CN112445981A - Social and consumption joint recommendation system, method, storage medium and computer equipment - Google Patents
Social and consumption joint recommendation system, method, storage medium and computer equipment Download PDFInfo
- Publication number
- CN112445981A CN112445981A CN202011218535.8A CN202011218535A CN112445981A CN 112445981 A CN112445981 A CN 112445981A CN 202011218535 A CN202011218535 A CN 202011218535A CN 112445981 A CN112445981 A CN 112445981A
- Authority
- CN
- China
- Prior art keywords
- social
- consumption
- preference
- joint
- characteristic
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 30
- 239000011159 matrix material Substances 0.000 claims abstract description 51
- 230000007246 mechanism Effects 0.000 claims abstract description 28
- 238000004458 analytical method Methods 0.000 claims abstract description 9
- 238000013528 artificial neural network Methods 0.000 claims description 22
- 238000001228 spectrum Methods 0.000 claims description 5
- 238000004590 computer program Methods 0.000 claims description 4
- 238000012545 processing Methods 0.000 claims description 4
- 230000004931 aggregating effect Effects 0.000 claims description 3
- 238000000354 decomposition reaction Methods 0.000 abstract description 6
- 238000013135 deep learning Methods 0.000 abstract description 3
- 230000003595 spectral effect Effects 0.000 abstract description 2
- 230000006399 behavior Effects 0.000 description 4
- 230000007547 defect Effects 0.000 description 4
- 230000031068 symbiosis, encompassing mutualism through parasitism Effects 0.000 description 4
- 230000011273 social behavior Effects 0.000 description 3
- 238000003491 array Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- RTAQQCXQSZGOHL-UHFFFAOYSA-N Titanium Chemical compound [Ti] RTAQQCXQSZGOHL-UHFFFAOYSA-N 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000001149 cognitive effect Effects 0.000 description 1
- 238000002790 cross-validation Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000008092 positive effect Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 238000012549 training Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9536—Search customisation based on social or collaborative filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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
Abstract
The invention belongs to the technical field of recommendation systems of deep learning applications, and discloses a social and consumption combined recommendation system, a method, a storage medium and computer equipment. The social and consumption joint recommendation system comprises: self-attention spatial layer, self-attention spectral layer, reciprocal analysis layer, prediction layer. The invention solves the limitation of sparse matrix decomposition; the introduction of the self-attention model fully considers the individual difference, so that the characteristics extracted by the first two layers are more fit with the real attributes of the user; the introduction of a reciprocal mechanism fully exerts the interactivity of the original information of the two recommendation systems, and improves the accuracy, recall rate and NDCG index of the recommendation of the prediction layer.
Description
Technical Field
The invention belongs to the technical field of recommendation systems for deep learning applications, and particularly relates to a social and consumption combined recommendation system, method, storage medium and computer equipment.
Background
At present: many social studies have shown that people's consumption behavior and social behavior are closely related. However, most current research either considers only social impact on consumption recommendations, explores recommendations for user relationships, or uses one of the recommendations as an auxiliary reference for another recommendation to improve accuracy. Only a few efforts have been made to treat social and consumption recommendations as a common problem, and these solutions are usually based on matrix decomposition and neural networks, and then simply use cascades or weighted sums to aggregate the consumption and social behavior information of users, so that the interaction relationship between the two cannot be sufficiently characterized, and many potential recommendation possibilities are missed.
Through the above analysis, the problems and defects of the prior art are as follows:
(1) in the prior art, a matrix decomposition technology is adopted to extract consumption preference characteristics and social preference characteristics, a recommendation system inevitably encounters the problems of new user recommendation and sparse data of new data warehousing, however, the matrix decomposition is not sensitive to a sparse matrix, and the recommendation problem in the case of sparse data cannot be solved through the consumption preference characteristics or the social preference characteristics extracted through the matrix decomposition;
(2) the existing joint recommendation technology adopts a direct cascade or weighted cascade mode to summarize and use the features extracted in the process, and researches show that the consumption behaviors of users are easily influenced by friends of the users, meanwhile, people with the same consumption behaviors are easier to establish social contact, and the characteristics are simply summarized by using cascade or weighted sum, so that the interactivity of the two cannot be fully reflected;
(3) the existing joint recommendation technology neglects the action and influence of individual subjective cognitive difference in a recommendation system, the influence of friends with different intimacy on consumption behaviors of the friends is different necessarily, and meanwhile, the social behaviors of the friends are also influenced necessarily by different love degrees of articles. If individual differences cannot be taken into account, the accuracy of the prediction result is inevitably affected.
