CN112862007A - Commodity sequence recommendation method and system based on user interest editing - Google Patents
Commodity sequence recommendation method and system based on user interest editing Download PDFInfo
- Publication number
- CN112862007A CN112862007A CN202110332569.8A CN202110332569A CN112862007A CN 112862007 A CN112862007 A CN 112862007A CN 202110332569 A CN202110332569 A CN 202110332569A CN 112862007 A CN112862007 A CN 112862007A
- Authority
- CN
- China
- Prior art keywords
- sequence
- interest
- commodity
- user
- prediction model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0631—Item recommendations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Finance (AREA)
- Accounting & Taxation (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (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 present disclosure provides a commodity sequence recommendation method and system based on user interest editing, including: acquiring a commodity history sequence generating interactive behaviors with a user; inputting the commodity historical sequence into a pre-trained sequence prediction model, and outputting a recommended commodity; the training process adopts an interest editing strategy to enable the sequence prediction model to learn the commonalities and the peculiarities among different commodity historical sequences to obtain a recombined sequence representation, and the recombined sequence is used for training the sequence prediction model. The scheme solves the problem of user interest extraction and representation in sequence recommendation through an automatic supervision technology, and forces a sequence recommendation model to be capable of distinguishing the commonality and the particularity of different sequences of a user in interaction with a recommendation system based on an interest editing strategy, so that more accurate user interest is obtained, and the accuracy of sequence recommendation is improved.
Description
Technical Field
The disclosure belongs to the technical field of commodity sequence recommendation, and particularly relates to a commodity sequence recommendation method and system based on user interest editing.
Background
The statements in this section merely provide background information related to the present disclosure and may not necessarily constitute prior art.
The sequential recommendation is a method for making a recommendation to a user by capturing short-term or long-term hobbies of the user according to browsing or purchasing records of the user within a certain period of time. The sequence recommendation plays a very important role in a recommendation system, and the method models the record of the purchasing or browsing behavior of the user so as to learn the interest expression and change of the user, so that the next click of the user can be predicted and recommended. In addition, related products utilizing the technologies are widely applied to various fields nowadays, such as Taobao recommends commodities meeting the interests of users, Internet-accessible cloud concerts recommend music meeting the moods of users, American parties recommend takeoffs and restaurants meeting the preferences of users, and the like, which greatly facilitates the lives of people and promotes the benefits of the sales industry. Meanwhile, as the big data age comes, the number of users and user records have increased, and in order to fully mine the association behind data, academic and industrial fields have paid much attention to this.
The inventor finds that, in sequence recommendation, existing methods based on technologies such as Convolutional Neural Network (CNN), cyclic neural network (RNN), and Transformer (Transformer) mostly assume that interests of users are concentrated and mixed, and they usually consider that there is a main user interest in a sequence, so they usually use a mixed vector to represent the overall interests of the users and make recommendations based on the main user interest, which does not distinguish and represent different interests of the users, resulting in low accuracy of recommendation results and no help to interpretability of the recommendations. Also, existing methods, which typically rely on the recording of the next click in a sequence of user interactions to supervise the model learning process, while effective in capturing user interest, ignore other associations inherent in the data. Previous methods using self-supervision technology simply search for the self-supervision signal of the surface layer in the sequence, and do not extend to the sequence to search for deeper association.
Disclosure of Invention
The scheme solves the problem of user interest extraction and representation in sequence recommendation through a self-supervision technology, and forces a sequence recommendation model to be capable of distinguishing the commonality and the uniqueness of different sequences of a user in interaction with a recommendation system based on an interest editing strategy, so that more accurate user interest is obtained, and the accuracy of sequence recommendation is improved.
