CN115757528A - Information recommendation method and device - Google Patents
Information recommendation method and device Download PDFInfo
- Publication number
- CN115757528A CN115757528A CN202310014861.4A CN202310014861A CN115757528A CN 115757528 A CN115757528 A CN 115757528A CN 202310014861 A CN202310014861 A CN 202310014861A CN 115757528 A CN115757528 A CN 115757528A
- Authority
- CN
- China
- Prior art keywords
- information
- data
- information data
- piece
- pieces
- 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 55
- 238000012545 processing Methods 0.000 claims description 67
- 238000003860 storage Methods 0.000 claims description 18
- 230000006870 function Effects 0.000 claims description 14
- 230000008569 process Effects 0.000 claims description 10
- 230000000052 comparative effect Effects 0.000 claims description 8
- 238000000605 extraction Methods 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 7
- 230000002776 aggregation Effects 0.000 claims description 4
- 238000004220 aggregation Methods 0.000 claims description 4
- 230000007246 mechanism Effects 0.000 claims description 4
- 238000012549 training Methods 0.000 claims description 4
- 230000005012 migration Effects 0.000 claims description 3
- 238000013508 migration Methods 0.000 claims description 3
- 238000004590 computer program Methods 0.000 description 11
- 238000010586 diagram Methods 0.000 description 10
- 238000005516 engineering process Methods 0.000 description 8
- 230000003993 interaction Effects 0.000 description 7
- 230000002452 interceptive effect Effects 0.000 description 6
- 230000008901 benefit Effects 0.000 description 5
- 230000000694 effects Effects 0.000 description 5
- 238000013528 artificial neural network Methods 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 239000013598 vector Substances 0.000 description 3
- 238000013473 artificial intelligence Methods 0.000 description 2
- 238000007405 data analysis Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000014759 maintenance of location Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 230000001902 propagating effect Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 238000009423 ventilation Methods 0.000 description 1
Images
Abstract
The invention discloses an information recommendation method and device, relates to the technical field of rail transit and the technical field of computers, and mainly aims to solve the problem that the conventional cross-domain recommendation system only focuses on information which is commonly interested among a plurality of users and ignores information which is independently interested by each user. The main technical scheme of the invention is as follows: acquiring a plurality of pieces of information data through a preset interface, wherein the information data come from a plurality of information systems, and the preset interface is an interface between a signal system and the plurality of information systems; determining a plurality of characteristic data corresponding to each piece of information data; comparing and learning a plurality of feature data corresponding to each piece of information data to obtain global feature data corresponding to each piece of information data; cross learning is carried out on global feature data among the information data to obtain target feature data; and determining target information data from the plurality of pieces of information data based on the plurality of pieces of target characteristic data, and outputting the target information data as an information recommendation result.
Description
Technical Field
The invention relates to the technical field of rail transit and the technical field of computers, in particular to an information recommendation method and device.
Background
Train arrival information is sometimes obtained by application when passengers take rail vehicles, but with the development of technology, passengers are not satisfied with obtaining only arrival information, but are gradually enriched in various aspects. However, a recommendation system for recommending other information does not exist in the technical field of rail transit at present, so that the recommendation system and the rail transit technology need to be fused, but because the data volume of the traditional recommendation system based on artificial intelligence and big data analysis is too small and the information recommendation effect is not ideal, the cross-domain recommendation system and the rail transit technology are fused.
Although the existing cross-domain recommendation system combines data of multiple fields, the data volume is rich, and the information recommendation effect is comprehensive compared with that of the traditional recommendation system. However, the current cross-domain recommendation system only focuses on information which is commonly interested among a plurality of users, and then recommends the common information to other users, and ignores the information which is individually interested by each user.
Disclosure of Invention
In view of the above problems, the present invention provides an information recommendation method and apparatus, and mainly aims to solve the problem that the current cross-domain recommendation system only focuses on information commonly interested among multiple users, and then recommends the common information to other users, but ignores information that is individually interested by each user.
In order to solve the technical problems, the invention provides the following scheme:
in a first aspect, the present invention provides an information recommendation method, where the method includes:
acquiring a plurality of pieces of information data through a preset interface, wherein the plurality of pieces of information data are from a plurality of information systems in rail transit, and the preset interface is an interface between a signal system and the plurality of information systems in the rail transit;
determining a plurality of characteristic data corresponding to each piece of information data;
comparing and learning the plurality of feature data corresponding to each piece of information data to obtain global feature data corresponding to each piece of information data;
performing cross learning processing on global feature data among the information data to obtain target feature data;
and determining target information data in the plurality of pieces of information data based on the plurality of pieces of target characteristic data, and outputting the target information data as an information recommendation result.
