CN113222687A - Deep learning-based recommendation method and device - Google Patents
Deep learning-based recommendation method and device Download PDFInfo
- Publication number
- CN113222687A CN113222687A CN202110433760.1A CN202110433760A CN113222687A CN 113222687 A CN113222687 A CN 113222687A CN 202110433760 A CN202110433760 A CN 202110433760A CN 113222687 A CN113222687 A CN 113222687A
- Authority
- CN
- China
- Prior art keywords
- target user
- user
- recommendation
- commodities
- commodity
- 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
- 238000013135 deep learning Methods 0.000 title claims abstract description 48
- 238000000034 method Methods 0.000 title claims abstract description 44
- 238000012216 screening Methods 0.000 claims abstract description 23
- 238000012545 processing Methods 0.000 claims abstract description 18
- 238000013136 deep learning model Methods 0.000 claims abstract description 14
- 238000000605 extraction Methods 0.000 claims abstract description 13
- 230000006399 behavior Effects 0.000 claims description 34
- 238000013507 mapping Methods 0.000 claims description 10
- 238000011156 evaluation Methods 0.000 claims description 9
- 238000004590 computer program Methods 0.000 claims description 8
- 238000010276 construction Methods 0.000 claims description 3
- 238000012958 reprocessing Methods 0.000 claims 1
- 230000009977 dual effect Effects 0.000 abstract description 3
- 230000036039 immunity Effects 0.000 abstract description 3
- 238000010586 diagram Methods 0.000 description 5
- 238000001179 sorption measurement Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 2
- 238000005065 mining Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 239000004615 ingredient Substances 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 235000016709 nutrition Nutrition 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0631—Item recommendations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Biophysics (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Mathematical Physics (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a deep learning-based recommendation method, which comprises the steps of obtaining a plurality of user figures and a plurality of commodity attributes, and locking a target user according to the user figures; extracting the display characteristics of the target user and the display characteristics of the plurality of commodity attributes, and then processing to generate a recommendation list; learning the hidden features of the target user and the hidden features of the plurality of commodity attributes by using a deep learning model and predicting the scores of the target user on the recommendation list according to the hidden features; and pre-recommending the target user by the commodities in the recommendation list, determining whether the target user receives the recommendation, and if so, recommending the relevant information of the commodities according to the grade of the recommendation list and the priority. The invention realizes multidimensional personalized screening, enables the user to independently select whether to receive the push or not, carries out effective and accurate push, solves the dual requirements of the user and a merchant, and has better noise immunity and effectiveness based on deep learning feature extraction.
Description
Technical Field
The invention relates to the technical field of recommendation, in particular to a deep learning-based recommendation method and device.
Background
It is proposed in the prior art to classify these models based on the form of input (methods using content information and methods without content information) and output (rating and ranking). However, with the continuous emergence of new research results, such a classification framework is no longer applicable, and a new inclusive framework is needed to better understand the research field, in the prior art, the similarity between the target user and the approximate user is calculated, the approximate user with higher similarity is determined as a recommendation direction, and then the item is recommended to the target user through the preference of the approximate user, which is a recommendation method based on useless content information, and a recommendation method based on content information is conceivable, namely, recommendation is recommended through the personal preference of the user. However, recommended objects are not screened, and whether the user receives the recommendations is not considered, which are reasons for low recommendation efficiency, and even the user needs are not considered really, so that potential needs of the user are mined, and recommendation with high accuracy and high satisfaction cannot be achieved.
Disclosure of Invention
The invention provides a deep learning-based recommendation method, and aims to solve the problems that marketing push mostly depends on short message and mail push in the prior art, if a user sets short message interception and junk mail judgment, the push content cannot be well seen and converted by the user, the user cannot choose not to receive the information, the user cannot be accurately screened for push, the requirements of the user and a merchant cannot be met, and the user experience is poor.
In order to achieve the purpose, the invention adopts the following technical scheme:
the invention relates to a deep learning-based recommendation method, which comprises the following steps of:
acquiring a plurality of user figures and a plurality of commodity attributes, and locking a target user according to the user figures;
extracting the display characteristics of the target user and the display characteristics of the plurality of commodity attributes, and then processing to generate a recommendation list;
learning the hidden features of the target user and the hidden features of the plurality of commodity attributes by using a deep learning model, and predicting the scores of the target user on the recommendation list according to the hidden features;
and pre-recommending the target user by the commodities in the recommendation list, determining whether the target user receives the recommendation, and if so, recommending the relevant information of the commodities according to the grade of the recommendation list and the priority.
Obtaining a plurality of user figures, wherein the user figures comprise behavior characteristics and preference characteristics of users, and locking the user figures according to a direction to be recommended and then locking a target user; extracting the display features of the target user and the display features of the plurality of commodity attributes, wherein the display features are visually represented, the features which do not need secondary mining can visually reflect the characteristics of the user and the characteristics of the commodities, and a recommendation list is generated according to the display features; then, learning the hidden features of the target user and the hidden features of the plurality of commodity attributes by using a deep learning model, and predicting the scores of the target user on the recommendation list according to the hidden features, wherein the hidden features are data which need to be developed secondarily, cannot visually reflect results, and need to be calculated; and pre-recommending the target user by the commodities in the recommendation list, determining whether the target user receives the recommendation, and if so, recommending the relevant information of the commodities according to the grade of the recommendation list and the priority.