According to a first aspect of the embodiments of the present disclosure, there is provided a method for recommending a commodity sequence edited based on user interests, including:
acquiring a commodity history sequence generating interactive behaviors with a user;
inputting the commodity historical sequence into a pre-trained sequence prediction model, and outputting a recommended commodity;
the sequence prediction model comprises a sequence encoder, an interest discriminator and a sequence decoder, wherein an interest editing strategy is adopted in the training process, so that the sequence prediction model learns the commonalities and the peculiarities among different commodity historical sequences to obtain a recombined sequence representation, and the recombined sequence is used for training the sequence prediction model.
Further, the interest editing strategy comprises two operations of interest separation and interest exchange, multiple interest representations of different sequences are obtained through an interest discriminator, then the sequence prediction model is forced to learn the commonality and uniqueness among the sequences through the interest separation operation, the respective commonality representation parts of the sequences are exchanged through the interest exchange operation, and the sequence representation after recombination is generated for each sequence.
Further, in the sequence encoder, a plurality of special marks are spliced for each commodity sequence, wherein each mark represents special interest of a user, and the marked commodity sequence is encoded into a hidden state representation through the sequence encoder.
Further, in the interest discriminator, the attention distribution of each special mark to all commodities in the sequence is calculated, and meanwhile, an interest coverage mechanism is introduced to avoid that different special marks focus on the same commodity.
According to a second aspect of the embodiments of the present disclosure, there is provided a commodity sequence recommendation system edited based on user interests, including:
the data acquisition unit is used for acquiring a commodity history sequence generating interactive behaviors with a user;
a commodity recommending unit for inputting the commodity history sequence into a pre-trained sequence prediction model and outputting a recommended commodity;
the sequence prediction model comprises a sequence encoder, an interest discriminator and a sequence decoder, wherein an interest editing strategy is adopted in the training process, so that the sequence prediction model learns the commonalities and the peculiarities among different commodity historical sequences to obtain a recombined sequence representation, and the recombined sequence is used for training the sequence prediction model.
According to a third aspect of the embodiments of the present disclosure, there is provided an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the memory, where the processor implements the method for recommending a commodity sequence based on user interest editing when executing the program.
According to a fourth aspect of the embodiments of the present disclosure, there is provided a non-transitory computer-readable storage medium having stored thereon a computer program, which when executed by a processor, implements the method for recommending a product sequence edited based on user interests.
Compared with the prior art, the beneficial effect of this disclosure is:
(1) the scheme disclosed by the invention solves the problem of user interest extraction and representation in sequence recommendation through a self-supervision technology, and forces a sequence recommendation model to be capable of distinguishing the commonality and the particularity of different sequences of a user in interaction with a recommendation system based on an interest editing strategy, so that more accurate user interest is obtained, and the accuracy of sequence recommendation is improved;
(2) the scheme creatively provides a brand-new self-supervision loss function to enhance the learning process, and simultaneously separately models hidden various interest representations in the user sequence. Compared with the existing sequence recommendation method, the scheme has the advantages that the incidence relation among the sequences is fully mined, so that a plurality of evaluation indexes on a plurality of data sets in the e-commerce recommendation field are improved;
(3) the scheme can be applied to recommendation scenes in multiple fields, can accurately judge and capture user interests, and enables a recommendation system to be more detailed and accurate, so that the use experience of users can be improved, and the income of e-commerce operation can also be improved.
Advantages of additional aspects of the disclosure will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the disclosure.
Drawings
The accompanying drawings, which are included to provide a further understanding of the disclosure, illustrate embodiments of the disclosure and together with the description serve to explain the disclosure and are not to limit the disclosure.
Fig. 1 is a flowchart of a commodity sequence recommendation method edited based on user interests in a first embodiment of the present disclosure.
Detailed Description
The present disclosure is further described with reference to the following drawings and examples.
It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the disclosure. 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 disclosure belongs.
It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments according to the present disclosure. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, and it should be understood that when the terms "comprises" and/or "comprising" are used in this specification, they specify the presence of stated features, steps, operations, devices, components, and/or combinations thereof, unless the context clearly indicates otherwise.
The embodiments and features of the embodiments in the present disclosure may be combined with each other without conflict.