The difficulty in solving the above problems and defects is:
(1) how to solve the problem of matrix sparsity while fully extracting consumption features and social features;
(2) how to reasonably characterize the interaction relationships of consumption features and social features;
(3) how to characterize personalized differences and reasonably introduce them into a joint recommendation system.
The significance of solving the problems and the defects is as follows: according to the analysis, the defects of the existing joint recommendation system have great influence on the accuracy of the recommendation system, so that the accuracy of the joint recommendation system is further improved by solving the problems.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides a social and consumption joint recommendation system, method, storage medium and computer equipment.
The invention is realized in such a way, and provides a social and consumption joint recommendation method, which comprises the following steps:
the graph neural network and the self-attention mechanism extract consumption preference characteristics from the input rating matrix R;
extracting social preference features from the input social matrix S by the graph volume network and the self-attention mechanism;
introducing a reciprocity graph neural network, simulating a symbiotic mechanism of a biological world, and combining the consumption preference characteristic and the social preference characteristic to obtain a joint consumption characteristic and a joint social characteristic;
and inputting data of the user needing prediction, the possible consumption items and the persons who are possible to establish the relationship, and predicting the potential consumption possibility and social possibility of the user.
Further, the graph neural network and the self-attention mechanism extract consumption preference characteristics from the input rating matrix R, and firstly, the rating matrix R is converted into the graph neural network; secondly, two non-Euclidean subgraphs were isolated from the transformed network: an item preference sub-graph and a user preference sub-graph; then, respectively extracting the item preference and the user preference characteristics from the data; finally, the two are combined into a consumption preference feature.
Further, the graph convolution network and the self-attention mechanism extract social preference features from an input social matrix S, and firstly, a regularized Laplacian matrix is constructed; then, constructing a convolution kernel of a spectrogram convolution network through the eigenvalue matrix and the eigenvector matrix of the matrix, and extracting social characteristics of each person; and finally, aggregating social characteristics of all friends of a certain user to obtain social preference characteristics.
Furthermore, the combined consumption feature and the combined social feature are obtained by combining the consumption preference feature and the social preference feature, an element wise mode is adopted, the consumption preference feature and the social preference feature are multiplied in an alignment mode to obtain a reciprocal feature, then the consumption preference feature is cascaded with the reciprocal feature to obtain a combined consumption feature, and the social preference feature is cascaded with the reciprocal feature to obtain a combined social feature.
Further, predicting potential consumption possibility and social possibility of the user, multiplying the combined consumption characteristics and the attributes of the goods to be predicted to obtain probabilities, performing descending ordering on the probabilities of all the goods to be predicted, taking top k as a prediction result, or taking an empirical value of 0.5 as a threshold value, and taking the consumption probabilities larger than 0.5 as the prediction results; similarly, the social characteristics are multiplied by the attributes of the people to be predicted to obtain the probability, and finally top k or data larger than the threshold value by 0.5 is taken as a result.
It is a further object of the invention to provide a computer device comprising a memory and a processor, the memory storing a computer program which, when executed by the processor, causes the processor to perform the steps of:
the graph neural network and the self-attention mechanism extract consumption preference characteristics from the input rating matrix R;
extracting social preference features from the input social matrix S by the graph volume network and the self-attention mechanism;
introducing a reciprocity graph neural network, simulating a symbiotic mechanism of a biological world, and combining the consumption preference characteristic and the social preference characteristic to obtain a joint consumption characteristic and a joint social characteristic;
and inputting data of the user needing prediction, the possible consumption items and the persons who are possible to establish the relationship, and predicting the potential consumption possibility and social possibility of the user.