Preferably, the acquiring a plurality of user figures and a plurality of merchandise attributes and locking a target user according to the plurality of user figures comprises:
the method comprises the steps of obtaining a plurality of user portraits and a plurality of commodity attributes, wherein the user portraits comprise behavior features and preference features, and the commodity attributes comprise basic commodity attributes and commodity evaluation;
and setting a multidimensional screening item according to the behavior characteristic and the preference characteristic, and locking a target user according to the multidimensional screening item.
Preferably, the extracting the apparent features of the target user and the apparent features of the plurality of commodity attributes, and then processing to generate the recommendation list includes:
acquiring behavior characteristics of the target user, browsing records and searching records of the plurality of commodities, and constructing a preference model of the target user;
simultaneously acquiring the display characteristics of the attributes of the commodities, and searching for similar commodities according to the display characteristics;
and caching the commodities which have mapping relation with the preference model in the approximate commodities into a recommendation list.
Preferably, the learning, by using a deep learning model, the hidden features of the target user and the hidden features of the plurality of product attributes, and predicting the rating of the target user on the recommendation list according to the hidden features includes:
obtaining comment data of the target user on the plurality of commodity attributes;
constructing K dimensional features according to the comment data of the target user on the plurality of commodity attributes, and determining K neighbor users of the target user on the K dimensional features according to the K dimensional features, wherein K is an integer greater than 1;
and obtaining comment data of the K neighbor users on the plurality of commodity attributes, and predicting the scores of the target user on the recommendation list according to the comment data of the K neighbor users on the plurality of commodity attributes.
A deep learning based recommendation device comprising:
an acquisition module: the system comprises a plurality of user portraits and a plurality of commodity attributes, and a target user is locked according to the user portraits;
an extraction module: the system is used for extracting the display features of the target user and the display features of the plurality of commodity attributes, which are locked by the acquisition module, and then processing the display features to generate a recommendation list;
a scoring module: the system comprises an extraction module, a deep learning module and a recommendation module, wherein the extraction module is used for extracting a recommendation list of a target user from a plurality of commodity attributes, and the recommendation list is used for carrying out deep learning on hidden features of the target user and hidden features of the commodity attributes by using the deep learning module;
a recommendation module: and the system is used for pre-recommending the target user by the commodities in the recommendation list, determining whether the target user receives the recommendation, and recommending the related information of the commodities according to the grade of the recommendation list and the priority if the target user receives the recommendation.
Preferably, the acquiring module specifically includes:
a first acquisition unit: the system comprises a plurality of user portraits and a plurality of commodity attributes, wherein the user portraits comprise behavior characteristics and preference characteristics, and the commodity attributes comprise basic commodity attributes and commodity evaluation;
a locking unit: the multi-dimensional screening item is set according to the behavior characteristics and the preference characteristics acquired by the first acquisition unit, and a target user is locked according to the multi-dimensional screening item.
Preferably, the acquisition module specifically includes:
a first acquisition unit: the preference model is used for acquiring the behavior characteristics of the target user, browsing records and searching records of the commodities and constructing a preference model of the target user;
a second acquisition unit: the system is used for simultaneously acquiring the display characteristics of the attributes of the commodities and searching for similar commodities according to the display characteristics;
a cache unit: and the commodity buffer module is used for caching commodities in the approximate commodities and having a mapping relation with the preference model into a recommendation list.
Preferably, the scoring module specifically includes:
a second acquisition unit: the comment data of the target user on the plurality of commodity attributes are acquired;
a construction unit: the system comprises a data acquisition module, a data processing module and a data processing module, wherein the data acquisition module is used for acquiring comment data of a target user on a plurality of commodity attributes, constructing K dimension characteristics according to the comment data of the target user on the plurality of commodity attributes, and determining K neighbor users of the target user on the K dimension characteristics according to the K dimension characteristics, wherein K is an integer greater than 1;
a scoring subunit: the system is used for acquiring comment data of the K neighbor users on the plurality of commodity attributes and predicting the scores of the target user on the recommendation list according to the comment data of the K neighbor users on the plurality of commodity attributes.
An electronic device comprising a memory and a processor, the memory for storing one or more computer instructions, wherein the one or more computer instructions are executed by the processor to implement a deep learning based recommendation method as claimed in any one of the preceding claims.
A computer-readable storage medium storing a computer program which, when executed, causes a computer to implement a deep learning based recommendation method as claimed in any one of the preceding claims.
The invention has the following beneficial effects:
the method and the system realize multidimensional personalized screening, enable the user to independently select whether to receive the pushing through the network platform, effectively and accurately push the marketing content, solve the dual requirements of the user and the merchant, and have better noise immunity and effectiveness based on the feature extraction of deep learning.
Drawings
FIG. 1 is a first flowchart of a method for implementing deep learning based recommendation according to an embodiment of the present invention;
FIG. 2 is a second flowchart of a method for implementing deep learning based recommendation according to an embodiment of the present invention;
FIG. 3 is a third flowchart of a recommendation method based on deep learning according to an embodiment of the present invention
FIG. 4 is a fourth flowchart of a deep learning-based recommendation method according to an embodiment of the present invention
FIG. 5 is a flowchart illustrating an embodiment of a recommendation method based on deep learning according to the present invention;
FIG. 6 is a schematic diagram of a deep learning-based recommendation apparatus according to an embodiment of the present invention;
FIG. 7 is a schematic diagram of an obtaining module of a deep learning-based recommendation apparatus according to an embodiment of the present invention;
FIG. 8 is a schematic diagram of an acquisition module of a deep learning based recommendation device according to an embodiment of the present invention;
FIG. 9 is a schematic diagram of a scoring module of a deep learning-based recommendation device according to an embodiment of the present invention;
fig. 10 is a schematic diagram of an electronic device implementing a deep learning based recommendation apparatus according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, rather than all embodiments, and all other embodiments obtained by those skilled in the art without any inventive work based on the embodiments of the present invention belong to the protection scope of the present invention.
The terms "first," "second," and the like in the claims and in the description of the present application are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order, it being understood that the terms so used are interchangeable under appropriate circumstances and are merely used to describe a distinguishing manner between similar elements in the embodiments of the present application and that the terms "comprising" and "having" and any variations thereof are intended to cover a non-exclusive inclusion such that a process, method, system, article, or apparatus that comprises a list of elements is not necessarily limited to those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs, and the terms used herein in the specification of the present application are for the purpose of describing particular embodiments only and are not intended to limit the present application.
Example 1
As shown in fig. 1, a deep learning based recommendation method includes the following steps:
s110, acquiring a plurality of user figures and a plurality of commodity attributes, and locking a target user according to the user figures;
s120, extracting the display features of the target user and the display features of the commodity attributes, and then processing to generate a recommendation list;
s130, learning the hidden features of the target user and the hidden features of the plurality of commodity attributes by using a deep learning model, and predicting the scores of the target user on the recommendation list according to the hidden features;
s140, pre-recommending the target user by the commodities in the recommendation list, determining whether the target user receives recommendation, and if so, recommending the relevant information of the commodities according to the grade of the recommendation list and the priority.
In embodiment 1, a plurality of user figures are obtained, wherein the user figures comprise preference characteristics and behavior characteristics of users, and the user figures are locked according to a direction to be recommended and then a target user is locked; extracting the display features of the target user and the display features of the commodity attributes according to a deep learning model, wherein the display features are visually represented, the features which do not need secondary mining can visually reflect the characteristics of the user and the characteristics of the commodity, and a recommendation list is generated according to the display features; learning the hidden features of the target user and the hidden features of the plurality of commodity attributes by using a neural network model, and predicting the scores of the target user on the recommendation list according to the hidden features, wherein the hidden features are data which need to be developed secondarily, cannot visually reflect results, and need to be calculated; and pre-recommending the target user by the commodities in the recommendation list, determining whether the target user receives the recommendation, and if so, recommending the relevant information of the commodities according to the grade of the recommendation list and the priority. The method and the system realize multidimensional personalized screening, enable the user to independently select whether to receive the pushing through the network platform, effectively and accurately push the marketing content, solve the dual requirements of the user and the merchant, and have better noise immunity and effectiveness based on the feature extraction of deep learning.
Example 2
As shown in fig. 2, a deep learning based recommendation method includes:
s210, obtaining a plurality of user portraits and a plurality of commodity attributes, wherein the user portraits comprise behavior characteristics and preference characteristics, and the commodity attributes comprise basic commodity attributes and commodity evaluation;
s220, setting a multidimensional screening item according to the behavior characteristic and the preference characteristic, and locking a target user according to the multidimensional screening item;
s230, extracting the display features of the target user and the display features of the commodity attributes, and then processing to generate a recommendation list;
s240, learning the hidden features of the target user and the hidden features of the plurality of commodity attributes by using a deep learning model, and predicting the scores of the target user on the recommendation list according to the hidden features;
s250, pre-recommending the target user by the commodities in the recommendation list, determining whether the target user receives recommendation, and if so, recommending the relevant information of the commodities according to the grade of the recommendation list and the priority.
As can be seen from embodiment 2, a plurality of user figures and a plurality of commodity attributes are obtained, each user figure includes behavior characteristics and preference characteristics, and also includes static characteristics, each static characteristic includes occupation, age, gender, and the like of a user, each behavior characteristic includes accumulated consumption amount, average consumption amount, accumulated consumption times, unconsumed duration, and the like, each preference characteristic refers to commodity preference including consumed commodities and the like, and each screening item has a multi-dimensional setting item including a time dimension, an amount interval, a commodity range, and the like; and accurately screening target users who want to push according to the characteristics, for example, users who want to push with stronger consumption capacity, setting screening options: occupation, accumulated sum of consumption, average sum of consumption; and for another example, pushing the user who just gets mom, setting a screening option: gender, age, amount differentiation, commodity range, time dimension, etc.; for example, the student party is pushed, and then screening items are set: occupation, age, time dimension, etc.; based on the pushing method with content, the target is pushed to the user, so that the re-recommending efficiency is half of that of the successful re-recommending, the accurate pushing requirement of a merchant is met, and a premise is created for the user to recommend highly satisfactory commodities in subsequent steps.