The first embodiment is as follows:
the embodiment aims to provide a commodity sequence recommendation method based on user interest editing.
A commodity sequence recommendation method based on user interest editing comprises the following steps:
acquiring a commodity history sequence generating interactive behaviors with a user;
inputting the commodity historical sequence into a pre-trained sequence prediction model, and outputting a recommended commodity;
the sequence prediction model comprises a sequence encoder, an interest discriminator and a sequence decoder, wherein an interest editing strategy is adopted in the training process, so that the sequence prediction model learns the commonalities and the peculiarities among different commodity historical sequences to obtain a recombined sequence representation, and the recombined sequence is used for training the sequence prediction model.
Specifically, for the convenience of understanding, the scheme of the present embodiment is described in detail below with reference to fig. 1:
the basic model of the present disclosure adopts the currently popular encoder-decoder framework based on deep learning, which is named as Multi-prediction Transformer (MrTransformer), and includes three modules, respectively: a sequence encoder, an interest resolver, and a sequence decoder. In the learning process of the model, an interest Editing strategy (Preference Editing) is adopted to guide model training, wherein the method comprises two operations: interest separation and interest exchange. They force the model to learn the commonality and peculiarities between different sequences, thus generating new self-supervision losses to guide model training. The basic model (mrtransform) and the interest Editing (Preference Editing) constitute the overall architecture of the present disclosure, called mrtransform (pe), and the workflow thereof is shown in fig. 1.
Next, the inputs and outputs of the definition MrTransformer (PE) are formalized.Let I ═ I1,i2,…,iτ,…,itDenotes a commodity set, let S ═ S1,S2,…SSDenotes a sequence set. Given an arbitrary sequence S ═ i1,i2,…,iτ,…,itIn which iτRepresenting the goods interacted by the user at the time tau, the model aims to model the implicit user interest representation in the sequence so as to predict the goods interacted by the user at the next time, and the process can be formally defined as follows:
P(it+1|S)～f(S) (1)
wherein, P (i)t+1I S) represents the probability calculation for recommending the next good, and f (S) is a function that models this probability.
Each part of MrTransformer (PE) is described in detail below:
(1) sequence encoder
For any sequence, unlike existing cyclic neural network (RNN) and Transformer based work, the scheme of the present disclosure splices K special tags ([ P1 ] before the sequence],[P2],…,[PK]) Wherein each token represents a particular representation of the user's interest (the user may reflect different interests in the sequence of interactions, e.g. during reading, he may prefer both the scientific and romantic types and the martial arts type, and each representation of interest we use the aforementioned particular token [ Pi]To represent), K represents the number of interests for the entire data set, not for a particular sequence. The scheme of the present disclosure defines the sequence after processing as follows:
S′＝{[P1],[P2],…,[PK],i1,i2,…iτ,…it} (2)
in this module, the sequence after processing is encoded into a hidden state representation, specifically: first, the scheme described in this disclosure initializes the representation of the sequence and the position representations as E and P, respectively, and then adds the position representation to the sequence representation as a whole: e ═ E + P. These sequential representations are then input into L-stacked bidirectional converter layers, each layer iteratively modifying the representation at all locations by exchanging information at certain specific locations by a mask matrix, which can be defined as:
El＝Trm(El-1,Maske) (3)
wherein Trm represents the converter layer, ElIs a sequence representation matrix, Mask, of the l-th layereIs a mask matrix. In particular, for each particular mark Pk]He can retrieve information from all locations, since it aims to capture the user-specific interest representation by observing the whole sequence, so the observation field is the whole sequence, whose mask vector is all ones. For each commodity in the sequence, the information it can accept is limited to between commodities, so the mask vector consists of K zeros and t ones. Finally, E of the uppermost layer can be obtainedlSequence representation as a whole.