It is another object of the present invention to provide a computer-readable storage medium storing a computer program which, when executed by a processor, causes the processor to perform the steps of:
the graph neural network and the self-attention mechanism extract consumption preference characteristics from the input rating matrix R;
extracting social preference features from the input social matrix S by the graph volume network and the self-attention mechanism;
introducing a reciprocity graph neural network, simulating a symbiotic mechanism of a biological world, and combining the consumption preference characteristic and the social preference characteristic to obtain a joint consumption characteristic and a joint social characteristic;
and inputting data of the user needing prediction, the possible consumption items and the persons who are possible to establish the relationship, and predicting the potential consumption possibility and social possibility of the user.
The invention also aims to provide an information data processing terminal, which is used for realizing the social and consumption joint recommendation method.
Another object of the present invention is to provide a social and consumption joint recommendation system for implementing the social and consumption joint recommendation method, the social and consumption joint recommendation system comprising:
the self-attention space layer is used for extracting consumption preference characteristics of the user from the rating matrix R;
the self-attention spectrum layer is used for extracting social preference characteristics of the user from the social matrix S;
and the reciprocal analysis layer is inspired by the symbiotic model, takes the consumption preference characteristics and the social preference characteristics as input, and aims to perform self-adaptive modeling on the mutual relation between the consumption preference of the user and the social connection to generate joint consumption characteristics and joint social characteristics.
And the prediction layer is used for applying the joint consumption characteristics and the joint social characteristics to the prediction layer so as to predict potential consumption and social possibility.
By combining all the technical schemes, the invention has the advantages and positive effects that: the graph neural network replaces the introduction of an adjacency matrix and a spectrogram convolution network, and the limitation of sparse matrix decomposition is solved; the introduction of the self-attention model fully considers the individual difference, so that the characteristic attributes extracted through the first two layers are more fit with the real attributes of the user; the introduction of a reciprocal mechanism fully exerts the interactivity of the original information of the two recommendation systems, and improves the accuracy, recall rate and NDCG index of the recommendation of the prediction layer.
For consumption recommendation tasks, the model obtains better recommendation performance in consumption prediction and reaches the highest value in the aspects of accuracy, recall rate and NDCG. Compared with the best methods DANSER and HERS at present, the accuracy, the recall rate and the NDCG are respectively improved by at least 5.70 percent and 3.23 percent, 7.31 percent and 6.02 percent, 5.67 percent and 3.39 percent.
For the social recommendation task, the model obtains better recommendation performance in social prediction, and the precision, the recall rate and the NDCG are averagely 14.15%, 19.95% and 11.58% higher than those of other methods. Compared with the best method at present, namely Attenwalk and MGNN, the accuracy, recall rate and NDCG are increased by at least 3.98 percent and 5.43 percent, 6.43 percent and 9.28 percent, 3.50 percent and 6.78 percent respectively.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the embodiments of the present application will be briefly described below, and it is obvious that the drawings described below are only some embodiments of the present application, and it is obvious for those skilled in the art that other drawings can be obtained from the drawings without creative efforts.
Fig. 1 is a flowchart of a social and consumption joint recommendation method according to an embodiment of the present invention.
FIG. 2 is a schematic structural diagram of a social and consumption joint recommendation system according to an embodiment of the present invention;
in fig. 2: 1. a self-attention space layer; 2. a self-attention spectral slice; 3. a reciprocal analysis layer; 4. and predicting the layer.
FIG. 3 is a schematic diagram of a social and consumption joint recommendation system according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail with reference to the following embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
In view of the problems in the prior art, the present invention provides a social and consumption joint recommendation system, method, storage medium, and computer device, and the following describes the present invention in detail with reference to the accompanying drawings.
As shown in fig. 1, the social and consumption joint recommendation method provided by the present invention includes the following steps:
s101: introducing a graph neural network and a self-attention mechanism, and extracting consumption preference characteristics from an input rating matrix R;
s102: introducing a graph convolution network and a self-attention mechanism, and extracting social preference characteristics from the input social matrix S;
s103: introducing a reciprocity graph neural network, simulating a symbiotic mechanism of a biological world, and combining the consumption preference characteristic and the social preference characteristic to obtain a joint consumption characteristic and a joint social characteristic;
s104: and inputting data of the user needing prediction, the possible consumption items and the persons who are possible to establish the relationship, and predicting the potential consumption possibility and social possibility of the user.