In a second aspect, the present invention provides an information recommendation apparatus, including:
the system comprises an acquisition unit, a processing unit and a processing unit, wherein the acquisition unit is used for acquiring a plurality of pieces of information data through a preset interface, the plurality of pieces of information data are sourced from a plurality of information systems in the rail transit, and the preset interface is an interface between a signal system and the plurality of information systems in the rail transit;
a determining unit configured to determine a plurality of feature data corresponding to each piece of information data acquired by the acquiring unit;
the first processing unit is used for performing comparison learning processing on a plurality of feature data corresponding to each piece of information data determined by the determining unit to obtain global feature data corresponding to each piece of information data;
the second processing unit is used for performing cross learning processing on global feature data among the plurality of pieces of information data obtained by the first processing unit to obtain a plurality of pieces of target feature data;
and the result output unit is used for determining target information data in the plurality of pieces of information data based on the plurality of pieces of target characteristic data processed by the second processing unit and outputting the target information data as an information recommendation result.
In order to achieve the above object, according to a third aspect of the present invention, there is provided a storage medium including a stored program, wherein when the program runs, a device on which the storage medium is located is controlled to execute the information recommendation method of the first aspect.
In order to achieve the above object, according to a fourth aspect of the present invention, there is provided a processor for executing a program, wherein the program executes to perform the information recommendation method of the first aspect.
By means of the technical scheme, the information recommendation method and the information recommendation device provided by the invention can firstly acquire a plurality of pieces of information data from a plurality of information systems through the preset interfaces between the signal systems and the plurality of information systems in rail transit, and then determine a plurality of feature data corresponding to each piece of information data; and then comparing and learning the plurality of characteristic data corresponding to each piece of information data, so that the plurality of characteristic data of each piece of information data are learned mutually, each characteristic can learn the characteristic which does not exist in other characteristics, and the global characteristic data corresponding to each piece of information data which is more evenly and abundantly distributed can be obtained. After the global feature data corresponding to each piece of information data is obtained, the global feature data among the pieces of information data can be subjected to cross learning processing, so that the data of each piece of information data can be more balanced and enriched, the features of other pieces of information data are fused, the finally obtained pieces of target feature data are more enriched and reliable, and finally, a final information recommendation result is obtained according to the pieces of target feature data. Compared with the prior art, the recommendation method and the recommendation system have the advantages that the common characteristics among the information data are reserved to the maximum extent, and the maximization of the characteristic data corresponding to each information data is reserved to the maximum extent.
The above description is only an overview of the technical solutions of the present invention, and the present invention can be implemented in accordance with the content of the description so as to make the technical means of the present invention more clearly understood, and the above and other objects, features, and advantages of the present invention will be more clearly understood.
Drawings
Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention. Also, like reference numerals are used to refer to like parts throughout the drawings. In the drawings:
fig. 1 shows a flowchart of an information recommendation method according to an embodiment of the present invention;
FIG. 2 is a flow chart of another information recommendation method provided by the embodiment of the invention;
fig. 3 is a block diagram illustrating components of an information recommendation apparatus according to an embodiment of the present invention;
fig. 4 is a block diagram illustrating another information recommendation apparatus according to an embodiment of the present invention.
Detailed Description
Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.
Train arrival information is sometimes obtained by application when passengers take rail vehicles, but with the development of technology, passengers are not satisfied with obtaining only arrival information, but are gradually enriched in various aspects. However, a recommendation system for recommending other information does not exist in the technical field of rail transit at present, so that the recommendation system and the rail transit technology need to be fused, but because the data volume of the traditional recommendation system based on artificial intelligence and big data analysis is too small and the information recommendation effect is not ideal, the cross-domain recommendation system and the rail transit technology are fused. Although the current cross-domain recommendation system combines data of a plurality of fields, the data volume is rich, and the information recommendation effect is comprehensive compared with that of the traditional recommendation system. However, the current cross-domain recommendation system only focuses on information which is commonly interested among a plurality of users, and then recommends the common information to other users, and ignores information which is individually interested by each user. Therefore, the invention provides an information recommendation method, which can solve the problems in the prior art, and the specific execution steps are shown in fig. 1 and comprise the following steps:
101. and acquiring a plurality of pieces of information data through a preset interface.