(2) Interest distinguisher
The attention distribution of each special marker for all the items in the sequence is calculated in this module, as follows:
P,A＝Ident(El,MaskI) (4)
wherein Ident is realized by a conversion layer, MaskIIs a mask matrix for interest discrimination. Unlike the mask matrix in the encoder, each special marker can only compute the interest distribution for the commodity, so the information of other markers needs to be removed. For each commodity, the mask vector is the same as the mask vector for the commodity in the encoder. P represents the multi-interest expression matrix of the first K special marks, and A is the interest distribution matrix generated by the marks on the commodity.
Meanwhile, in order to avoid that different special marks concern the same commodity, the scheme of the disclosure introduces an interest coverage mechanism, specifically, maintains K coverage vectors
(Vector)
Recorded in a special mark Pk]The sum of the previous attention distributions of all markers for all items in the sequence, which represents the degree of coverage of those items from the attention mechanism, is calculated as follows:
wherein, akIs composed of [ Pk]The attention distribution vectors generated for all the commodities are represented and
is a zero vector, which means that at the first time step, no goods are covered.
(3) Sequence decoder
In order to use the obtained representation of the multiple interests of the user to make recommendations, the present embodiment calculates by the following formula:
P(it+1|S)＝softmax(pW+b) (6)
where P is the overall vector representation resulting from the summation of the multiple interest representation matrices P, W is the embedded representation matrix for all the commodities, and b is the bias term.
(4) Model training objective function
Consistent with most sequence recommendations, the primary goal of the solution described in this disclosure is to predict the next item for each position of the input sequence. The scheme uses negative log-likelihood as the recommended loss function, which is calculated as follows:
where θ is all parameters of MrTransformer. In addition, the solution of the present disclosure further defines a coverage loss to punish the situations of different interests concerning the same commodity:
and finally, multiplying the coverage loss function by the super parameter alpha, and adding the recommended loss function as the overall loss function to guide the whole learning process:
L(θ)＝Lrec(θ)+α·Lcov(θ) (9)
where alpha controls the ratio of the coverage loss function.
(5) Interest editing learning strategy
The learning strategy aims to mine the association relation between commodities among different interaction sequences.
As shown in FIG. 1, interest editing involves two operations: interest separation and interest exchange. First, two sequences need to be sampled from a sequence set, which needs to ensure that the two sequences have a certain degree of coincidence and respective uniqueness. By the interest resolvers in the basic model, multiple interest representations of different sequences can be obtained. The interest separation operation then forces the model to learn the commonalities and uniqueness between the sequences, and the interest exchange operation exchanges the respective commonalities representation parts to generate a re-assembled sequence representation for each of the original sequences.
Interest separation operation: for each pair of sequences SxAnd SyIn order to measure their similarity, their correlation is expressed by calculating a similarity matrix, each element IijCan be calculated as follows:
wherein p isiAnd pjRespectively a multi-interest representation matrix PxAnd PyThe vector of (a) is determined,
representing element multiplication operations. Based on the similarity matrix, an attention matrix A may be calculatedxAnd ByThey reflect the attention distribution of one sequence's interest representation over another sequence's interest representation, as calculated as follows:
Ax＝softmaxrow(I) (11)
By＝softmaxcol(I)
wherein, softmaxrowAnd softmaxcolRespectively representing the calculation of the softmax function in rows and columns. Then, the common and specific interest representation for each sequence can be calculated as follows:
wherein, CxAnd CyRepresenting a common interest representation, U, for each sequencexAnd UyRepresenting the respective unique interest representation of the sequences.
Interest exchange operation: for both sequences, the sequence representation after recombination is obtained by exchanging the previously obtained common interest representation, as follows:
based on the recombined new sequence representation, the scheme of the present disclosure defines two kinds of self-supervision signals for training:
LSSL(θ)＝Lpred(θ)+Lapp(θ) (14)
wherein L ispred(theta) representing P based on the recombined sequencex' and PyTo predict the next commodity, the calculation uses the negative log-likelihood function in the basic model:
Lpred(θ)＝Lx rec+Ly rec (15)
another term is a regularization term, which contains three parts:
wherein L isapp(θ)xAnd Lapp(θ)yEnsuring that the sequence representation after recombination approaches the original sequence representation indefinitely, Lapp(θ)cEnsuring that the commonality between the two sequences indicates that the parts are close enough.