Persons of ordinary skill in the art can also implement the social and consumption joint recommendation method provided by the present invention by using other steps, and the social and consumption joint recommendation method provided by the present invention in fig. 1 is only one specific embodiment.
As shown in fig. 2, the social and consumption joint recommendation system provided by the present invention includes:
and the self-attention space layer 1 is used for extracting the consumption preference characteristics of the user from the rating matrix R.
And the self-attention spectrum layer 2 is used for extracting the social preference characteristics of the user from the social network S.
And the reciprocal analysis layer 3 is inspired by the symbiotic model, takes the consumption preference characteristics and the social preference characteristics as input, and aims to perform self-adaptive modeling on the mutual relation between the consumption preference of the user and the social connection to generate joint consumption characteristics and joint social characteristics.
And the prediction layer 4 is used for applying the reciprocal preference attribute and the reciprocal social attribute to the prediction layer so as to predict potential consumption and social possibility.
The technical solution of the present invention is further described below with reference to the accompanying drawings.
The invention analogizes the symbiosis phenomenon of the biological world with two recommendation systems with potential association, migrates the symbiosis mechanism to fully utilize the interactivity to mine the potential connection between data, and falls on the ground of a social and consumption joint recommendation method, comprising four components: from the attention space layer 1, the attention spectrum layer 2, the reciprocal analysis layer 3 and the prediction layer 4, new ideas and possibilities are injected into the recommendation system field.
As shown in fig. 3, which is a floor detail of the joint recommendation system, the social and consumption joint recommendation system based on the reciprocal graph neural network is used for processing the input rating matrix R and the social network matrix S, and further predicting the potential consumption possibility and social possibility of the user.
The self-attention space layer 1 is intended to extract the consumption preference characteristics of the user from the rating matrix R. For this reason, the rating matrix is firstly converted into a graph neural network, so that the problem that the sparse matrix is insensitive to feature extraction is solved; secondly, two non-Euclidean subgraphs were isolated from the transformed network: an item preference sub-graph and a user preference sub-graph; then introducing a self-attention model, and respectively extracting item preference and user preference characteristics from the self-attention model; finally, the two are combined into a consumption preference feature.
The self-attention spectrum layer 2 aims to extract social preference features of the user from the social network S. The core framework of the layer is a spectrogram convolution network. Firstly, constructing a regularized Laplacian matrix, then constructing a convolution kernel of a spectrogram convolution network through an eigenvalue matrix and an eigenvector matrix of the matrix, extracting social characteristics of each person, and finally, introducing a self-attention model, aggregating social characteristics of all friends of a certain user and obtaining social preference characteristics.
The reciprocity analysis layer 3 is inspired by the symbiosis phenomenon, introduces a self-attention model and a symbiosis reciprocity model, and obtains a reciprocity characteristic by carrying out counterpoint multiplication on the consumption preference characteristic and the social preference characteristic in an element wise mode; the consumption preference characteristics are cascaded with the reciprocal characteristics to obtain combined consumption characteristics, the social preference characteristics are cascaded with the reciprocal characteristics to obtain combined social characteristics, and the whole framework adopts a multilayer perceptron idea to fully represent the interactivity of the two characteristics.
The prediction layer 4 requires input of data of users who need prediction, consumer goods that may occur, and people who may establish relationships, making predictions of potential consumer and social possibilities for the users. Multiplying the combined consumption characteristics and the attributes of the articles to be predicted to obtain probabilities, performing descending ordering on the probabilities of all the articles to be predicted, taking top k as a prediction result, or taking an empirical value of 0.5 as a threshold value, and taking the consumption probabilities larger than 0.5 as the prediction results; similarly, the social characteristics are multiplied by the attributes of the people to be predicted to obtain the probability, and finally top k or data larger than the threshold value by 0.5 is taken as a result.