The information data are from a plurality of information systems in rail transit, and the information data in different information systems are different in type. The preset interface is an interface between the signal system and the plurality of information systems.
Because the signal system in the rail transit and other information systems in the rail transit are provided with communication interfaces, the signal system can collect data in the other information systems, and therefore the invention can collect a plurality of pieces of information data from a plurality of information systems through the preset interfaces between the signal system and the other information systems, and the quantity of the collected data is wide.
In addition, when the signal system acquires information data in a plurality of information systems through the preset interface, scores submitted by the user aiming at specific item information in each information system can be acquired through the preset interface, the interest degree of the user on the specific items in the system can be known through the scores, and the interest degree can also be embodied as the interaction times between the user and information corresponding to the specific items. That is, the information data may be a record of an interaction between a user and item information in an information system.
Wherein the information system includes a comprehensive monitoring system, a vehicle system, a passenger information system, a broadcasting system, an advertising system, a network information system, a geographic information system, a government information system, an entertainment system, etc., and the information system mentioned in the present invention includes, but is not limited to, the above information systems.
Illustratively, the information data in the integrated monitoring system can be interactive records of passengers on station temperature, humidity, ventilation and the like by combining the sources of the information; the information data in the vehicle system can be interactive records of the temperature of the passenger for the carriage, the congestion degree of the carriage and the like; the information data in the passenger information system can be interactive records of passengers on arrival information, early-late point information, vehicle stop-and-go information, first-and-last-class information and the like of the vehicles; the information data in the broadcasting system can be interactive records of passengers on station broadcasting information and the like; the information data in the advertising system may include records of passenger interactions with commercial information and the like; the information data in the entertainment system, the geographic information system and the government affair information system can be the interactive records of the passengers on the peripheral information, the entertainment information, the government affair information distribution and the like of the station. Data in other systems are not listed here.
102. And determining a plurality of characteristic data corresponding to each piece of information data.
In this step, since there are a plurality of pieces of collected information data, in order to make the subsequent recommendation result as comprehensive as possible, it is necessary to determine a plurality of feature data corresponding to each piece of information data for each piece of information data.
Specifically, when a plurality of feature data corresponding to each piece of information data are determined, each piece of information data may be input into a trained model, and the model may use a convolutional neural network or a graph neural network.
For example, if a piece of information data comes from the entertainment system, the name of a song that a user frequently listens to: the extracted information data may correspond to a plurality of feature data such as "game", "mythical story", and the like.
103. And performing comparison learning processing on the plurality of feature data corresponding to each piece of information data to obtain global feature data corresponding to each piece of information data.
In this step, after the plurality of feature data corresponding to each piece of information data are extracted, the plurality of feature data may be compared and learned, which may also be understood as performing various combinations among the plurality of feature data to enrich the corresponding information data, and finally obtaining the global feature data corresponding to each piece of data.
Specifically, during the comparative learning, the plurality of feature data corresponding to each piece of information data may be embodied in the form of a graph data structure, and then each piece of feature data is taken as a starting point to randomly walk in the graph structure, so that different walking paths can generate different path graphs, the plurality of graphs can be subjected to comparative learning, and finally, more balanced global feature data corresponding to each piece of information data can be obtained.
104. And performing cross learning processing on the global feature data among the plurality of pieces of information data to obtain a plurality of pieces of target feature data.
In the step, after the global feature data in each piece of information data is obtained, the global feature data in each piece of information data is processed through a fully-connected neural network preset in a signal system, a nonlinear relation between a user and data corresponding to each piece of information data is obtained, and therefore stable information input is provided for an information output layer.
Specifically, since the information data may be an interaction record of a user for a specific item, the cross-learning process may be performed by combining the features of the user in a specific domain and the weights between the features of multiple users in common in multiple domains.
105. And determining target information data in the plurality of pieces of information data based on the plurality of pieces of target characteristic data, and outputting the target information data as an information recommendation result.
In this step, after the plurality of pieces of target feature data are obtained, the plurality of pieces of target data may be processed to determine target information data that matches the plurality of pieces of target feature data among the plurality of pieces of information data, and the target information data may be recommended to the user as a final information recommendation result.
The target characteristic data is obtained by combining the global characteristic data of each piece of information data for cross learning, so that the obtained target characteristic data is more comprehensive, and the finally presented information recommendation result is more balanced.
Illustratively, the recommendation result may be station periphery information, car temperature, or the like.