Finally, the training loss function for MrTransformer (PE) is calculated as follows:
Lall(θ)＝L(θ)+LSSL(θ) (17)
where L (θ) is used for interest extraction and representation, LSSL(θ) for interest editing.
The scheme disclosed by the disclosure aims to innovatively apply an automatic supervision learning theory to the field of sequence recommendation, creatively provides a brand-new automatic supervision loss function to enhance the learning process, and simultaneously separately models various hidden interest expressions in a user sequence. Compared with the existing sequence recommendation method, the scheme disclosed by the invention has the advantages that the incidence relation among the sequences is fully mined, so that a plurality of evaluation indexes on a plurality of data sets in the field of e-commerce recommendation are improved. Meanwhile, the scheme can be applied to recommendation scenes in multiple fields, user interest can be judged and captured more accurately, and the recommendation system is more detailed and accurate, so that the use experience of users can be improved, and the income of e-commerce operation can also be improved.
Example two:
the embodiment aims to provide a commodity sequence recommendation system edited based on user interest.
A system for recommending a sequence of goods edited based on user's interest, comprising:
the data acquisition unit is used for acquiring a commodity history sequence generating interactive behaviors with a user;
a commodity recommending unit for inputting the commodity history sequence into a pre-trained sequence prediction model and outputting a recommended commodity;
the sequence prediction model comprises a sequence encoder, an interest discriminator and a sequence decoder, wherein an interest editing strategy is adopted in the training process, so that the sequence prediction model learns the commonalities and the peculiarities among different commodity historical sequences to obtain a recombined sequence representation, and the recombined sequence is used for training the sequence prediction model.
In further embodiments, there is also provided:
an electronic device comprising a memory and a processor, and computer instructions stored on the memory and executed on the processor, the computer instructions when executed by the processor performing the method of embodiment one. For brevity, no further description is provided herein.
It should be understood that in this embodiment, the processor may be a central processing unit CPU, and the processor may also be other general purpose processors, digital signal processors DSP, application specific integrated circuits ASIC, off-the-shelf programmable gate arrays FPGA or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, and so on. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
The memory may include both read-only memory and random access memory, and may provide instructions and data to the processor, and a portion of the memory may also include non-volatile random access memory. For example, the memory may also store device type information.
A computer readable storage medium storing computer instructions which, when executed by a processor, perform the method of embodiment one.
The method in the first embodiment may be directly implemented by a hardware processor, or may be implemented by a combination of hardware and software modules in the processor. The software modules may be located in ram, flash, rom, prom, or eprom, registers, among other storage media as is well known in the art. The storage medium is located in a memory, and a processor reads information in the memory and completes the steps of the method in combination with hardware of the processor. To avoid repetition, it is not described in detail here.
Those of ordinary skill in the art will appreciate that the various illustrative elements, i.e., algorithm steps, described in connection with the embodiments disclosed herein may be implemented as electronic hardware or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present disclosure.
The commodity sequence recommendation method and system based on user interest editing provided by the embodiment can be realized, and have a wide application prospect.
The above description is only a preferred embodiment of the present disclosure and is not intended to limit the present disclosure, and various modifications and changes may be made to the present disclosure by those skilled in the art. Any modification, equivalent replacement, improvement and the like made within the spirit and principle of the present disclosure should be included in the protection scope of the present disclosure.
Although the present disclosure has been described with reference to specific embodiments, it should be understood that the scope of the present disclosure is not limited thereto, and those skilled in the art will appreciate that various modifications and changes can be made without departing from the spirit and scope of the present disclosure.