The technical effects of the present invention will be described in detail with reference to experiments.
The model is realized through a famous open source deep learning platform Pythrch in the experiment, and hardware is based on a Titan Xp GPU. To test the performance of the model, the performance of the model and the existing methods were evaluated using 4 sets of true data set cross validation. In each run, 75% of the rating and social matrices were randomly drawn as a training set, with the remaining 25% used as tests. And selecting the performance of the data set to be evaluated according to the precision, the recall rate and the NDCG index.
For the consumption recommendation task, the result shows that the model obtains better recommendation performance in consumption prediction and reaches the highest value in the aspects of precision, recall rate and NDCG. Compared with the best methods DANSER and HERS at present, the accuracy, the recall rate and the NDCG are respectively improved by at least 5.70 percent and 3.23 percent, 7.31 percent and 6.02 percent, 5.67 percent and 3.39 percent.
For the social recommendation task, the result shows that the model obtains better recommendation performance in social prediction, and the accuracy, recall rate and NDCG are averagely 14.15%, 19.95% and 11.58% higher than those of other methods. Compared with the best method at present, namely Attenwalk and MGNN, the accuracy, recall rate and NDCG are increased by at least 3.98 percent and 5.43 percent, 6.43 percent and 9.28 percent, 3.50 percent and 6.78 percent respectively.
It should be noted that the embodiments of the present invention can be realized by hardware, software, or a combination of software and hardware. The hardware portion may be implemented using dedicated logic; the software portions may be stored in a memory and executed by a suitable instruction execution system, such as a microprocessor or specially designed hardware. Those skilled in the art will appreciate that the apparatus and methods described above may be implemented using computer executable instructions and/or embodied in processor control code, such code being provided on a carrier medium such as a disk, CD-or DVD-ROM, programmable memory such as read only memory (firmware), or a data carrier such as an optical or electronic signal carrier, for example. The apparatus and its modules of the present invention may be implemented by hardware circuits such as very large scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., or by software executed by various types of processors, or by a combination of hardware circuits and software, e.g., firmware.
The above description is only for the purpose of illustrating the present invention and the appended claims are not to be construed as limiting the scope of the invention, which is intended to cover all modifications, equivalents and improvements that are within the spirit and scope of the invention as defined by the appended claims.
Claims (9)
1. A social and consumption joint recommendation method is characterized by comprising the following steps:
the graph neural network and the self-attention mechanism extract consumption preference characteristics from the input rating matrix R;
extracting social preference features from the input social matrix S by the graph volume network and the self-attention mechanism;
introducing a reciprocity graph neural network, simulating a symbiotic mechanism of a biological world, and combining the consumption preference characteristic and the social preference characteristic to obtain a joint consumption characteristic and a joint social characteristic;
and inputting data of the user needing prediction, the possible consumption items and the persons who are possible to establish the relationship, and predicting the potential consumption possibility and social possibility of the user.
2. The social and consumption combined recommendation method according to claim 1, wherein the graph neural network and the self-attention mechanism extract consumption preference features from an input rating matrix R, and first convert the rating matrix R into the graph neural network; secondly, two non-Euclidean subgraphs were isolated from the transformed network: an item preference sub-graph and a user preference sub-graph; then, respectively extracting the item preference and the user preference characteristics from the data; finally, the two are combined into a consumption preference feature.
3. The social and consumption joint recommendation method of claim 1, wherein the graph convolution network and the self-attention mechanism extract social preference features from the input social matrix S, and first, a regularized Laplacian matrix is constructed; then, constructing a convolution kernel of a spectrogram convolution network through the eigenvalue matrix and the eigenvector matrix of the matrix, and extracting social characteristics of each person; and finally, aggregating social characteristics of all friends of a certain user to obtain social preference characteristics.
4. The social and consumption combined recommendation method according to claim 1, wherein the consumption preference feature and the social preference feature are combined to obtain a combined consumption feature and a combined social feature, an element wise manner is adopted, the consumption preference feature and the social preference feature are subjected to counterpoint multiplication to obtain a reciprocal feature, then the consumption preference feature is cascaded with the reciprocal feature to obtain a combined consumption feature, and the social preference feature is cascaded with the reciprocal feature to obtain a combined social feature.