Example 3
As shown in fig. 3, a deep learning based recommendation method includes:
s310, acquiring a plurality of user figures and a plurality of commodity attributes, and locking a target user according to the user figures;
s320, collecting behavior characteristics of the target user, browsing records and searching records of the commodities, and constructing a preference model of the target user;
s330, simultaneously collecting the display characteristics of the attributes of the commodities, and searching for similar commodities according to the display characteristics;
s340, caching the commodities which have a mapping relation with the preference model in the approximate commodities into a recommendation list;
s350, learning the hidden features of the target user and the hidden features of the commodity attributes by using a deep learning model, and predicting the scores of the target user on the recommendation list according to the hidden features;
s360, pre-recommending the target user by the commodities in the recommendation list, determining whether the target user receives recommendation, and if so, recommending the relevant information of the commodities according to the grade of the recommendation list and the priority.
In embodiment 3, behavior characteristics of the target user, browsing records and search records of the plurality of commodities are collected, and a preference model of the target user is directly constructed, where the preference model includes a preference that directly reflects commodities that the target user has purchased or does not purchase a desire to purchase. And simultaneously acquiring the display characteristics of the attributes of the commodities, wherein the display characteristics directly reflect the characteristics of the commodities, such as commodity types, commodity speakers, commodity nutritional ingredients and the like, searching for similar commodities according to the display characteristics, classifying the commodities according to a certain characteristic to form a connection map of the commodities, and caching the commodities with a mapping relation between the similar commodities and the preference model into a recommendation list.
Example 4
As shown in fig. 4, a deep learning based recommendation method includes:
s410, acquiring a plurality of user figures and a plurality of commodity attributes, and locking a target user according to the user figures;
s420, extracting the display features of the target user and the display features of the commodity attributes, and then processing to generate a recommendation list;
s430, obtaining comment data of the target user on the plurality of commodity attributes;
s440, constructing K dimensional features according to the comment data of the target user on the plurality of commodity attributes, and determining K neighbor users of the target user on the K dimensional features according to the K dimensional features, wherein K is an integer greater than 1;
s450, obtaining comment data of the K neighbor users on the plurality of commodity attributes, and predicting the scores of the target user on the recommendation list according to the comment data of the K neighbor users on the plurality of commodity attributes;
s460, pre-recommending the target user by the commodities in the recommendation list, determining whether the target user receives recommendation, and if so, recommending the relevant information of the commodities according to the grade of the recommendation list and the priority.
In embodiment 4, all comment data of a target user on multiple product attributes are obtained, where the multiple products have a comment that the target user purchased and a comment that the user did not purchase, and the ratio of the two exists is not limited, and all comment data includes comments of the target user on a single purchased product and comments of the target user on a single unpurchased product. For example, a user has bought a certain piece of paper and commented under the paper that "this piece of paper is not good, but not good as" this piece of paper has poor adsorption, "this piece of paper is not suitable for a baby," etc., and a mapping relationship with the comment is established based on visible attributes of a plurality of commodities and hidden features of a plurality of commodities, so that the pieces of paper of "certain", "good adsorption piece of paper", "suitable for a baby" are added to a recommendation list, wherein the visible attributes visually represent features such as price, shape, color, etc., and the hidden features represent features of little concern such as adsorption, what materials to use, etc. The preference model also records the behavior characteristics of the user, such as consumption behavior, for example, if the consumption ability is strong, the behavior characteristics of the target user are defined as high price and high quality, if the consumption ability is strong, the behavior characteristics of the target user are defined as moderate quality and low price, and for example, if the target user has special preference for a certain type of commodities, such as sports commodities, the behavior characteristics of the user are defined as sports. Meanwhile, comment data of users except the target user on the plurality of commodity attributes are collected, comment data of other users on the plurality of commodity attributes form scores for a certain commodity, and a commodity feature model is formed by combining the plurality of commodity attributes, for example, the feature of a piece of paper is defined as 'higher price, good adsorptivity and applicable to a baby', the scores are evaluations left by other users who use the paper or evaluations left by other users who do not use the paper but know the paper, so that commodities recommended to the target user are not only commodities possibly required by the user but also commodities with higher scores.
And inputting training data by using a deep learning model, and training a deep network by using an unsupervised learning method, wherein the deep model has four layers including a visible layer v1, a hidden layer h1, a hidden layer h2 and a hidden layer h3, and the top layer and the hidden layer h3 form an undirected associative memory layer. After deep mapping is performed on data through a depth structure, the original M dimension (M is the dimension of input data) is changed into the topmost K dimension (K is the dimension of the topmost neuron of a depth model), and therefore the original high-dimensional data is considered to be mapped through a feature detection group, and the features implicit in the data are mapped into a K-dimensional space. For example, for the commodity scoring data, the mapped K-dimensional space can be understood as the preference characteristics of the user to M commodities, 1-K dimensions represent the degree of the user who likes a certain component in the commodity, the degree of the user who likes a certain speaker, the degree of the user who likes a certain type of commodity, and the like, respectively, and clustering or similarity comparison is performed on the data of each user in the K dimensions, so that the accuracy of the similarity comparison result is higher than that of the similarity comparison result performed by using the original data.