Claims (10)
1. A commodity sequence recommendation method based on user interest editing is characterized by comprising the following steps:
acquiring a commodity history sequence generating interactive behaviors with a user;
inputting the commodity historical sequence into a pre-trained sequence prediction model, and outputting a recommended commodity;
the sequence prediction model comprises a sequence encoder, an interest discriminator and a sequence decoder, wherein an interest editing strategy is adopted in the training process, so that the sequence prediction model learns the commonalities and the peculiarities among different commodity historical sequences to obtain a recombined sequence representation, and the recombined sequence is used for training the sequence prediction model.
2. The method as claimed in claim 1, wherein the interest editing strategy includes two operations of interest separation and interest exchange, multiple interest representations of different sequences are obtained through an interest resolver, and then the sequence prediction model is forced to learn the commonality and uniqueness among the sequences through the interest separation operation, and the interest exchange operation exchanges respective common representation parts to generate the recombined sequence representation for each previous sequence.
3. The commodity sequence recommendation method based on user interest editing according to claim 2, wherein the interest classification operation specifically comprises: for each pair of sequences SxAnd SyThe degree of correlation is expressed by calculating a similarity matrix; based on the similarity matrix, an attention matrix A is calculatedxAnd ByAnd obtaining a representation part with commonality and uniqueness of each sequence according to the attention matrix.
4. The commodity sequence recommendation method based on user interest editing according to claim 2, wherein the interest exchange operation specifically is: and exchanging the obtained public representation parts for the two sequences to obtain the sequence representation after recombination.
5. The method as claimed in claim 1, wherein in the sequence encoder, a plurality of special marks are spliced for each commodity sequence, wherein each mark represents a special interest of the user, and the marked commodity sequence is encoded into a hidden state representation by the sequence encoder.
6. The commodity sequence recommendation method based on user interest editing of claim 1, wherein in the interest discriminator, the attention distribution of each special mark to all commodities in the sequence is calculated, and meanwhile, an interest coverage mechanism is introduced to avoid that different special marks focus on the same commodity.
7. The commodity sequence recommendation method based on user interest editing according to claim 1, wherein an objective function adopted by the sequence prediction model training is a combination of a recommendation loss function and a coverage loss function, wherein the recommendation loss function adopts negative log likelihood estimation, and the coverage loss function is used for punishing a situation that different interests concern the same commodity.
8. A system for recommending a sequence of goods edited based on user's interest, comprising:
the data acquisition unit is used for acquiring a commodity history sequence generating interactive behaviors with a user;
a commodity recommending unit for inputting the commodity history sequence into a pre-trained sequence prediction model and outputting a recommended commodity;
the sequence prediction model comprises a sequence encoder, an interest discriminator and a sequence decoder, wherein an interest editing strategy is adopted in the training process, so that the sequence prediction model learns the commonalities and the peculiarities among different commodity historical sequences to obtain a recombined sequence representation, and the recombined sequence is used for training the sequence prediction model.
9. An electronic device comprising a memory, a processor and a computer program stored and executed on the memory, wherein the processor implements a method for recommending a product sequence based on user's interest as claimed in any one of claims 1-7 when executing the program.