5. The social and consumption combined recommendation method according to claim 1, wherein the potential consumption possibility and the social possibility of the user are predicted, the combined consumption characteristics are multiplied by the attributes of the items to be predicted to obtain probabilities, the probabilities of all the items to be predicted are sorted in a descending manner, top k is taken as a prediction result, or an empirical value 0.5 is taken as a threshold value, and the consumption probabilities larger than 0.5 are both taken as prediction results; similarly, the social characteristics are multiplied by the attributes of the people to be predicted to obtain the probability, and finally top k or data larger than the threshold value by 0.5 is taken as a result.
6. A computer device, characterized in that the computer device comprises a memory and a processor, the memory storing a computer program which, when executed by the processor, causes the processor to carry out the steps of:
the graph neural network and the self-attention mechanism extract consumption preference characteristics from the input rating matrix R;
extracting social preference features from the input social matrix S by the graph volume network and the self-attention mechanism;
introducing a reciprocity graph neural network, simulating a symbiotic mechanism of a biological world, and combining the consumption preference characteristic and the social preference characteristic to obtain a joint consumption characteristic and a joint social characteristic;
and inputting data of the user needing prediction, the possible consumption items and the persons who are possible to establish the relationship, and predicting the potential consumption possibility and social possibility of the user.
7. A computer-readable storage medium storing a computer program which, when executed by a processor, causes the processor to perform the steps of:
the graph neural network and the self-attention mechanism extract consumption preference characteristics from the input rating matrix R;
extracting social preference features from the input social matrix S by the graph volume network and the self-attention mechanism;
introducing a reciprocity graph neural network, simulating a symbiotic mechanism of a biological world, and combining the consumption preference characteristic and the social preference characteristic to obtain a joint consumption characteristic and a joint social characteristic;
and inputting data of the user needing prediction, the possible consumption items and the persons who are possible to establish the relationship, and predicting the potential consumption possibility and social possibility of the user.
8. An information data processing terminal, characterized in that the information data processing terminal is used for realizing the social and consumption joint recommendation method of any one of claims 1 to 5.
9. A social and consumption joint recommendation system for implementing the social and consumption joint recommendation method according to any one of claims 1 to 5, the social and consumption joint recommendation system comprising:
the self-attention space layer is used for extracting consumption preference characteristics of the user from the rating matrix R;
the self-attention spectrum layer is used for extracting social preference characteristics of the user from the social matrix S;
the mutual benefit analysis layer is inspired by the symbiotic model, takes the consumption preference characteristic and the social preference characteristic as input, and aims to perform self-adaptive modeling on the mutual relation between the consumption preference of the user and the social connection to generate a joint consumption characteristic and a joint social characteristic;
and the prediction layer is used for applying the joint consumption characteristics and the joint social characteristics to the prediction layer so as to predict potential consumption and social possibility.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011218535.8A CN112445981A (en) | 2020-11-04 | 2020-11-04 | Social and consumption joint recommendation system, method, storage medium and computer equipment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011218535.8A CN112445981A (en) | 2020-11-04 | 2020-11-04 | Social and consumption joint recommendation system, method, storage medium and computer equipment |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112445981A true CN112445981A (en) | 2021-03-05 |
Family
ID=74736856
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011218535.8A Pending CN112445981A (en) | 2020-11-04 | 2020-11-04 | Social and consumption joint recommendation system, method, storage medium and computer equipment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112445981A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113158088A (en) * | 2021-04-16 | 2021-07-23 | 桂林电子科技大学 | Position recommendation method based on graph neural network |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109767301A (en) * | 2019-01-14 | 2019-05-17 | 北京大学 | Recommended method and system, computer installation, computer readable storage medium |
US10356035B1 (en) * | 2013-03-15 | 2019-07-16 | Cox Communications, Inc | Recommendation system based on common interests in social networks |
CN110503531A (en) * | 2019-08-30 | 2019-11-26 | 中国科学技术大学 | The dynamic social activity scene recommended method of timing perception |
CN111428147A (en) * | 2020-03-25 | 2020-07-17 | 合肥工业大学 | Social recommendation method of heterogeneous graph volume network combining social and interest information |
CN111737537A (en) * | 2020-07-21 | 2020-10-02 | 杭州欧若数网科技有限公司 | POI recommendation method, device and medium based on graph database |
CN111881363A (en) * | 2020-06-23 | 2020-11-03 | 北京工业大学 | Recommendation method based on graph interaction network |
-
2020
- 2020-11-04 CN CN202011218535.8A patent/CN112445981A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10356035B1 (en) * | 2013-03-15 | 2019-07-16 | Cox Communications, Inc | Recommendation system based on common interests in social networks |
CN109767301A (en) * | 2019-01-14 | 2019-05-17 | 北京大学 | Recommended method and system, computer installation, computer readable storage medium |
CN110503531A (en) * | 2019-08-30 | 2019-11-26 | 中国科学技术大学 | The dynamic social activity scene recommended method of timing perception |
CN111428147A (en) * | 2020-03-25 | 2020-07-17 | 合肥工业大学 | Social recommendation method of heterogeneous graph volume network combining social and interest information |
CN111881363A (en) * | 2020-06-23 | 2020-11-03 | 北京工业大学 | Recommendation method based on graph interaction network |
CN111737537A (en) * | 2020-07-21 | 2020-10-02 | 杭州欧若数网科技有限公司 | POI recommendation method, device and medium based on graph database |
Non-Patent Citations (1)
Title |
---|
YANG XIAO: "MGNN: Mutualistic Graph Neural Network for Joint Friend and Item Recommendation", 《 IEEE INTELLIGENT SYSTEMS》 * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113158088A (en) * | 2021-04-16 | 2021-07-23 | 桂林电子科技大学 | Position recommendation method based on graph neural network |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Fang et al. | Exploring deep neural networks via layer-peeled model: Minority collapse in imbalanced training | |
CN111914569B (en) | Fusion map-based prediction method and device, electronic equipment and storage medium | |
Kim et al. | Deep hybrid recommender systems via exploiting document context and statistics of items | |
CN110717098B (en) | Meta-path-based context-aware user modeling method and sequence recommendation method | |
CN111797321B (en) | Personalized knowledge recommendation method and system for different scenes | |
Jamil et al. | A literature survey of benchmark functions for global optimisation problems | |
Zhang et al. | Enabling kernel-based attribute-aware matrix factorization for rating prediction | |
WO2022016556A1 (en) | Neural network distillation method and apparatus | |
Puccetti et al. | Technology identification from patent texts: A novel named entity recognition method | |
CN112529071B (en) | Text classification method, system, computer equipment and storage medium | |
CN113761359B (en) | Data packet recommendation method, device, electronic equipment and storage medium | |
CN113190702B (en) | Method and device for generating information | |
KR101635283B1 (en) | Method for analyzing data based on matrix factorization model and apparatus therefor | |
Xia et al. | Modeling consumer buying decision for recommendation based on multi-task deep learning | |
Gu et al. | Robust weighted SVD-type latent factor models for rating prediction | |
CN112328909A (en) | Information recommendation method and device, computer equipment and medium | |
Najar et al. | Exact fisher information of generalized dirichlet multinomial distribution for count data modeling | |
CN110826327A (en) | Emotion analysis method and device, computer readable medium and electronic equipment | |
CN112445981A (en) | Social and consumption joint recommendation system, method, storage medium and computer equipment | |
CN113254788B (en) | Big data based recommendation method and system and readable storage medium | |
CN115618079A (en) | Session recommendation method, device, electronic equipment and storage medium | |
Rana et al. | Exploring numerical calculations with calcnet | |
CN116501993B (en) | House source data recommendation method and device | |
CN116703498B (en) | Commodity recommendation method and device, electronic equipment and storage medium | |
CN115062230B (en) | Data processing method, apparatus, device, medium, and program product |
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: 20210305 |
|
RJ01 | Rejection of invention patent application after publication |