Example 5
As shown in fig. 5, one specific embodiment may be:
s510, obtaining a plurality of user figures and a plurality of commodity attributes, and locking a target user according to the user figures;
in a management background of super-members of a merchant, a member accurate marketing function is added, a merchant can select a designated member or screen consumption behaviors and commodity preferences of the member, the consumption behaviors comprise accumulated consumption amount, average consumption amount, accumulated consumption times, unconsumed duration and the like, the commodity preferences comprise consumed commodities and the like, and each screening item has a multi-dimensional setting item comprising a time dimension, an amount interval, a commodity range and the like.
S520, extracting the display features of the target user and the display features of the commodity attributes, and then processing to generate a recommendation list;
s530, learning the hidden features of the target user and the hidden features of the plurality of commodity attributes by using a deep learning model, and predicting the scores of the target user on the recommendation list according to the hidden features;
s540, pre-recommending the target user by the commodities in the recommendation list, determining whether the target user receives recommendation, and if so, recommending the relevant information of the commodities according to the grade of the recommendation list and the priority.
Because the basic information, consumption behaviors and commodity preferences of the members are stored data based on the digital business super system, the target members can be inquired after the system is screened, but the users need to push and touch the users and actively select to receive messages on the network platform, and the users can decide by themselves according to the stipulation of the network platform, for example, the small program messages on the WeChat platform can be received only once or for a long time by the users.
After the system screens out the member information, the system further screens accurately by combining whether the member information is receivable push users or not. And adding the coupons appointed by the merchants, and pushing the coupons through a message template set by the system.
The user receives the push message through the network platform, can jump to a user coupon center interface of a shopping mall on the supermarket online by clicking, and can operate the coupon to realize the marketing purpose of the merchant.
The process not only achieves the purpose of marketing push, but also achieves the overall good marketing operation of the merchant due to the fact that independent information bases such as members, commodities and coupons of the merchant super-system are opened, and the user can accurately and sustainably reach the target user due to the fact that the user actively receives the push, and user experience is improved.
Example 6
As shown in fig. 6, a deep learning based recommendation apparatus includes:
the acquisition module 10: an acquisition module: the system comprises a plurality of user portraits and a plurality of commodity attributes, and a target user is locked according to the user portraits;
the extraction module 20: the recommendation system is used for extracting the display features of the target user and the display features of the plurality of commodity attributes locked by the acquisition module 10, and then processing the display features to generate a recommendation list;
the scoring module 30: the system is used for learning the hidden features of the target user and the hidden features of the plurality of commodity attributes by using a deep learning model, and predicting the grade of the target user on the recommendation list obtained by the extraction module 20 according to the hidden features;
the recommending module 40: and the system is used for pre-recommending the target user by the commodities in the recommendation list, determining whether the target user receives the recommendation, and recommending the related information of the commodities according to the grade of the recommendation list and the priority if the target user receives the recommendation.
One embodiment of the above apparatus may be: the acquisition module 10: the acquisition module acquires a plurality of user figures and a plurality of commodity attributes, and locks a target user according to the user figures; the extraction module 20 extracts the display features of the target user and the display features of the plurality of commodity attributes locked by the acquisition module 10, and then processes the display features to generate a recommendation list; the scoring module 30 learns the hidden features of the target user and the hidden features of the plurality of commodity attributes by using a deep learning model, and predicts the scoring of the target user on the recommendation list obtained by the extraction module 20 according to the hidden features; the recommending module 40 pre-recommends the commodities in the recommendation list to the target user, determines whether the target user receives the recommendation, and recommends the related information of the commodities according to the grade of the recommendation list and the priority if the target user receives the recommendation.
Example 7
As shown in fig. 7, an obtaining module 10 of a deep learning based recommendation device includes:
the first acquisition unit 12: the system comprises a plurality of user portraits and a plurality of commodity attributes, wherein the user portraits comprise behavior characteristics and preference characteristics, and the commodity attributes comprise basic commodity attributes and commodity evaluation;
the locking unit 14: the multi-dimensional screening item is set according to the behavior characteristics and the preference characteristics acquired by the first acquisition unit, and a target user is locked according to the multi-dimensional screening item.
One embodiment of the acquisition module 10 of the above apparatus may be: the first acquisition unit 12 acquires a plurality of user figures including behavior features and preference features and a plurality of commodity attributes including commodity basic attributes and commodity evaluations; the locking unit 14 sets a multidimensional screening item according to the behavior feature and the preference feature acquired by the first acquiring unit, and locks a target user according to the multidimensional screening item.
Example 8
As shown in fig. 8, an acquisition module 20 of a deep learning based recommendation device includes:
the first acquisition unit 22: the preference model is used for acquiring the behavior characteristics of the target user, browsing records and searching records of the commodities and constructing a preference model of the target user;
the second acquisition unit 24: the system is used for simultaneously acquiring the display characteristics of the attributes of the commodities and searching for similar commodities according to the display characteristics;
the cache unit 26: and the commodity buffer module is used for caching commodities in the approximate commodities and having a mapping relation with the preference model into a recommendation list.
One embodiment of the acquisition module 10 of the above apparatus may be: the first acquisition unit 22 acquires the behavior characteristics of the target user, browsing records and searching records of the plurality of commodities, and constructs a preference model of the target user; the second acquisition unit 24 acquires the display characteristics of the attributes of the plurality of commodities at the same time, and searches for an approximate commodity according to the display characteristics; the caching unit 26 caches the items in the approximate items, which have a mapping relation with the preference model, in a recommendation list.
Example 9
As shown in fig. 9, a scoring module 30 of a deep learning based recommendation apparatus includes:
the second acquisition unit 32: the comment data of the target user on the plurality of commodity attributes are acquired;
the construction unit 34: the system comprises a data acquisition module, a data processing module and a data processing module, wherein the data acquisition module is used for acquiring comment data of a target user on a plurality of commodity attributes, constructing K dimension characteristics according to the comment data of the target user on the plurality of commodity attributes, and determining K neighbor users of the target user on the K dimension characteristics according to the K dimension characteristics, wherein K is an integer greater than 1;
the scoring unit 36: the system is used for acquiring comment data of the K neighbor users on the plurality of commodity attributes and predicting the scores of the target user on the recommendation list according to the comment data of the K neighbor users on the plurality of commodity attributes.
One embodiment of the scoring module 30 of the above apparatus may be: the second acquiring unit 32 acquires comment data of the target user on the plurality of commodity attributes; the constructing unit 34 constructs K dimensional features according to the comment data of the target user on the plurality of commodity attributes, and determines K neighboring users of the target user on the K dimensional features according to the K dimensional features, wherein K is an integer greater than 1; the scoring unit 36 obtains comment data of the K neighboring users on the plurality of product attributes, and predicts the scoring of the target user on the recommendation list according to the comment data of the K neighboring users on the plurality of product attributes.
Example 10
As shown in fig. 10, an electronic device includes a memory 1001 and a processor 1002, where the memory 1001 is used for storing one or more computer instructions, and the one or more computer instructions are executed by the processor 1002 to implement one of the deep learning based recommendation methods described above.
It can be clearly understood by those skilled in the art that, for convenience and brevity of description, the specific working process of the electronic device described above may refer to the corresponding process in the foregoing method embodiment, and is not described herein again.
A computer-readable storage medium storing a computer program which, when executed, causes a computer to implement a deep learning based recommendation method as described above.
Illustratively, a computer program may be divided into one or more modules/units, one or more modules/units are stored in the memory 1001 and executed by the processor 1002, and the I/O interface transmission of data is performed by the input interface 1005 and the output interface 1006 to accomplish the present invention, and one or more of the modules/units may be a series of computer program instruction segments describing the execution of the computer program in a computer device, which can accomplish specific functions.
The computer device may be a desktop computer, a notebook, a palm computer, a cloud server, or other computing devices. The computer device may include, but is not limited to, the memory 1001 and the processor 1002, and those skilled in the art will appreciate that the present embodiment is only an example of the computer device, and does not constitute a limitation of the computer device, and may include more or less components, or combine some components, or different components, for example, the computer device may further include the input device 1007, a network access device, a bus, and the like.
The processor 1002 may be a Central Processing Unit (CPU), or may be other general-purpose processor 1002, a digital signal processor 1002 (DSP), an Application Specific Integrated Circuit (ASIC), an off-the-shelf programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or the like. The general purpose processor 1002 may be a microprocessor 1002 or the processor 1002 may be any conventional processor 1002 or the like.
The storage 1001 may be an internal storage unit of the computer device, such as a hard disk or a memory of the computer device. The memory 1001 may also be an external storage device of the computer device, such as a plug-in hard disk, a Smart Memory Card (SMC), a Secure Digital (SD) card, a flash memory card (FlashCard) and the like provided on the computer device, further, the memory 1001 may also include both an internal storage unit and an external storage device of the computer device, the memory 1001 is used for storing computer programs and other programs and data required by the computer device, the memory 1001 may also be used for temporarily storing the program codes in the outputter 1008, and the aforementioned storage media include various media capable of storing program codes, such as a usb disk, a removable hard disk, a read only memory ROM1003, a random access memory RAM1004, a disk and an optical disk.
The above description is only an embodiment of the present invention, but the technical features of the present invention are not limited thereto, and any changes or modifications within the technical field of the present invention by those skilled in the art are covered by the claims of the present invention.
Claims (10)
1. A deep learning-based recommendation method is characterized by comprising the following steps:
acquiring a plurality of user figures and a plurality of commodity attributes, and locking a target user according to the user figures;
extracting the display characteristics of the target user and the display characteristics of the plurality of commodity attributes, and then processing to generate a recommendation list;
learning the hidden features of the target user and the hidden features of the plurality of commodity attributes by using a deep learning model, and predicting the scores of the target user on the recommendation list according to the hidden features;
and pre-recommending the target user by the commodities in the recommendation list, determining whether the target user receives the recommendation, and if so, recommending the relevant information of the commodities according to the grade of the recommendation list and the priority.
2. The deep learning-based recommendation method of claim 1, wherein the obtaining a plurality of user figures and a plurality of commodity attributes, and locking a target user according to the plurality of user figures comprises:
the method comprises the steps of obtaining a plurality of user portraits and a plurality of commodity attributes, wherein the user portraits comprise behavior features and preference features, and the commodity attributes comprise basic commodity attributes and commodity evaluation;
and setting a multidimensional screening item according to the behavior characteristic and the preference characteristic, and locking a target user according to the multidimensional screening item.
3. The deep learning-based recommendation method according to claim 1, wherein the extracting the apparent features of the target user and the apparent features of the plurality of commodity attributes, and the reprocessing to generate the recommendation list comprises:
acquiring behavior characteristics of the target user, browsing records and searching records of the plurality of commodities, and constructing a preference model of the target user;
simultaneously acquiring the display characteristics of the attributes of the commodities, and searching for similar commodities according to the display characteristics;
and caching the commodities which have mapping relation with the preference model in the approximate commodities into a recommendation list.
4. The deep learning-based recommendation method according to claim 1, wherein the learning of the target user's hidden features and the hidden features of the plurality of product attributes by using a deep learning model and the prediction of the target user's rating of the recommendation list according to the hidden features comprises:
obtaining comment data of the target user on the plurality of commodity attributes;
constructing K dimensional features according to the comment data of the target user on the plurality of commodity attributes, and determining K neighbor users of the target user on the K dimensional features according to the K dimensional features, wherein K is an integer greater than 1;
and obtaining comment data of the K neighbor users on the plurality of commodity attributes, and predicting the scores of the target user on the recommendation list according to the comment data of the K neighbor users on the plurality of commodity attributes.
5. A deep learning based recommendation device, comprising:
an acquisition module: the system comprises a plurality of user portraits and a plurality of commodity attributes, and a target user is locked according to the user portraits;
an extraction module: the system is used for extracting the display features of the target user and the display features of the plurality of commodity attributes, which are locked by the acquisition module, and then processing the display features to generate a recommendation list;
a scoring module: the system comprises an extraction module, a deep learning module and a recommendation module, wherein the extraction module is used for extracting a recommendation list of a target user from a plurality of commodity attributes, and the recommendation list is used for carrying out deep learning on hidden features of the target user and hidden features of the commodity attributes by using the deep learning module;
a recommendation module: and the system is used for pre-recommending the target user by the commodities in the recommendation list, determining whether the target user receives the recommendation, and recommending the related information of the commodities according to the grade of the recommendation list and the priority if the target user receives the recommendation.
6. The deep learning-based recommendation device according to claim 5, wherein the obtaining module specifically comprises:
a first acquisition unit: the system comprises a plurality of user portraits and a plurality of commodity attributes, wherein the user portraits comprise behavior characteristics and preference characteristics, and the commodity attributes comprise basic commodity attributes and commodity evaluation;
a locking unit: the multi-dimensional screening item is set according to the behavior characteristics and the preference characteristics acquired by the first acquisition unit, and a target user is locked according to the multi-dimensional screening item.
7. The deep learning-based recommendation device according to claim 5, wherein the acquisition module specifically comprises:
a first acquisition unit: the preference model is used for acquiring the behavior characteristics of the target user, browsing records and searching records of the commodities and constructing a preference model of the target user;
a second acquisition unit: the system is used for simultaneously acquiring the display characteristics of the attributes of the commodities and searching for similar commodities according to the display characteristics;
a cache unit: and the commodity buffer module is used for caching commodities in the approximate commodities and having a mapping relation with the preference model into a recommendation list.
8. The deep learning-based recommendation device according to claim 5, wherein the scoring module specifically comprises:
a second acquisition unit: the comment data of the target user on the plurality of commodity attributes are acquired;
a construction unit: the system comprises a data acquisition module, a data processing module and a data processing module, wherein the data acquisition module is used for acquiring comment data of a target user on a plurality of commodity attributes, constructing K dimension characteristics according to the comment data of the target user on the plurality of commodity attributes, and determining K neighbor users of the target user on the K dimension characteristics according to the K dimension characteristics, wherein K is an integer greater than 1;
a scoring subunit: the system is used for acquiring comment data of the K neighbor users on the plurality of commodity attributes and predicting the scores of the target user on the recommendation list according to the comment data of the K neighbor users on the plurality of commodity attributes.
9. An electronic device comprising a memory and a processor, the memory configured to store one or more computer instructions, wherein the one or more computer instructions are executed by the processor to implement a deep learning based recommendation method as claimed in any one of claims 1-4.
10. A computer-readable storage medium storing a computer program, wherein the computer program is configured to enable a computer to implement a deep learning based recommendation method according to any one of claims 1 to 4 when executed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110433760.1A CN113222687A (en) | 2021-04-22 | 2021-04-22 | Deep learning-based recommendation method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110433760.1A CN113222687A (en) | 2021-04-22 | 2021-04-22 | Deep learning-based recommendation method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113222687A true CN113222687A (en) | 2021-08-06 |
Family
ID=77088429
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110433760.1A Pending CN113222687A (en) | 2021-04-22 | 2021-04-22 | Deep learning-based recommendation method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113222687A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114139052A (en) * | 2021-11-19 | 2022-03-04 | 北京百度网讯科技有限公司 | Ranking model training method for intelligent recommendation, intelligent recommendation method and device |
| CN114912948A (en) * | 2022-04-24 | 2022-08-16 | 深圳船奇科技有限公司 | Cloud service-based cross-border e-commerce big data intelligent processing method, device and equipment |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109087177A (en) * | 2018-08-21 | 2018-12-25 | 北京京东尚科信息技术有限公司 | To the method, apparatus and computer readable storage medium of target user's Recommendations |
| CN109359244A (en) * | 2018-10-30 | 2019-02-19 | 中国科学院计算技术研究所 | A kind of recommendation method for personalized information and device |
| CN110210933A (en) * | 2019-05-21 | 2019-09-06 | 清华大学深圳研究生院 | A kind of enigmatic language justice recommended method based on generation confrontation network |
| CN111191122A (en) * | 2019-12-20 | 2020-05-22 | 重庆邮电大学 | Learning resource recommendation system based on user portrait |
| CN111310029A (en) * | 2020-01-20 | 2020-06-19 | 哈尔滨理工大学 | Mixed recommendation method based on user commodity portrait and potential factor feature extraction |
-
2021
- 2021-04-22 CN CN202110433760.1A patent/CN113222687A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109087177A (en) * | 2018-08-21 | 2018-12-25 | 北京京东尚科信息技术有限公司 | To the method, apparatus and computer readable storage medium of target user's Recommendations |
| CN109359244A (en) * | 2018-10-30 | 2019-02-19 | 中国科学院计算技术研究所 | A kind of recommendation method for personalized information and device |
| CN110210933A (en) * | 2019-05-21 | 2019-09-06 | 清华大学深圳研究生院 | A kind of enigmatic language justice recommended method based on generation confrontation network |
| CN111191122A (en) * | 2019-12-20 | 2020-05-22 | 重庆邮电大学 | Learning resource recommendation system based on user portrait |
| CN111310029A (en) * | 2020-01-20 | 2020-06-19 | 哈尔滨理工大学 | Mixed recommendation method based on user commodity portrait and potential factor feature extraction |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114139052A (en) * | 2021-11-19 | 2022-03-04 | 北京百度网讯科技有限公司 | Ranking model training method for intelligent recommendation, intelligent recommendation method and device |
| CN114912948A (en) * | 2022-04-24 | 2022-08-16 | 深圳船奇科技有限公司 | Cloud service-based cross-border e-commerce big data intelligent processing method, device and equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Ahmed et al. | Scalable distributed inference of dynamic user interests for behavioral targeting | |
| CN109299994B (en) | Recommendation method, device, equipment and readable storage medium | |
| US10423648B2 (en) | Method, system, and computer readable medium for interest tag recommendation | |
| KR101419504B1 (en) | System and method providing a suited shopping information by analyzing the propensity of an user | |
| TWI512653B (en) | Information providing method and apparatus, method and apparatus for determining the degree of comprehensive relevance | |
| WO2018014759A1 (en) | Method, device and system for presenting clustering data table | |
| Chehal et al. | Implementation and comparison of topic modeling techniques based on user reviews in e-commerce recommendations | |
| US20230214895A1 (en) | Methods and systems for product discovery in user generated content | |
| CN106294500B (en) | Content item pushing method, device and system | |
| EP3061017A1 (en) | Systems and methods for determining influencers in a social data network | |
| Zhang et al. | Multimodal marketing intent analysis for effective targeted advertising | |
| CN108885624A (en) | Information recommendation system and method | |
| CN103870973A (en) | Information push and search method and apparatus based on electronic information keyword extraction | |
| CN116703485B (en) | Advertisement accurate marketing method and system based on big data | |
| CN109325179A (en) | A kind of method and device that content is promoted | |
| CN108073667B (en) | Method for generating user browsing attributes, and non-transitory computer readable medium | |
| CN108415913A (en) | Crowd's orientation method based on uncertain neighbours | |
| Ma et al. | Measuring the coverage and redundancy of information search services on e-commerce platforms | |
| CN113222687A (en) | Deep learning-based recommendation method and device | |
| Xu et al. | Do adjective features from user reviews address sparsity and transparency in recommender systems? | |
| Kim et al. | Competitive intelligence in Korean Ramen market using text mining and sentiment analysis | |
| CN110851694A (en) | Personalized recommendation system based on user memory network and tree structure depth model | |
| Choi et al. | Fake review identification and utility evaluation model using machine learning | |
| Sardar et al. | Recommender system for journal articles using opinion mining and semantics | |
| CN111445302A (en) | Commodity sorting method, system and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210806 |
|
| RJ01 | Rejection of invention patent application after publication |