10. A non-transitory computer readable storage medium having stored thereon a computer program, wherein the program when executed by a processor implements a method for recommending a sequence of items based on user's interests according to any of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110332569.8A CN112862007B (en) | 2021-03-29 | 2021-03-29 | Commodity sequence recommendation method and system based on user interest editing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110332569.8A CN112862007B (en) | 2021-03-29 | 2021-03-29 | Commodity sequence recommendation method and system based on user interest editing |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112862007A true CN112862007A (en) | 2021-05-28 |
| CN112862007B CN112862007B (en) | 2022-12-13 |
Family
ID=75993093
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110332569.8A Active CN112862007B (en) | 2021-03-29 | 2021-03-29 | Commodity sequence recommendation method and system based on user interest editing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112862007B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114004340A (en) * | 2021-11-15 | 2022-02-01 | 南开大学 | Cyclic neural network interpretable method and system based on sequence transformation |
Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109509054A (en) * | 2018-09-30 | 2019-03-22 | 平安科技（深圳）有限公司 | Method of Commodity Recommendation, electronic device and storage medium under mass data |
| CN109710805A (en) * | 2018-12-13 | 2019-05-03 | 百度在线网络技术（北京）有限公司 | Video interactive method and device based on interest cluster |
| WO2019205795A1 (en) * | 2018-04-26 | 2019-10-31 | 腾讯科技（深圳）有限公司 | Interest recommendation method, computer device, and storage medium |
| CN110516160A (en) * | 2019-08-30 | 2019-11-29 | 中国科学院自动化研究所 | User modeling method, the sequence of recommendation method of knowledge based map |
| CN110807156A (en) * | 2019-10-23 | 2020-02-18 | 山东师范大学 | Interest recommendation method and system based on user sequence click behaviors |
| CN111127142A (en) * | 2019-12-16 | 2020-05-08 | 东北大学秦皇岛分校 | Article recommendation method based on generalized neural attention |
| CN111242729A (en) * | 2020-01-07 | 2020-06-05 | 西北工业大学 | Serialization recommendation method based on long-term and short-term interests |
| CN111368210A (en) * | 2020-05-27 | 2020-07-03 | 腾讯科技（深圳）有限公司 | Information recommendation method and device based on artificial intelligence and electronic equipment |
| CN111709810A (en) * | 2020-06-17 | 2020-09-25 | 腾讯云计算（北京）有限责任公司 | Object recommendation method and device based on recommendation model |
| CN111815410A (en) * | 2020-07-07 | 2020-10-23 | 中国人民解放军军事科学院国防科技创新研究院 | Commodity recommendation method based on selective neighborhood information |
| CN111932336A (en) * | 2020-07-17 | 2020-11-13 | 重庆邮电大学 | Commodity list recommendation method based on long-term and short-term interest preference |
| CN112381581A (en) * | 2020-11-17 | 2021-02-19 | 东华理工大学 | Advertisement click rate estimation method based on improved Transformer |
| CN112395504A (en) * | 2020-12-01 | 2021-02-23 | 中国计量大学 | Short video click rate prediction method based on sequence capsule network |
| CN112528147A (en) * | 2020-12-10 | 2021-03-19 | 北京百度网讯科技有限公司 | Content recommendation method and apparatus, training method, computing device, and storage medium |
-
2021
- 2021-03-29 CN CN202110332569.8A patent/CN112862007B/en active Active
Patent Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019205795A1 (en) * | 2018-04-26 | 2019-10-31 | 腾讯科技（深圳）有限公司 | Interest recommendation method, computer device, and storage medium |
| CN109509054A (en) * | 2018-09-30 | 2019-03-22 | 平安科技（深圳）有限公司 | Method of Commodity Recommendation, electronic device and storage medium under mass data |
| CN109710805A (en) * | 2018-12-13 | 2019-05-03 | 百度在线网络技术（北京）有限公司 | Video interactive method and device based on interest cluster |
| CN110516160A (en) * | 2019-08-30 | 2019-11-29 | 中国科学院自动化研究所 | User modeling method, the sequence of recommendation method of knowledge based map |
| CN110807156A (en) * | 2019-10-23 | 2020-02-18 | 山东师范大学 | Interest recommendation method and system based on user sequence click behaviors |
| CN111127142A (en) * | 2019-12-16 | 2020-05-08 | 东北大学秦皇岛分校 | Article recommendation method based on generalized neural attention |
| CN111242729A (en) * | 2020-01-07 | 2020-06-05 | 西北工业大学 | Serialization recommendation method based on long-term and short-term interests |