Based on the implementation manner shown in fig. 1, it can be seen that, according to the information recommendation method provided by the present invention, a plurality of pieces of information data from a plurality of information systems can be obtained through preset interfaces between a signal system and the plurality of information systems in rail transit, and the types of the information data in each information system are different from each other, and then a plurality of feature data corresponding to each information data are determined; and then, comparing and learning the plurality of characteristic data corresponding to each piece of information data, so that the plurality of characteristic data are learned mutually, each characteristic can learn the characteristics which are not possessed by other characteristics, and the global characteristic data corresponding to each piece of information data which are more evenly and abundantly distributed can be obtained. After the global features corresponding to each piece of information data are obtained, the global feature data among the information data can be subjected to cross learning processing, so that the data of each piece of information data can be more balanced and enriched, the features of other information data are fused, the finally obtained target feature data are more abundant and reliable, and finally the final information recommendation result is obtained according to the target feature data. Compared with the prior art, the recommendation method and the recommendation system have the advantages that the common characteristics among the information data are reserved to the maximum extent, and the maximization of the characteristic data corresponding to each information data is reserved to the maximum extent.
Further, as a refinement and an extension of the embodiment shown in fig. 1, an embodiment of the present invention further provides another information recommendation method, as shown in fig. 2, which includes the following specific steps:
201. and acquiring a plurality of pieces of information data through a preset interface.
The implementation manner of step 201 is the same as that of step 101, and the same technical effect can be achieved, so as to solve the same technical problem, which is not repeated herein.
202. And determining a plurality of characteristic data corresponding to each piece of information data.
In the invention, a trained information recommendation model is deployed in a signal system in advance, wherein the information recommendation model comprises a feature extraction layer and is used for extracting feature data of information data and performing comparison learning and other work among a plurality of feature data; and the characteristic migration layer is used for performing cross learning work among characteristic data.
Therefore, in the invention, each piece of information data can be input into the feature extraction layer in the information recommendation model, and a plurality of feature data corresponding to each piece of information data can be extracted.
203. In a feature extraction layer in a pre-trained information recommendation model, a uniform noise processing mode is utilized to perform enhancement processing on a plurality of feature data corresponding to each piece of information data, and a plurality of enhanced feature data corresponding to each piece of information data are obtained.
204. And performing comparative learning processing on the plurality of enhanced feature data corresponding to each piece of information data to obtain global feature data corresponding to each piece of information data.
In steps 203 to 204, after obtaining a plurality of feature data corresponding to each piece of information data, enhancement processing may be performed on the feature data, and during the enhancement processing, random noise may be added to each piece of feature data, and then a plurality of feature enhancement data corresponding to each piece of information data is obtained, where a specific formula may be:
Where H (-) is an encoder function that encodes connection information to represent learning. δ ^ 'and δ ^' are randomly generated noise vectors. The LightGCN is adopted as a graph encoder to perform convolution operation on a plurality of input feature data, capture the feature data preference of nodes by propagating information between feature data of adjacent nodes and amplify the influence of variance.
Then, the multiple enhanced feature data corresponding to each piece of information data may be subjected to comparative learning processing, so as to obtain global feature data corresponding to each piece of information data.
Specifically, because a plurality of enhanced feature data corresponding to each piece of information data can be embodied in a graph structure, each piece of feature enhanced data randomly walks in the graph structure as a starting point, and because a plurality of walking paths exist, a plurality of views corresponding to each piece of feature enhanced data can be generated, and then consistency of different views corresponding to each piece of feature enhanced data and divergence between different nodes are realized by using a contrast loss function InfoNCE, so as to complete comparison learning work among a plurality of enhanced feature data corresponding to each piece of information data. The comparison loss InfoNCE formula is as follows:
Where i, j are users/items in a sample batch. τ >0 is a hyper-parameter, referred to as temperature in softmax. This allows similar items to be brought closer together after data enhancement and dissimilar data to be pushed away.
Graph encoder LightGCN was then employed and Bayesian Personalized Ranking (BPR) loss was used as the main objective function loss:
Wherein
It is the user that embeds the matrix,
and
representing positive and negative term embedding matrices, respectively. σ is a sigmoid function.
Then, after comparative learning is carried out, a plurality of designated enhanced feature data corresponding to each piece of information data are obtained, and then a multi-task training function is utilized to process the plurality of designated enhanced feature data corresponding to each piece of information data, so that global feature data corresponding to each piece of information data are obtained. Because the information data is the interactive record of the user and the item information in the information system, the finally generated global feature data can be a feature vector matrix corresponding to the user and the item information.
The multi-task training function is:
Wherein Θ is
λ _1 is a contrast learning rate for adjusting
And
loss ratio ofThe rate, λ _2, is a hyper-parameter that controls L2 regularization.
205. And performing cross learning processing on the global feature data among the plurality of pieces of information data to obtain a plurality of pieces of target feature data.
In the step, after the global feature data is obtained, the global feature data in each piece of information data is processed through a fully-connected neural network preset in a signal system, a nonlinear relation between a user and data corresponding to each piece of information data is obtained, and therefore stable information input is provided for an information output layer.
Specifically, when the cross attention mechanism is used to perform the cross learning processing on the global feature data among the plurality of pieces of information data, the cross learning processing may be performed on the global feature data among the plurality of pieces of information data by using a cross attention formula to obtain a plurality of pieces of target feature data, and then the combination processing may be performed on the plurality of pieces of target feature data by using a preset aggregation calculation formula and a weight formula to obtain processed target feature data. Specifically, the preset weight calculation formula may be:
Wherein, the cross attention formula is:
Wherein the content of the first and second substances,
、
、
vectors respectively corresponding to global feature data generated using contrast learning
Which is indicative of the result of the projection,
and 
is [0, 1 ]]A hyper-parameter within a range. And obtaining the common characteristic E _ C of the user in two different fields through a formula five, obtaining the characteristics of the user in two respective fields through a formula six and a formula seven, and controlling the retention rate of the characteristics through the use of the hyper-parameters. For example, when λ ^ ((a)) = λ ^ ((b)) =1.0, it indicates that 100% of the user features in the two domains are preserved, so the specificity of the user features in the two domains is the greatest. When λ ^ ((a)) = λ ^ ((b)) = 0.5, the specificity of the user's feature disappears, and the same user has the same feature in both domains.
The preset aggregation calculation formula may be:
Wherein, is the element multiplication.
、
And the network weight matrixes are respectively the global characteristic data corresponding to different information data.
And the learnable matrixes correspond to the information data in different fields. U ̃ ^ a is the characteristic value of the aggregated feature data, and U ̃ ^ b is the characteristic value of the global feature corresponding to each information data.
Wherein a and b represent the fields in which different information data are located, respectively.
206. And determining target information data in the plurality of pieces of information data based on the plurality of pieces of target characteristic data, and outputting the target information data as an information recommendation result.
In this step, after the plurality of processed target feature data are obtained in step 205, the plurality of processed target feature data may be input to a result generation layer in the information recommendation model trained in advance, and the information recommendation result may be output.
Since the information data is the interaction record of the user and the specific project information, the target feature data is also the feature related to the interaction record of the user and the specific project information, for example, if the information data is the interaction record of the user and a certain song, the target feature data may be the time length for the user to listen to the certain song in the entertainment system, and the like.
Therefore, in the result generation layer, the probability that the user and the item information may interact with each other may be generated from the target feature data, and then, information data corresponding to the item information having a high probability may be output as the target information data and the target information data may be output as the information recommendation result.
Further, as an implementation of the method shown in fig. 1, an embodiment of the present invention further provides an information recommendation apparatus, which is used for implementing the method shown in fig. 1. The embodiment of the apparatus corresponds to the embodiment of the method, and for convenience of reading, details in the embodiment of the apparatus are not described again one by one, but it should be clear that the apparatus in the embodiment can correspondingly implement all the contents in the embodiment of the method. As shown in fig. 3, the apparatus includes:
an obtaining unit 301, configured to obtain multiple pieces of information data through a preset interface, where the multiple pieces of information data are derived from multiple information systems in rail transit, and the preset interface is an interface between a signal system and the multiple information systems in rail transit;
a determining unit 302, configured to determine a plurality of feature data corresponding to each piece of information data acquired by the acquiring unit 301;
a first processing unit 303, configured to perform comparison and learning processing on multiple pieces of feature data corresponding to each piece of information data determined by the determining unit 302 to obtain global feature data corresponding to each piece of information data;
a second processing unit 304, configured to perform cross learning processing on global feature data among the pieces of information data processed by the first processing unit 303 to obtain pieces of target feature data;
a result output unit 305, configured to determine target information data among the plurality of pieces of information data based on the plurality of pieces of target feature data processed by the second processing unit 304, and output the target information data as an information recommendation result.
Further, as an implementation of the method shown in fig. 2, an embodiment of the present invention further provides another information recommendation apparatus, which is used for implementing the method shown in fig. 2. The embodiment of the apparatus corresponds to the embodiment of the method, and for convenience of reading, details in the embodiment of the apparatus are not repeated one by one, but it should be clear that the apparatus in the embodiment can correspondingly implement all the contents in the embodiment of the method. As shown in fig. 4, the apparatus includes:
an obtaining unit 301, configured to obtain multiple pieces of information data through a preset interface, where the multiple pieces of information data are derived from multiple information systems in rail transit, and the preset interface is an interface between a signal system and the multiple information systems in rail transit;
a determining unit 302, configured to determine a plurality of feature data corresponding to each piece of information data acquired by the acquiring unit 301;
a first processing unit 303, configured to perform comparison and learning processing on multiple pieces of feature data corresponding to each piece of information data determined by the determining unit 302 to obtain global feature data corresponding to each piece of information data;
a second processing unit 304, configured to perform cross learning processing on global feature data among the pieces of information data processed by the first processing unit 303 to obtain pieces of target feature data;
a result output unit 305, configured to determine target information data among the plurality of pieces of information data based on the plurality of pieces of target feature data processed by the second processing unit 304, and output the target information data as an information recommendation result.
In an optional implementation manner, the determining unit 302 is specifically configured to:
inputting each piece of information data into a feature extraction layer in a pre-trained information recommendation model, and extracting a plurality of feature data corresponding to each piece of information data, wherein the information recommendation model is located in a signal system.
In an optional implementation, the first processing unit 303 includes:
a feature enhancing module 3031, configured to perform enhancement processing on multiple feature data corresponding to each piece of information data in a feature extraction layer in a pre-trained information recommendation model in a uniform noise processing manner, so as to obtain multiple enhanced feature data corresponding to each piece of information data;
a comparison learning module 3032, configured to perform comparison learning processing on multiple pieces of enhanced feature data corresponding to each piece of information data obtained by the feature enhancement module 3031, so as to obtain global feature data corresponding to each piece of information data.
In an optional implementation manner, the comparative learning module 3032 is specifically configured to:
comparing and learning the plurality of enhanced feature data corresponding to each piece of information data by using a comparison loss function to obtain a plurality of specified enhanced feature data corresponding to each piece of information data;
and processing the plurality of specified enhanced feature data corresponding to each piece of information data by using a multi-task training function to obtain global feature data corresponding to each piece of information data.
In an optional implementation, the second processing unit 304 includes:
the cross processing module 3041 is configured to input global feature data of multiple pieces of information data into a feature migration layer in a pre-trained information recommendation model, and perform cross learning processing on the global feature data among the multiple pieces of information data by using a cross attention mechanism to obtain multiple pieces of target feature data.
In an optional implementation manner, the intersection processing module 3041 is specifically configured to:
performing cross learning processing on global feature data among the plurality of pieces of information data by using a cross attention formula to obtain a plurality of pieces of target feature data;
and combining the plurality of pieces of target characteristic data by using a preset weight calculation formula and a preset aggregation calculation formula to obtain processed target characteristic data.
Further, an embodiment of the present invention further provides a storage medium, where the storage medium is used to store a computer program, where the computer program controls, when running, a device in which the storage medium is located to execute the information recommendation method described in fig. 1-2.
Further, an embodiment of the present invention further provides a processor, where the processor is configured to execute a program, where the program executes the information recommendation method described in fig. 1-2.
In the foregoing embodiments, the descriptions of the respective embodiments have respective emphasis, and for parts that are not described in detail in a certain embodiment, reference may be made to related descriptions of other embodiments.
It will be appreciated that the relevant features of the method and apparatus described above are referred to one another. In addition, "first", "second", and the like in the above embodiments are for distinguishing the embodiments, and do not represent merits of the embodiments.
It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described systems, apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.
The algorithms and displays presented herein are not inherently related to any particular computer, virtual machine, or other apparatus. Various general purpose systems may also be used with the teachings herein. The required structure for constructing such a system will be apparent from the description above. Moreover, the present invention is not directed to any particular programming language. It is appreciated that a variety of programming languages may be used to implement the teachings of the present invention as described herein, and any descriptions of specific languages are provided above to disclose the best mode of the invention.
In addition, the memory may include volatile memory in a computer readable medium, random Access Memory (RAM) and/or nonvolatile memory such as Read Only Memory (ROM) or flash memory (flash RAM), and the memory includes at least one memory chip.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and so forth) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
In a typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.
The memory may include forms of volatile memory in a computer readable medium, random Access Memory (RAM) and/or non-volatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM). The memory is an example of a computer-readable medium.
Computer-readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device. As defined herein, a computer readable medium does not include a transitory computer readable medium such as a modulated data signal and a carrier wave.
It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising a … …" does not exclude the presence of another identical element in a process, method, article, or apparatus that comprises the element.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The above are merely examples of the present application and are not intended to limit the present application. Various modifications and changes may occur to those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.
Claims (10)
1. An information recommendation method, characterized in that the method comprises:
acquiring a plurality of pieces of information data through a preset interface, wherein the plurality of pieces of information data are from a plurality of information systems in rail transit, and the preset interface is an interface between a signal system and the plurality of information systems in the rail transit;
determining a plurality of characteristic data corresponding to each piece of information data;
comparing and learning the plurality of feature data corresponding to each piece of information data to obtain global feature data corresponding to each piece of information data;
performing cross learning processing on global feature data among the information data to obtain target feature data;
and determining target information data in the plurality of pieces of information data based on the plurality of pieces of target characteristic data, and outputting the target information data as an information recommendation result.
2. The method of claim 1, wherein determining a plurality of feature data corresponding to each piece of information data comprises:
inputting each piece of information data into a feature extraction layer in a pre-trained information recommendation model, and extracting a plurality of feature data corresponding to each piece of information data, wherein the information recommendation model is located in a signal system.
3. The method according to claim 1, wherein performing a comparative learning process between a plurality of feature data corresponding to each piece of information data to obtain global feature data corresponding to each piece of information data includes:
in a feature extraction layer in a pre-trained information recommendation model, performing enhancement processing on a plurality of feature data corresponding to each piece of information data by using a uniform noise processing mode to obtain a plurality of enhanced feature data corresponding to each piece of information data;
and performing comparative learning processing on the plurality of enhanced feature data corresponding to each piece of information data to obtain global feature data corresponding to each piece of information data.
4. The method according to claim 3, wherein performing a contrast learning process between the plurality of enhanced feature data corresponding to each piece of information data to obtain global feature data corresponding to each piece of information data includes:
performing contrast learning processing on the plurality of enhanced feature data corresponding to each piece of information data by using a contrast loss function to obtain a plurality of specified enhanced feature data corresponding to each piece of information data;
and processing the plurality of specified enhanced feature data corresponding to each piece of information data by using a multi-task training function to obtain global feature data corresponding to each piece of information data.
5. The method of claim 1, wherein performing cross-learning processing on global feature data among the plurality of pieces of information data to obtain a plurality of pieces of target feature data comprises:
and inputting the global feature data of the plurality of pieces of information data into a feature migration layer in a pre-trained information recommendation model, and performing cross learning processing on the global feature data among the plurality of pieces of information data by using a cross attention mechanism to obtain a plurality of pieces of target feature data.
6. The method of claim 5, wherein performing cross-learning processing on global feature data among the plurality of pieces of information data using a cross-attention mechanism to obtain a plurality of pieces of target feature data comprises:
performing cross learning processing on global feature data among the plurality of pieces of information data by using a cross attention formula to obtain a plurality of pieces of target feature data;
and combining the plurality of pieces of target characteristic data by using a preset weight calculation formula and a preset aggregation calculation formula to obtain processed target characteristic data.
7. An information recommendation apparatus, characterized in that the apparatus comprises:
the system comprises an acquisition unit, a processing unit and a processing unit, wherein the acquisition unit is used for acquiring a plurality of pieces of information data through a preset interface, the plurality of pieces of information data are sourced from a plurality of information systems in the rail transit, and the preset interface is an interface between a signal system and the plurality of information systems in the rail transit;
a determining unit configured to determine a plurality of feature data corresponding to each piece of information data acquired by the acquiring unit;
the first processing unit is used for performing comparison learning processing on a plurality of pieces of feature data corresponding to each piece of information data determined by the determining unit to obtain global feature data corresponding to each piece of information data;
the second processing unit is used for performing cross learning processing on global feature data among the plurality of pieces of information data obtained by the first processing unit to obtain a plurality of pieces of target feature data;
and the result output unit is used for determining target information data in the plurality of pieces of information data based on the plurality of pieces of target characteristic data processed by the second processing unit and outputting the target information data as an information recommendation result.
8. The apparatus according to claim 7, wherein the determining unit is specifically configured to:
inputting each piece of information data into a feature extraction layer in a pre-trained information recommendation model, and extracting a plurality of feature data corresponding to each piece of information data, wherein the information recommendation model is located in a signal system.
9. A storage medium characterized by comprising a stored program, wherein a device on which the storage medium is located is controlled to execute the information recommendation method according to any one of claims 1 to 6 when the program runs.
10. A processor, characterized in that the processor is configured to execute a program, wherein the program executes the information recommendation method according to any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310014861.4A CN115757528A (en) | 2023-01-06 | 2023-01-06 | Information recommendation method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310014861.4A CN115757528A (en) | 2023-01-06 | 2023-01-06 | Information recommendation method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115757528A true CN115757528A (en) | 2023-03-07 |
Family
ID=85348501
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310014861.4A Pending CN115757528A (en) | 2023-01-06 | 2023-01-06 | Information recommendation method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115757528A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112417319A (en) * | 2020-11-24 | 2021-02-26 | 清华大学 | Site recommendation method and device based on difficulty sampling meta-learning |
| CN114117029A (en) * | 2021-11-24 | 2022-03-01 | 国网山东省电力公司信息通信公司 | Solution recommendation method and system based on multi-level information enhancement |
| CN115203264A (en) * | 2022-05-13 | 2022-10-18 | 青岛文达通科技股份有限公司 | Urban tourism route recommendation method and system based on LightGBM algorithm |
| US20220374776A1 (en) * | 2021-12-27 | 2022-11-24 | Beijing Baidu Netcom Science Technology Co., Ltd. | Method and system for federated learning, electronic device, and computer readable medium |
-
2023
- 2023-01-06 CN CN202310014861.4A patent/CN115757528A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112417319A (en) * | 2020-11-24 | 2021-02-26 | 清华大学 | Site recommendation method and device based on difficulty sampling meta-learning |
| CN114117029A (en) * | 2021-11-24 | 2022-03-01 | 国网山东省电力公司信息通信公司 | Solution recommendation method and system based on multi-level information enhancement |
| US20220374776A1 (en) * | 2021-12-27 | 2022-11-24 | Beijing Baidu Netcom Science Technology Co., Ltd. | Method and system for federated learning, electronic device, and computer readable medium |
| CN115203264A (en) * | 2022-05-13 | 2022-10-18 | 青岛文达通科技股份有限公司 | Urban tourism route recommendation method and system based on LightGBM algorithm |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11709902B2 (en) | Recommendation method and apparatus, and storage medium | |
| CN109816032B (en) | Unbiased mapping zero sample classification method and device based on generative countermeasure network | |
| US11907817B2 (en) | System and methods for machine learning training data selection | |
| CN115545114A (en) | Training method of multi-task model, content recommendation method and device | |
| Park et al. | Exploiting negative preference in content-based music recommendation with contrastive learning | |
| CN114332550A (en) | Model training method, system, storage medium and terminal equipment | |
| CN109344319B (en) | Online content popularity prediction method based on ensemble learning | |
| CN112148994A (en) | Information push effect evaluation method and device, electronic equipment and storage medium | |
| CN115757528A (en) | Information recommendation method and device | |
| CN115186750A (en) | Model training method, device, equipment and storage medium | |
| CN117150053A (en) | Multimedia information recommendation model training method, recommendation method and device | |
| CN113065067A (en) | Article recommendation method and device, computer equipment and storage medium | |
| CN115482021A (en) | Multimedia information recommendation method and device, electronic equipment and storage medium | |
| CN113298283A (en) | Content object prediction method and device and content object recommendation method | |
| CN112035740A (en) | Project use duration prediction method, device, equipment and storage medium | |
| CN111062477A (en) | Data processing method, device and storage medium | |
| CN113205369B (en) | User consumption duration prediction method and device, electronic equipment and storage medium | |
| CN110990522B (en) | Legal document determining method and system | |
| CN116485505B (en) | Method and device for training recommendation model based on user performance fairness | |
| CN113836407B (en) | Recommendation method and related device | |
| CN115878884A (en) | Recommendation method based on dynamic difference diagram | |
| CN117407529A (en) | Node classification method, device, equipment, medium and product based on decoupling graph | |
| CN115130654A (en) | Neural network structure searching method, apparatus, computer device and storage medium | |
| CN106708901B (en) | Clustering method and device for search words in website | |
| CN117851684A (en) | Real-time article recommending method and system based on knowledge graph and incremental learning |
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: 20230307 |
|
| RJ01 | Rejection of invention patent application after publication |