired interface, a Wireless interface (e.g., a Wireless-Fidelity (Wi-Fi) interface). The Memory 1005 may be a high-speed random access Memory (Random Access Memory, RAM) Memory or a stable nonvolatile Memory (NVM), such as a disk Memory. The memory 1005 may also optionally be a storage device separate from the processor 1001 described above.
It will be appreciated by those skilled in the art that the structure shown in fig. 1 is not limiting of the merchandise recommendation apparatus and may include more or fewer components than shown, or certain components in combination, or a different arrangement of components.
As shown in fig. 1, an operating system, a network communication module, a user interface module, and a commodity recommendation program may be included in the memory 1005 as one type of storage medium.
In the merchandise recommendation apparatus shown in fig. 1, the network interface 1004 is mainly used for data communication with a network server; the user interface 1003 is mainly used for data interaction with a user; the processor 1001 and the memory 1005 in the commodity recommendation device of the present invention may be provided in the commodity recommendation device, and the commodity recommendation device calls the commodity recommendation program stored in the memory 1005 through the processor 1001 and executes the commodity recommendation method provided by the embodiment of the present invention.
An embodiment of the invention provides a commodity recommendation method, referring to fig. 2, fig. 2 is a schematic flow chart of a first embodiment of the commodity recommendation method of the invention.
In this embodiment, the commodity recommendation method includes the following steps:
step S10: user characteristic data and tag data are determined based on the user history data.
It should be noted that, the execution body of the embodiment is an intelligent terminal, for example: and the computer is provided with a commodity recommendation program in the intelligent terminal, and commodity recommendation is realized based on diversity preference of users by running the commodity recommendation program.
It is understood that the user history data refers to relevant history data of the user, including user behavior data and user attribute data. The user behavior data, namely historical data related to the user behavior, comprises a user click sequence, namely a set of all clicked commodities of a user and a set of all purchased commodities, wherein the user click sequence comprises a plurality of clicked commodities, namely the commodities clicked by the user, and the user purchase sequence comprises a plurality of purchased commodities, namely the commodities purchased by the user. The user attribute data is data (numerical value or character string) corresponding to the attribute of the user, the attribute of the user comprises a natural attribute and a platform attribute, the natural attribute can be gender, age, region and the like, the platform attribute can be a user identification code (for example, user ID (Identity document)), registration time, new and old user states, registration channels and the like, therefore, the user attribute data comprises the natural attribute data and the platform attribute data, the natural attribute data at least comprises gender data, age data and region data, and the platform attribute data at least comprises user identification data, registration time data, new and old user state data and registration channel data.
It should be appreciated that in order to boost the purchase conversion objective, the commodity purchased by the user may be up-sampled to enhance the weight of the purchase behavior, and the user behavior data in this embodiment has been data enhanced.
The user feature data is feature data required for model training, and is input as a training sample, and the tag data is tag data required for model training, and in this embodiment, the tag data is a classification tag of a commodity attribute.
Further, the user characteristic data in the present embodiment includes two parts: behavior feature data and attribute feature data, i.e., behavior-related features and attribute-related features, typically use user attribute data as attribute feature data, which is further determined according to the user behavior data.
In a specific implementation, the clicked commodity and the purchased commodity are arranged according to time sequence, sampling is performed through a sliding window after arrangement is completed, and the length and the interval of the sliding window can be set according to actual requirements, which is not limited in this embodiment. If the length of the sliding window is n, the number of commodities in the sliding window is also n, wherein each commodity has m types of attributes, the commodities in the sliding window can be converted into m attribute sequences according to the m types of attributes respectively, the attribute identification codes of the n commodities in each attribute sequence are the behavior feature data, and the classification labels of the m commodity attributes are determined according to the attribute sequences.
Step S20: and establishing a user diversity preference model based on the user characteristic data and the tag data.
Further, the step S20 includes: inputting the user characteristic data into the input layer for data processing to obtain behavior characteristics and attribute characteristics; the behavior characteristics and the attribute characteristics are spliced and then input into the intermediate conversion layer to carry out nonlinear conversion, so that training characteristic data are obtained; and inputting the training feature data and the tag data into the loss function layer for model training to obtain the user diversity preference model.
It can be appreciated that the user diversity preference model is a model constructed to determine the diversity preference value of the user, which in this embodiment is for the merchandise attribute. The user diversity preference model includes three parts: an input layer, an intermediate transform layer, and a loss function layer. The input layer contains two types of inputs: the behavior feature data and the attribute feature data generally need to be processed, and the behavior feature and the attribute feature are processed behavior feature data and attribute feature data. The input behavior characteristic data, namely attribute sequences of m commodities, wherein the length of each attribute sequence is n, a search layer corresponding to an identification code and an embedded vector one by one is added to each attribute sequence, the dimension of the embedded vector is d, m matrixes of n x d are obtained, and then a summation layer is connected to obtain m vectors of 1*d as behavior characteristics. The input attribute feature data, namely user attribute data, is used for carrying out category coding on the data of the character string type, normalizing the data of the numerical value type to obtain attribute features, and then splicing the behavior features and the attribute features in the dimension of the row to be used as the input of an intermediate transformation layer. The intermediate transformation layer is composed of k full-connected layers for nonlinear transformation, and the obtained data (training feature data) is used for training at the loss function layer. Since there are m commodity attributes, the number of classifications is m, and the loss function is a multi-classification loss function, and in this embodiment, the loss function layer adopts cross entropy as the loss function.
In a specific implementation, user characteristic data and tag data are input into a model (initial model) for training, so that a final user diversity preference model is obtained, the model is used for predicting the preference degree of users for diversity, commodity recommendation is carried out according to the preference degree of the users for diversity, and the recommendation accuracy can be improved.
Step S30: and determining diversity preference data of the target user based on the user diversity preference model.
It should be understood that the target user refers to a user who needs to make a commodity recommendation, the diversity preference data, i.e., the diversity preference value.
Further, the step S30 includes: acquiring an interactive commodity sequence of the target user; determining target commodities in the interactive commodity sequence according to the selected quantity, and converting commodity attribute data of the target commodities into a target attribute sequence; and inputting the target attribute sequence and the user attribute data of the target user into the user diversity preference model to obtain diversity preference data of the target user.
It should be noted that, the interactive commodity sequence refers to a commodity sequence that the target user recently interacts (clicks and purchases), and the time range can be set according to the actual requirement, which is not limited in this embodiment. The selection quantity is the quantity of commodities selected from the interactive commodity sequence, and because the sliding window is used for sampling during model training, the selection quantity is required to be consistent with the length of the sliding window during model application. The target commodity is selected from the interactive commodity sequence, and if the selected number is n, the most recent n interactive commodities are generally selected. The commodity attribute data of the target commodity is data corresponding to each commodity attribute of the target commodity, and is generally represented by an attribute identification code (for example, attribute ID), and a sequence formed by commodity attribute data corresponding to all commodity attributes is a target attribute sequence.
In the specific implementation, the target attribute sequence of the target user and the user attribute data are used as model input, so that the diversity preference value of the target user on each commodity attribute can be obtained and used for determining key parameters required by scattering.
Step S40: and determining a breaking-up parameter of a breaking-up stage based on the diversity preference data of the target user.
It is understood that the breaking-up phase refers to a phase in which breaking-up processing is required, that is, a phase in which a breaking-up strategy is used, and in this embodiment, the breaking-up phase includes a merging multi-recall result phase and a post-fine rearrangement phase.
It should be understood that the breaking-up parameters are key parameters required for each breaking-up stage, and may also be considered as key parameters in the recommendation process, including at least the product breaking-up score.
Step S50: and recommending the commodity to the target user based on the scattering parameters of the scattering stage.
In a specific implementation, in each breaking stage, the calculated breaking parameters are used to break up the commodity sequence to be broken up, and the commodity sequence is usually ordered according to the breaking parameters.
In this embodiment, user feature data and tag data are determined based on user history data, a user diversity preference model is established based on the user feature data and the tag data, diversity preference data of a target user is determined based on the user diversity preference model, scattering parameters of a scattering stage are determined based on the diversity preference data of the target user, and commodity recommendation is performed on the target user based on the scattering parameters of the scattering stage. According to the method and the device, a user diversity preference model is built according to user historical behaviors, the preference degree of users for diversity is predicted, so that key parameters of a scattering strategy are automatically updated, recommendation accuracy can be improved, the modeling takes users as dimensions, personalized requirements of the users can be met, in addition, automatic updating can be carried out according to periods, dynamic changes of user preferences can be captured, and recommendation accuracy is further improved.
Referring to fig. 3, fig. 3 is a flowchart illustrating a commodity recommendation method according to a second embodiment of the present invention.
Based on the above embodiment, the step S10 includes:
step S101: and sequencing the user behavior data according to the time data corresponding to the user behavior data to obtain a sample commodity sequence.
The time data refers to the behavior transmission time, and the time data corresponding to the user behavior data is the click time corresponding to the clicked commodity and the purchase time corresponding to the purchased commodity. The sorting in this embodiment is a sequence obtained by sorting all the clicked commodities and the purchased commodities according to the time sequence, and the commodities in the sample commodity sequence may be represented by using a commodity identification code (for example, a commodity ID).
In the specific implementation, all the clicked commodities and purchased commodities in the user behavior data are ordered according to the time sequence, so that a sample commodity sequence is formed.
Step S102: and sampling the sample commodity sequence based on the sliding window, and determining window commodities and label commodities corresponding to the sliding window.
It will be appreciated that the length and interval of the sliding window may be set according to practical requirements, and the present embodiment is not limited thereto, for example: length n and spacing w. And sampling by using a sliding window, wherein in the sample commodity sequence, commodities in the sliding window are window commodities, and the later commodity in the sliding window is a label commodity.
Step S103: and converting commodity attribute data of the window commodity into an attribute sequence corresponding to commodity attributes, determining the behavior characteristic data according to the attribute sequence, and determining the attribute characteristic data according to the user attribute data.
It should be understood that the commodity attribute data of the window commodity is a numerical value corresponding to each commodity attribute of the window commodity, and may be represented by using an attribute identification code. The attribute sequence is a sequence formed by commodity attribute data, and when the commodity attribute data is converted, commodity attribute data corresponding to each window commodity is found according to the commodity attribute to form an attribute sequence corresponding to the commodity attribute, for example: and according to the commodity attribute A, finding out the value corresponding to the commodity attribute A of each window commodity, and combining the values into a sequence to obtain an attribute sequence of the commodity attribute A. If there are m commodity attributes in total, an attribute sequence of m commodity attributes can be obtained.
In a specific implementation, the obtained attribute sequence of the m commodity attributes is m behavior feature data, and the attribute feature data is user attribute data.
Step S104: and determining an attribute classification tag corresponding to the commodity attribute based on the attribute sequence of the window commodity and commodity attribute data of the tag commodity.
The article attribute data of the tag article is a value corresponding to each article attribute of the tag article, and may be represented by using an attribute identifier. The attribute classification labels refer to classification of commodity attributes, including positive sample labels, which are typically marked with a "1", and negative sample labels, which are typically marked with a "0".
Further, the step S104 includes: according to the commodity attribute, determining label attribute data corresponding to the commodity attribute in commodity attribute data of the label commodity, and determining a window attribute sequence corresponding to the commodity attribute in attribute sequences of the window commodity; determining whether the same attribute data exists in a window attribute sequence corresponding to the commodity attribute according to the label attribute data corresponding to the commodity attribute; when the same attribute data exist, determining that the attribute classification label of the commodity attribute is a negative sample label; and when the same attribute data does not exist, determining the attribute classification label of the commodity attribute as a positive sample label.
It will be understood that the tag attribute data refers to the article attribute data of the tag article corresponding to each article attribute, and the window attribute sequence refers to the attribute sequence corresponding to each article attribute. For each commodity attribute, it is necessary to determine whether the same data as the corresponding commodity attribute data exists in the corresponding window attribute sequence, if so, the attribute classification label of the commodity attribute is a negative sample label, labeled as "0", and if not, the attribute classification label of the commodity attribute is a positive sample label, labeled as "1".
In a specific implementation, for each attribute, if the attribute ID of the tag commodity is not the same as any attribute ID in the attribute sequence of the attribute, the attribute is a positive sample, marked as "1", otherwise "0", and if there are m commodity attributes, m attribute classification tags can be obtained.
Step S105: and determining the tag data based on the attribute classification tag corresponding to the commodity attribute.
It should be understood that the tag data corresponding to the attribute sequence, that is, the tag data of the behavior feature data, can be obtained after determining all attribute classification tags.
In this embodiment, according to time data corresponding to user behavior data, ordering the user behavior data to obtain a sample commodity sequence, sampling the sample commodity sequence based on a sliding window, determining window commodity and label commodity corresponding to the sliding window, converting commodity attribute data of the window commodity into attribute sequence corresponding to commodity attribute, determining behavior feature data as attribute sequence, determining attribute feature data as user attribute data, determining attribute classification label corresponding to commodity attribute based on the attribute sequence of the window commodity and commodity attribute data of the label commodity, and determining label data based on attribute classification label corresponding to commodity attribute. According to the embodiment, the characteristic data and the tag data are determined according to the historical behaviors of the user, and the characteristic data and the tag data are used for establishing a user diversity preference model so as to predict the diversity preference degree of the user, so that key parameters of a recommendation strategy are automatically updated, the recommendation accuracy can be improved, the modeling takes the user as a dimension, and the personalized requirements of the user can be met.
Referring to fig. 4, fig. 4 is a flowchart illustrating a third embodiment of a commodity recommendation method according to the present invention.
Based on the above embodiment, the step S40 includes:
step S401: and determining the number of the commodities with the same attribute corresponding to the target commodity according to the target attribute sequence and the target commodity sequence.
The target commodity sequence is a sequence formed by the target commodities according to time sequence, and the number of commodities with the same attribute refers to the number of commodities with the same attribute in all commodities in front of each target commodity in the target commodity sequence.
Step S402: and acquiring commodity sorting scores of the target commodities in the scattering stage, and acquiring corresponding relations among diversity preference data, the number of commodities with the same attribute, commodity sorting scores and commodity scattering scores.
It will be appreciated that the item ranking score refers to the user's score for each target item at the ranking stage, with different ranking stages possibly having different scores. The corresponding relation among the diversity preference data, the number of commodities with the same attribute, the commodity sorting score and the commodity scattering score, namely the calculation relation of the commodity scattering score is as follows:
wherein p is j Representing the diversity preference value, w, of the user for the j-th commodity attribute ij Is the number k of the articles with the same attribute j in all articles in front of the ith article in the target article sequence i The score of the user on the commodity i in the sorting stage represents the commodity scattering score of the ith commodity.
Step S403: and determining the commodity scattering score of the target commodity in the scattering stage according to the diversity preference data, the commodity quantity with the same attribute, the commodity sequencing score and the corresponding relation.
It should be understood that, substituting the diversity preference data, the number of commodities with the same attribute, and the commodity sorting score into the calculation relation of the commodity breaking score can obtain the corresponding commodity breaking score.
Further, the step S50 includes:
step S501: and according to the scattering parameters, the target commodity is subjected to descending order sorting in the scattering stage, so that a scattering commodity sequence is obtained.
In the scattering process, the scattered commodity sequences can be obtained by sorting from high to low according to commodity scattering scores.
Step S502: and determining a recommended commodity sequence of the target user based on the scattered commodity sequence, and recommending commodities to the target user based on the recommended commodity sequence.
It can be understood that after being scattered, other processing can be performed, so that the commodity recommended to the target user is finally obtained, the sequence formed by the commodity is a recommended commodity sequence, and the commodity is recommended to the target user according to the recommended commodity sequence.
It should be understood that the interactive commodity sequence of the target user can be automatically acquired according to the period, the diversity preference data can be automatically updated, the key parameters of the scattering strategy can be automatically updated, the dynamic change of the user preference can be captured, and the recommendation accuracy is improved.
In this embodiment, the number of the same attribute commodities corresponding to the target commodity is determined according to the target attribute sequence, the commodity sorting score of the target commodity in the scattering stage is obtained, the corresponding relationship between the diversity preference data, the number of the same attribute commodities, the commodity sorting score and the commodity scattering score is obtained, and the commodity scattering score of the target commodity in the scattering stage is determined according to the diversity preference data, the number of the same attribute commodities, the commodity sorting score and the corresponding relationship. According to the embodiment, a user diversity preference model is built according to the user history behaviors, the preference degree of the user for diversity is predicted, so that the scattering parameters are automatically updated in real time, the preference of the user for recommending diversity can be captured in real time, and the user experience and recommendation accuracy are improved.
In addition, the embodiment of the invention also provides a storage medium, wherein the storage medium stores a commodity recommendation program, and the commodity recommendation program realizes the steps of the commodity recommendation method when being executed by a processor.
Referring to fig. 5, fig. 5 is a block diagram showing a first embodiment of the commodity recommendation device according to the present invention.
As shown in fig. 5, the commodity recommendation device provided in the embodiment of the present invention includes:
the sample construction module 10 is configured to determine user characteristic data and tag data based on user history data.
The model building module 20 is configured to build a user diversity preference model based on the user feature data and the tag data.
The recommendation application module 30 is configured to determine diversity preference data of the target user based on the user diversity preference model.
The recommendation application module 30 is further configured to determine a break-up parameter of a break-up stage based on the diversity preference data of the target user.
The recommendation application module 30 is further configured to recommend the commodity to the target user based on the scattering parameter.
In this embodiment, user feature data and tag data are determined based on user history data, a user diversity preference model is established based on the user feature data and the tag data, diversity preference data of a target user is determined based on the user diversity preference model, scattering parameters of a scattering stage are determined based on the diversity preference data of the target user, and commodity recommendation is performed on the target user based on the scattering parameters of the scattering stage. According to the method and the device, a user diversity preference model is built according to user historical behaviors, the preference degree of users for diversity is predicted, so that key parameters of a scattering strategy are automatically updated, recommendation accuracy can be improved, the modeling takes users as dimensions, personalized requirements of the users can be met, in addition, automatic updating can be carried out according to periods, dynamic changes of user preferences can be captured, and recommendation accuracy is further improved.
In an embodiment, the user history data includes user behavior data and user attribute data, the user feature data includes behavior feature data and attribute feature data, and the sample construction module 10 is further configured to sort the user behavior data according to time data corresponding to the user behavior data, so as to obtain a sample commodity sequence;
sampling the sample commodity sequence based on a sliding window, and determining window commodities and label commodities corresponding to the sliding window;
converting commodity attribute data of the window commodity into an attribute sequence corresponding to commodity attributes, determining the behavior characteristic data as the attribute sequence, and determining the attribute characteristic data as the user attribute data;
determining an attribute classification tag corresponding to the commodity attribute based on the attribute sequence of the window commodity and commodity attribute data of the tag commodity;
and determining the tag data based on the attribute classification tag corresponding to the commodity attribute.
In an embodiment, the sample construction module 10 is further configured to determine, according to the commodity attribute, tag attribute data corresponding to the commodity attribute from commodity attribute data of the tag commodity, and determine a window attribute sequence corresponding to the commodity attribute from attribute sequences of the window commodity;
tag attribute data corresponding to the commodity attributes are used for determining whether the same attribute data exists in a window attribute sequence corresponding to the commodity attributes;
when the same attribute data exist, determining that the attribute classification label of the commodity attribute is a negative sample label;
and when the same attribute data does not exist, determining the attribute classification label of the commodity attribute as a positive sample label.
In an embodiment, the recommendation application module 30 is further configured to obtain an interactive merchandise sequence of the target user;
determining target commodities in the interactive commodity sequence according to the selected quantity, and converting commodity attribute data of the target commodities into a target attribute sequence;
and inputting the target attribute sequence and the user attribute data of the target user into the user diversity preference model to obtain diversity preference data of the target user.
In an embodiment, the recommendation application module 30 is further configured to determine, according to the target attribute sequence and the target commodity sequence, the number of commodities with the same attribute corresponding to the target commodity;
acquiring commodity sorting scores of the target commodity in the scattering stage;
acquiring the corresponding relation among diversity preference data, the number of commodities with the same attribute, commodity sorting scores and commodity scattering scores;
and determining the commodity scattering score of the target commodity in the scattering stage according to the diversity preference data, the commodity quantity with the same attribute, the commodity sequencing score and the corresponding relation.
In an embodiment, the recommendation application module 30 is further configured to sort the target commodities in a descending order in the scattering stage according to the scattering parameter, so as to obtain a scattered commodity sequence;
determining a recommended commodity sequence of the target user based on the scattered commodity sequence;
and recommending the commodity to the target user based on the commodity recommending sequence.
In an embodiment, the model building module 20 is further configured to input the user feature data into the input layer for data processing, so as to obtain a behavior feature and an attribute feature;
the behavior characteristics and the attribute characteristics are spliced and then input into the intermediate conversion layer to carry out nonlinear conversion, so that training characteristic data are obtained;
and inputting the training feature data and the tag data into the loss function layer for model training to obtain the user diversity preference model.
It should be understood that the foregoing is illustrative only and is not limiting, and that in specific applications, those skilled in the art may set the invention as desired, and the invention is not limited thereto.
It should be noted that the above-described working procedure is merely illustrative, and does not limit the scope of the present invention, and in practical application, a person skilled in the art may select part or all of them according to actual needs to achieve the purpose of the embodiment, which is not limited herein.
In addition, technical details not described in detail in the present embodiment may refer to the commodity recommendation method provided in any embodiment of the present invention, and are not described herein.
Furthermore, it should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or system. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or system that comprises the element.
The foregoing embodiment numbers of the present invention are merely for the purpose of description, and do not represent the advantages or disadvantages of the embodiments.
From the above description of the embodiments, it will be clear to those skilled in the art that the above-described embodiment method may be implemented by means of software plus a necessary general hardware platform, but of course may also be implemented by means of hardware, but in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art in the form of a software product stored in a storage medium (e.g. Read Only Memory)/RAM, magnetic disk, optical disk) and including several instructions for causing a terminal device (which may be a mobile phone, a computer, a server, or a network device, etc.) to perform the method according to the embodiments of the present invention.
The foregoing description is only of the preferred embodiments of the present invention, and is not intended to limit the scope of the invention, but rather is intended to cover any equivalents of the structures or equivalent processes disclosed herein or in the alternative, which may be employed directly or indirectly in other related arts.
Claims (10)
1. A commodity recommendation method, characterized in that the commodity recommendation method comprises:
determining user characteristic data and tag data based on the user history data;
establishing a user diversity preference model based on the user characteristic data and the tag data;
determining diversity preference data of a target user based on the user diversity preference model;
determining a breaking-up parameter of a breaking-up stage based on the diversity preference data of the target user;
and recommending the commodity to the target user based on the scattering parameters.
2. The method of claim 1, wherein the user history data comprises user behavior data and user attribute data, the user feature data comprises behavior feature data and attribute feature data, and the determining user feature data and tag data based on the user history data comprises:
sequencing the user behavior data according to the time data corresponding to the user behavior data to obtain a sample commodity sequence;
sampling the sample commodity sequence based on a sliding window, and determining window commodities and label commodities corresponding to the sliding window;
converting commodity attribute data of the window commodity into an attribute sequence corresponding to commodity attributes;
determining the behavior characteristic data according to the attribute sequence, and determining the attribute characteristic data according to the user attribute data;
determining an attribute classification tag corresponding to the commodity attribute based on the attribute sequence of the window commodity and commodity attribute data of the tag commodity;
and determining the tag data based on the attribute classification tag corresponding to the commodity attribute.
3. The method of claim 2, wherein the determining the attribute classification tag corresponding to the commodity attribute based on the sequence of attributes of the window commodity and commodity attribute data of the tagged commodity comprises:
according to the commodity attribute, determining label attribute data corresponding to the commodity attribute in commodity attribute data of the label commodity, and determining a window attribute sequence corresponding to the commodity attribute in attribute sequences of the window commodity;
determining whether the same attribute data exists in a window attribute sequence corresponding to the commodity attribute according to the label attribute data corresponding to the commodity attribute;
when the same attribute data exist, determining that the attribute classification label of the commodity attribute is a negative sample label;
and when the same attribute data does not exist, determining the attribute classification label of the commodity attribute as a positive sample label.
4. The method of claim 1, wherein the determining diversity preference data for the target user based on the user diversity preference model comprises:
acquiring an interactive commodity sequence of the target user;
determining target commodities in the interactive commodity sequence according to the selected quantity, and converting commodity attribute data of the target commodities into a target attribute sequence;
and inputting the target attribute sequence and the user attribute data of the target user into the user diversity preference model to obtain diversity preference data of the target user.
5. The method of claim 4, wherein the break-up parameters include at least a commodity break-up score, the break-up stage includes a merge multi-recall result stage and a post-fine rearrangement stage, the determining the break-up parameters of the break-up stage based on the diversity preference data of the target user includes:
determining the number of the commodities with the same attribute corresponding to the target commodity according to the target attribute sequence and the target commodity sequence;
acquiring commodity sorting scores of the target commodity in the scattering stage;
acquiring the corresponding relation among diversity preference data, the number of commodities with the same attribute, commodity sorting scores and commodity scattering scores;
and determining the commodity scattering score of the target commodity in the scattering stage according to the diversity preference data, the commodity quantity with the same attribute, the commodity sequencing score and the corresponding relation.
6. The method of claim 5, wherein the recommending goods to the target user based on the break-up parameters comprises:
according to the scattering parameters, the target commodities are subjected to descending order sorting in the scattering stage to obtain a scattering commodity sequence;
determining a recommended commodity sequence of the target user based on the scattered commodity sequence;
and recommending the commodity to the target user based on the commodity recommending sequence.
7. The method of any one of claims 1 to 6, wherein the user diversity preference model includes an input layer, an intermediate transformation layer, and a loss function layer, the building a user diversity preference model based on the user characteristic data and the tag data, comprising:
inputting the user characteristic data into the input layer for data processing to obtain behavior characteristics and attribute characteristics;
the behavior characteristics and the attribute characteristics are spliced and then input into the intermediate conversion layer to carry out nonlinear conversion, so that training characteristic data are obtained;
and inputting the training feature data and the tag data into the loss function layer for model training to obtain the user diversity preference model.
8. A commodity recommendation device, characterized in that the commodity recommendation device comprises:
the sample construction module is used for determining user characteristic data and label data based on the user history data;
the model building module is used for building a user diversity preference model based on the user characteristic data and the tag data;
the recommendation application module is used for determining diversity preference data of the target user based on the user diversity preference model;
the recommendation application module is further used for determining scattering parameters of a scattering stage based on the diversity preference data of the target user;
and the recommendation application module is also used for recommending the commodity to the target user based on the scattering parameters.
9. A commodity recommendation device, the device comprising: a memory, a processor and a commodity recommendation program stored on the memory and executable on the processor, the commodity recommendation program configured to implement the steps of the commodity recommendation method according to any one of claims 1 to 7.
10. A storage medium having stored thereon a commodity recommendation program which when executed by a processor performs the steps of the commodity recommendation method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311110267.1A CN117057886A (en) | 2023-08-30 | 2023-08-30 | Commodity recommendation method, commodity recommendation device, commodity recommendation equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311110267.1A CN117057886A (en) | 2023-08-30 | 2023-08-30 | Commodity recommendation method, commodity recommendation device, commodity recommendation equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117057886A true CN117057886A (en) | 2023-11-14 |
Family
ID=88653458
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311110267.1A Pending CN117057886A (en) | 2023-08-30 | 2023-08-30 | Commodity recommendation method, commodity recommendation device, commodity recommendation equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117057886A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117273871A (en) * | 2023-11-23 | 2023-12-22 | 深圳市铱云云计算有限公司 | High-quality commodity recommendation system and method based on big data |
-
2023
- 2023-08-30 CN CN202311110267.1A patent/CN117057886A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117273871A (en) * | 2023-11-23 | 2023-12-22 | 深圳市铱云云计算有限公司 | High-quality commodity recommendation system and method based on big data |
| CN117273871B (en) * | 2023-11-23 | 2024-03-08 | 深圳市铱云云计算有限公司 | High-quality commodity recommendation system and method based on big data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111444428A (en) | Information recommendation method and device based on artificial intelligence, electronic equipment and storage medium | |
| CN109582876B (en) | Tourist industry user portrait construction method and device and computer equipment | |
| JP7043786B2 (en) | Sales activity support system, sales activity support method and sales activity support program | |
| CN111144723A (en) | Method and system for recommending people's job matching and storage medium | |
| CN109816483B (en) | Information recommendation method and device and readable storage medium | |
| CN110647696B (en) | Business object sorting method and device | |
| WO2006102381A2 (en) | Constrained tree structure method and system | |
| CN111105209A (en) | Job resume matching method and device suitable for post matching recommendation system | |
| CN111950593A (en) | Method and device for recommending model training | |
| CN110852785B (en) | User grading method, device and computer readable storage medium | |
| CN110288350A (en) | User's Value Prediction Methods, device, equipment and storage medium | |
| CN110598120A (en) | Behavior data based financing recommendation method, device and equipment | |
| CN117057886A (en) | Commodity recommendation method, commodity recommendation device, commodity recommendation equipment and storage medium | |
| CN111695024A (en) | Object evaluation value prediction method and system, and recommendation method and system | |
| CN114861050A (en) | Feature fusion recommendation method and system based on neural network | |
| CN112487284A (en) | Bank customer portrait generation method, equipment, storage medium and device | |
| CN115619448A (en) | User loss prediction method and device, computer equipment and storage medium | |
| CN115587739A (en) | Client list distribution method and device, computer equipment and storage medium | |
| CN113032676B (en) | Recommendation method and system based on micro-feedback | |
| CN116205700A (en) | Recommendation method and device for target product, computer equipment and storage medium | |
| CN115496420A (en) | New user quality evaluation method and device, computer equipment and storage medium | |
| CN115587830A (en) | Work task excitation method and device, computer equipment and storage medium | |
| CN116823303A (en) | Sales data analysis method, device, equipment and storage medium based on big data | |
| CN112313679A (en) | Information processing apparatus, information processing method, and program | |
| US20180108045A1 (en) | Offer selection using sequential selection operations |
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 |