| CN111368210A (en) * | 2020-05-27 | 2020-07-03 | 腾讯科技（深圳）有限公司 | Information recommendation method and device based on artificial intelligence and electronic equipment |
| CN111709810A (en) * | 2020-06-17 | 2020-09-25 | 腾讯云计算（北京）有限责任公司 | Object recommendation method and device based on recommendation model |
| CN111815410A (en) * | 2020-07-07 | 2020-10-23 | 中国人民解放军军事科学院国防科技创新研究院 | Commodity recommendation method based on selective neighborhood information |
| CN111932336A (en) * | 2020-07-17 | 2020-11-13 | 重庆邮电大学 | Commodity list recommendation method based on long-term and short-term interest preference |
| CN112381581A (en) * | 2020-11-17 | 2021-02-19 | 东华理工大学 | Advertisement click rate estimation method based on improved Transformer |
| CN112395504A (en) * | 2020-12-01 | 2021-02-23 | 中国计量大学 | Short video click rate prediction method based on sequence capsule network |
| CN112528147A (en) * | 2020-12-10 | 2021-03-19 | 北京百度网讯科技有限公司 | Content recommendation method and apparatus, training method, computing device, and storage medium |
Non-Patent Citations (4)
| Title |
|---|
| MUYANG MA等: "π-Net: A Parallel Information-sharing Network for Shared-account Cross-domain Sequential Recommendations", 《SIGIR "19》 * |
| PENGJIE REN等: "Neural Attentive Session-based Recommendation", 《CIKM’17》 * |
| 唐诚: "基于电商数据的用户兴趣挖掘与POI个性化推荐研究", 《中国优秀博硕士学位论文全文数据库(硕士)信息科技辑》 * |
| 孟祥福等: "基于地理一社会关系的多样性与个性化兴趣点推荐", 《计算机学报》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114004340A (en) * | 2021-11-15 | 2022-02-01 | 南开大学 | Cyclic neural network interpretable method and system based on sequence transformation |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112862007B (en) | 2022-12-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Apostolidis et al. | Combining global and local attention with positional encoding for video summarization | |
| Kim et al. | Multi-object tracking with neural gating using bilinear lstm | |
| WO2019210695A1 (en) | Model training and service recommendation | |
| US20150254554A1 (en) | Information processing device and learning method | |
| CN108921657B (en) | Knowledge-enhanced memory network-based sequence recommendation method | |
| CN106202329A (en) | Sample data process, data identification method and device, computer equipment | |
| CN108984555B (en) | User state mining and information recommendation method, device and equipment | |
| CN110196928B (en) | Fully parallelized end-to-end multi-turn dialogue system with domain expansibility and method | |
| CN113191838B (en) | Shopping recommendation method and system based on heterogeneous graph neural network | |
| CN114036406A (en) | Recommendation method and system based on graph contrast learning and social network enhancement | |
| CN111052128A (en) | Descriptor learning method for detecting and locating objects in video | |
| KR20210043881A (en) | Method and Device for Completing Social Network Using Artificial Neural Network | |
| CN110689110A (en) | Method and device for processing interaction event | |
| Agbinya | Applied Data Analytics-Principles and Applications | |
| CN112862007B (en) | Commodity sequence recommendation method and system based on user interest editing | |
| CN114925270A (en) | Session recommendation method and model | |
| US20240037133A1 (en) | Method and apparatus for recommending cold start object, computer device, and storage medium | |
| Wang et al. | Dmfp: a dynamic multi-faceted fine-grained preference model for recommendation | |
| CN117251622A (en) | Method, device, computer equipment and storage medium for recommending objects | |
| CN116501979A (en) | Information recommendation method, information recommendation device, computer equipment and computer readable storage medium | |
| CN116522232A (en) | Document classification method, device, equipment and storage medium | |
| CN116361643A (en) | Model training method for realizing object recommendation, object recommendation method and related device | |
| CN113032612B (en) | Construction method of multi-target image retrieval model, retrieval method and device | |
| CN113343041B (en) | Message reply relation judgment system based on graph model representation learning | |
| CN114417161A (en) | Virtual article time sequence recommendation method, device, medium and equipment based on special-purpose map |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |