Public culture resource recommendation method and device based on graph neural network

Abstract

The invention provides a public culture resource recommendation method and device based on a graph neural network, and the method comprises the steps: processing a plurality of public culture resource data, obtaining a resource feature and a resource anchor point of each piece of public culture resource data, and enabling each piece of public culture resource data to comprise a plurality of culture resource points; according to the resource features and the resource anchor points of each piece of public culture resource data, constructing a first relation network of the plurality of culture resource points and the resource anchor points; according to the first relation network, constructing a second relation network between the user and the resource anchor point; obtaining a relation change prediction result of the user and the resource anchor point according to the second relation network; and according to the relationship change prediction result and the first relationship network, obtaining a recommendation result of the public culture resources. By means of the method, the problems that due to the fact that the number of cultural relics in the museum is large and the number of registered users is large, a traditional recommendation method cannot cope with massive data, joint recommendation between museums is difficult, and the recommendation requirements of the users cannot be met are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of cultural resource recommendation, and in particular to a method and device for recommending public cultural resources based on a graph neural network. Background Art

[0002] With the progress of society and the development of science and technology, as well as the advancement of the digital process of museums and cultural centers, the recommendation model for museum collections has received more and more attention and plays an important role in the development and dissemination of public culture. However, due to the large number of cultural relics in the collection and the large number of registered users, traditional recommendation methods cannot cope with such a large amount of data. They can often only build local recommendations for exhibits or cultural relics around a single cultural center or museum, which cannot meet the recommendation needs of users. Summary of the invention

[0003] The present invention provides a method and device for recommending public cultural resources based on graph neural network, so as to solve the problem in the prior art that due to the large number of cultural relics in collections and a large number of registered users, traditional recommendation methods are unable to cope with such a large amount of data, making it difficult to make joint recommendations between libraries and unable to meet the recommendation needs of users.

[0004] The present invention provides a method for recommending public cultural resources based on a graph neural network, comprising the following steps: Processing a plurality of public cultural resource data to obtain resource characteristics and resource anchor points of each public cultural resource data, wherein each public cultural resource data includes a plurality of cultural resource points; According to the resource characteristics and resource anchor points of each public cultural resource data, a first relationship network between multiple cultural resource points and resource anchor points is constructed, and the first relationship network is used to characterize the resource similarity between multiple cultural resource points and resource anchor points in each public cultural resource data; Based on the first relationship network, construct a second relationship network of users and resource anchors; According to the second relationship network, obtaining the relationship change prediction result between the user and the resource anchor point; According to the relationship change prediction result and the first relationship network, a recommendation result of the public cultural resource is obtained to recommend the public cultural resource to the user according to the recommendation result.

[0005] According to the public cultural resource recommendation method based on graph neural network provided by the present invention, the first relationship network between cultural resource points and resource anchor points is constructed according to the resource characteristics and resource anchor points of each public cultural resource data, including: According to the resource characteristics of each public cultural resource data, the characteristic similarity between each cultural resource point and the resource anchor point is calculated; According to the feature similarity between each cultural resource point and the resource anchor point, a first relationship network is obtained.

[0006] According to the public cultural resource recommendation method based on graph neural network provided by the present invention, the second relationship network of users and resource anchor points is constructed according to the first relationship network, including: Acquire cultural resource points browsed by the user in a historical period to obtain at least one first cultural resource point; According to the first relationship network, query the first resource anchor point corresponding to the first cultural resource point; at the same time, obtain the resource similarity between the first cultural resource point and the first resource anchor point; Obtaining an association weight between the user and the first resource anchor point according to the resource similarity between the first cultural resource point and the first resource anchor point; A second relationship network is constructed according to the association weights between the user and the first resource anchor point.

[0007] The public cultural resource recommendation method based on graph neural network provided by the present invention also includes: When the user generates a browsing record of a cultural resource point after the historical period, obtaining a second cultural resource point browsed by the user; According to the first relationship network, a second resource anchor point corresponding to the second cultural resource point is obtained, and at the same time, a resource similarity between the second cultural resource point and the second resource anchor point is obtained; The second relationship network is updated according to the resource similarity between the second cultural resource point and the second resource anchor point.

[0008] According to the public cultural resource recommendation method based on graph neural network provided by the present invention, obtaining the predicted result of the relationship change between the user and the resource anchor point according to the second relationship network includes: Acquire the second relationship network multiple times at preset time intervals to obtain a first relationship network sequence between the user and the resource anchor point; and acquire the second relationship network sequences between other users and the resource anchor points to obtain multiple second relationship network sequences; Calculate the relationship network similarity between the second relationship network most recently obtained in the first relationship network sequence and each second relationship network in each second relationship network sequence to obtain a calculation result; According to the calculation result, at least one similar user is selected from other users to obtain a third relationship network sequence between the similar users and the resource anchor points; According to the first relationship network sequence and the third relationship network sequence, a relationship change prediction result between the user and the resource anchor point is obtained.

[0009] According to the public cultural resource recommendation method based on graph neural network provided by the present invention, obtaining the relationship change prediction result between the user and the resource anchor point according to the first relationship network sequence and the third relationship network sequence includes: In each third relationship network sequence, the second relationship networks are sorted from large to small according to the relationship network similarity, and the similar relationship network ranked first is obtained; In each third relationship network sequence, a second relationship network obtained after a preset time interval of each similar relationship network is obtained to obtain a predicted relationship network; According to the predicted relationship network, the predicted results of the relationship changes between users and resource anchors are obtained.

[0010] The public cultural resource recommendation method based on graph neural network provided by the present invention also includes: When the similar relationship network does not have a second relationship network after a preset time interval, the user corresponding to the similar relationship network is excluded from the similar users.

[0011] The present invention also provides a public cultural resource recommendation device based on graph neural network, comprising the following modules: A data processing module is used to process a plurality of public cultural resource data to obtain resource characteristics and resource anchor points of each public cultural resource data, wherein each public cultural resource data includes a plurality of cultural resource points; A first relationship building module, used to build a first relationship network between multiple cultural resource points and resource anchor points according to the resource characteristics and resource anchor points of each public cultural resource data, wherein the first relationship network is used to characterize the resource similarity between multiple cultural resource points and resource anchor points in each public cultural resource data; A second relationship building module, used to build a second relationship network between users and resource anchors based on the first relationship network; A relationship prediction module, used to obtain a prediction result of a relationship change between a user and a resource anchor point according to the second relationship network; The resource recommendation module is used to obtain the recommendation results of public cultural resources according to the relationship change prediction results and the first relationship network, so as to recommend the public cultural resources to the user according to the recommendation results.

[0012] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, a method for recommending public cultural resources based on graph neural networks as described above is implemented.

[0013] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a method for recommending public cultural resources based on a graph neural network as described in any one of the above.

[0014] The present invention also provides a computer program product, including a computer program, which, when executed by a processor, implements any of the above-mentioned public cultural resource recommendation methods based on graph neural networks.

[0015] The public cultural resource recommendation device method and apparatus based on graph neural network provided by the present invention predicts the resource anchor points that the user is interested in in the future by constructing a first relationship network for characterizing the resource similarities between multiple cultural resource points and resource anchor points in each public cultural resource data, and constructing a second relationship network between users and resource anchor points. This makes it possible to recommend the public cultural resources that the user is most likely to be interested in based on the resource anchor points that the user is interested in and further based on the first relationship network, thereby solving the problem that there are too many public cultural resources and too many registered users to meet the recommendation needs of users. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0017] Figure 1 It is a flowchart of a public cultural resource recommendation method based on graph neural network provided by the present invention.

[0018] Figure 2 It is a schematic diagram of the principle of a feature extraction model provided by the present invention.

[0019] Figure 3 It is a schematic diagram of the principle of aligning resource features of various resource anchor points provided by the present invention.

[0020] Figure 4 It is a principle schematic diagram of a relationship network sequence provided by the present invention.

[0021] Figure 5 It is a schematic diagram of the principle of an RNN prediction model provided by the present invention.

[0022] Figure 6 It is a structural schematic diagram of a public cultural resource recommendation device based on graph neural network provided by the present invention.

[0023] Figure 7 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0025] The terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the descriptions used in this way can be interchanged where appropriate, so that the embodiments can be implemented in a sequence other than that illustrated or described in the present application. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or modules is not necessarily limited to those steps or modules that are clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products or devices. The naming or numbering of the steps that appear in the present application does not mean that the steps in the method flow must be executed in the time / logical sequence indicated by the naming or numbering. The process steps that have been named or numbered can change the execution order according to the technical purpose to be achieved, as long as the same or similar technical effects can be achieved. The division of modules in this application is a logical division. There may be other division methods when it is implemented in actual applications. For example, multiple modules can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, and the indirect coupling or communication connection between units can be electrical or other similar forms, which are not limited in this application. In addition, the modules or submodules described as separate components may or may not be physically separated, may or may not be physical units, or may be distributed in multiple circuit units, and some or all of the units may be selected according to actual needs to achieve the purpose of the present application.

[0026] Combine the following Figure 1-Figure 7 The specific contents of the present invention are described.

[0027] Figure 1 This is one of the flow charts of the public cultural resource recommendation method based on graph neural network provided by the present invention, such as Figure 1 As shown, the method includes steps S101-S105.

[0028] Step S101, processing a plurality of public cultural resource data to obtain resource characteristics and resource anchor points of each public cultural resource data, wherein each public cultural resource data includes a plurality of cultural resource points.

[0029] In an embodiment of the present invention, public cultural resources may be various cultural relics and exhibits in a museum, a cultural center, or an exhibition hall. Each individual museum, cultural center, or exhibition hall corresponds to a public cultural resource data, and any public cultural resource data includes multiple cultural resource points. Exemplarily, for Museum 1, cultural resource points such as cultural relics A, cultural relics B, and cultural relics C constitute multiple cultural resource points existing in the public cultural resource data. In an embodiment of the present invention, a museum is regarded as a public cultural resource of the present invention, and each cultural relic in the museum is regarded as a cultural resource point. The above description is only exemplary and is not a limitation of the present invention.

[0030] In an embodiment of the present invention, multiple public cultural resource data are processed, that is, semantic features are extracted from the cultural relic resource data of each museum, and a cultural relic anchor database is selected and constructed for each museum, which mainly includes three parts: local cultural relic feature search, cultural relic anchor selection, and cultural relic anchor feature alignment. Among them, the cultural relic resource data is mainly digital data such as pictures, videos, and texts related to cultural relics. When analyzing and processing, the pictures and texts of cultural relics are used as the analysis basis.

[0031] Specifically, step S101 includes steps S1011-S1013.

[0032] Step S1011, using CLIP as the main framework, extracting semantic features of cultural resource points according to a resource feature extraction model to obtain resource features of each cultural resource point.

[0033] like Figure 2 As shown in the figure, the entire feature extraction model training consists of two stages. In the first stage of training, the existing cultural relic image-text data pairs are used to conduct the first stage of training based on the idea of ​​CLIP contrastive learning. In the second stage of training, after the training is stable, the cultural relics are labeled. The labeling uses the existing cultural relic classification model for multi-label labeling, and the image features and text features are fused by addition. All model parameters except the last layer are frozen to train the classification model. After the training is completed, the fused features are output as the final cultural relic semantic features, and the semantic feature dimension is 1024. In practice, in order to facilitate data expansion and cope with massive data training, a separate cultural relic feature extraction model is trained for each museum data, taking a single museum resource data as a unit.

[0034] Step S1012, selecting a resource anchor point from multiple cultural resource points of each public cultural resource data.

[0035] In the present invention, when the cultural resource points are various cultural relics, the resource anchor points are cultural relic anchor points. The cultural relic anchor point selection is to select representative data points from a large number of cultural relic data resources, and make recommendations and feature alignment around these anchor points.

[0036] For example, based on existing classification methods and experience, existing cultural relics are simply divided according to year, material, and type. In each divided subclass, a deep clustering method is used to generate several cluster centers for each subclass. For each cluster center, the most similar cultural relic is selected as the anchor cultural relic, and its features are used as the anchor features. Then, resource anchors are selected from multiple cultural resource points of each public cultural resource data.

[0037] Step S1013: selecting a reference anchor point from multiple resource anchor points, and aligning resource features of each resource anchor point according to the reference anchor point.

[0038] Figure 3 This is a schematic diagram of the principle of aligning resource features of various resource anchor points provided by the present invention. The feature alignment of cultural relic anchor points is used to align the feature extraction models of cultural relics in different museums. The framework structure is as follows Figure 3 As shown in the figure, with the data features of Museum 1 as the target, for the cultural relic feature extraction model 2, a feature transformation module consisting of two fully connected layers is added before the last classification layer to realize the alignment of the cultural relic feature extraction model 2 to the cultural relic feature extraction model 1, and freeze the parameters of other layers. The specific steps are as follows: obtain the features of the cultural relic anchor point of Museum 1 under the cultural relic feature extraction model 1, that is, the anchor feature of Museum 1; obtain the features of the cultural relic anchor point of Museum 1 under the cultural relic feature extraction model 2 and after the feature transformation module ( Figure 3 The dotted line in the middle is the anchor point feature of Museum 1 after transformation; the transformed feature of the anchor point of Museum 2 under the feature extraction model 2 of the cultural relic is obtained, which is the anchor point feature of Museum 2 after transformation. Figure 3 , calculate the alignment loss and classification loss respectively, add the alignment loss and classification loss as the final loss, optimize the parameters of the feature transformation module, so that the cultural relic feature extraction model 2 is aligned with the cultural relic feature extraction model 1.

[0039] For other public cultural resources such as Museum 3 and Museum 4, the resource anchor point of Museum 1 is also used as the benchmark anchor point, and the above method is used in turn to complete the resource feature alignment of each resource anchor point, thereby achieving the model parameter alignment of the cultural relics feature extraction models corresponding to each museum.

[0040] In the embodiment of the present invention, the cultural relic feature extraction model used as a benchmark is generally constructed from cultural relic data of a richer variety, ensuring that other museum anchor cultural relics can be matched in the benchmark.

[0041] Step S102, constructing a first relationship network of multiple cultural resource points and resource anchor points based on the resource characteristics and resource anchor points of each public cultural resource data, wherein the first relationship network is used to characterize the resource similarity between multiple cultural resource points and resource anchor points in each public cultural resource data.

[0042] In the embodiment of the present invention, for each museum, there is a corresponding first relationship network, which is used to characterize the resource similarity between each cultural relic and the cultural relic anchor point in the museum. That is, according to the resource characteristics and resource anchor points of each public cultural resource data, the resource similarity between multiple cultural resource points and the resource anchor points in each public cultural resource data is calculated. For details, please refer to the following embodiment content.

[0043] In a possible implementation, based on the resource characteristics and resource anchor points of each public cultural resource data, a first relationship network of cultural resource points and resource anchor points is constructed, including: Step S1021, based on the resource characteristics of each public cultural resource data, calculate the characteristic similarity between each cultural resource point and the resource anchor point.

[0044] For example,

[0045] For each museum, after the cultural relic anchor points are constructed, the cultural relic features of each cultural relic are normalized, and the similarity between the cultural relic features and the anchor point features is calculated in turn. For cultural relic 1, the similarity is ; For artifact 2, the similarity is ; For cultural relic n, the similarity is ; in , arrive are the cultural relic characteristics of cultural relic 1, cultural relic 2 to cultural relic n, is the anchor feature.

[0046] Step S1022: obtaining a first relationship network according to the feature similarity between each cultural resource point and the resource anchor point.

[0047] Specifically, cultural relics are taken as graph nodes, cultural relics and anchor points are taken as edges, and the similarity between cultural relics and anchor points is taken as the edge weight. A cultural relic-anchor point relationship network is constructed for each museum to obtain the first relationship network.

[0048] Step S103: construct a second relationship network of users and resource anchors based on the first relationship network.

[0049] In an embodiment of the present invention, the second relationship network between users and resource anchors is used to record the browsing history tendency of users, wherein users and resource anchors are graph nodes, and the weights of the edges between users and resource anchors serve as the user's interest in the resource anchors.

[0050] For example, a user may view artifact data from multiple museums, so the anchor artifacts associated with the user cover all museum anchor artifact data, and the alignment of duplicate artifact anchors is removed.

[0051] In a possible implementation, constructing a second relationship network of users and resource anchors based on the first relationship network includes steps S1031 to S1034. In steps S1031 to S1034, an initial second relationship network of users and resource anchors is obtained.

[0052] S1031, obtaining cultural resource points browsed by the user in a historical period, and obtaining at least one first cultural resource point.

[0053] Specifically, it includes querying the types of cultural relics in museums that the user has browsed in the historical period, including cultural relic 1 in museum 1 and cultural relic 2 and cultural relic 3 in museum 2.

[0054] Step S1032: query the first resource anchor point corresponding to the first cultural resource point according to the first relationship network; and at the same time, obtain the resource similarity between the first cultural resource point and the first resource anchor point.

[0055] In the first relationship network of museum 1, query the cultural relic anchor point corresponding to cultural relic 1 to obtain the resource similarity Q1 between cultural relic 1 and cultural relic anchor point 1; at the same time, query the cultural relic anchor points corresponding to cultural relic 2 and cultural relic 3 in the first relationship network of museum 2 to obtain the resource similarities Q2 and Q3 between cultural relic 2 and cultural relic 3 and cultural relic anchor point 2 respectively.

[0056] Step S1033: Obtain an association weight between the user and the first resource anchor point according to the resource similarity between the first cultural resource point and the first resource anchor point.

[0057] At this time, for the cultural relic anchor point 1 of museum 1, the associated weight is Q1; for the cultural relic anchor point 2 of museum 2, the associated weight is Q2+Q3.

[0058] Step S1034: construct a second relationship network according to the association weight between the user and the first resource anchor point.

[0059] Finally, the second relationship network is constructed based on the association weight Q1 between the user and cultural relic anchor point 1, and the association weight Q2+Q3 between the user and cultural relic anchor point 2.

[0060] In a possible implementation, constructing a second relationship network between users and resource anchors according to the first relationship network further includes steps S1035-S1037.

[0061] In the embodiment of the present invention, each time a user browses cultural relic data, according to the first relationship network of cultural relic-anchor point, the edge weight calculation and update steps of the relationship between the user and the resource anchor point are as follows: Step S1035, when the user generates a browsing record of a cultural resource point after the historical period, a second cultural resource point browsed by the user is obtained.

[0062] Exemplarily, when a user generates a browsing record of a cultural resource point after a historical period, it is recorded that the user browsed cultural relics 4 in a museum 3 .

[0063] Step S1036: acquiring the second resource anchor point corresponding to the second cultural resource point according to the first relationship network, and acquiring the resource similarity between the second cultural resource point and the second resource anchor point.

[0064] According to the first relationship network of museum 3, the second resource anchor point (cultural relic anchor point 3) corresponding to the second cultural resource point (cultural relic 4) is obtained, and at the same time, the resource similarity (Q4) between the second cultural resource point and the second resource anchor point is obtained.

[0065] Step S1037: updating the second relationship network according to the resource similarity between the second cultural resource point and the second resource anchor point.

[0066] At this time, in the second relationship network, a weight correspondence between the cultural relic anchor point 3 and the user is added.

[0067] It should be noted that when a user generates a new browsing record, the following situations may occur: When the user generates a browsing record of a cultural resource point after a historical period, it is recorded that the user browsed cultural relic 5 of museum 1.

[0068] According to the first relationship network of museum 1, the resource similarity Q5 between cultural relic 5 and cultural relic anchor point 1 in museum 1 is obtained.

[0069] At this time, the association weight between the user and cultural relic anchor point 1 is updated to A*Q1+B*Q5. For example, A is 99% and B is 1%.

[0070] Specifically, in the update of the second relationship network, the main source of the association weight corresponding to the resource anchor points that already exist in the second relationship network should be the cultural resource points that the user has recently browsed, but this update needs to be gradual. Therefore, in the process of updating the association weight between the user and the cultural relic anchor point 1, for the cultural resource points corresponding to the same resource anchor point, the weight ratio B of the newly browsed cultural resource points should be much smaller than the weight ratio A of the historically browsed cultural resource points, so as to achieve the gradual update of the second relationship network.

[0071] Step S104: acquiring a predicted result of a relationship change between a user and a resource anchor point according to the second relationship network.

[0072] In a possible implementation, obtaining a prediction result of a relationship change between a user and a resource anchor point according to the second relationship network includes steps S1041 to S1044.

[0073] S1041, acquiring the second relationship network multiple times at a preset time interval to obtain a first relationship network sequence between the user and the resource anchor point; and acquiring the second relationship network sequences between other users and the resource anchor points to obtain multiple second relationship network sequences.

[0074] In an embodiment of the present invention, the second relationship network of the user is obtained multiple times at a preset time interval to obtain a first relationship network sequence between the user and the resource anchor point. , the first relational network sequence include M is the total number of the second relationship networks of the acquired users. At the same time, the second relationship network sequences of other users and resource anchors are obtained, and multiple second relationship network sequences are obtained. For user 1 to user n, the second relationship network sequences are respectively , for the nth other user, the second relationship network sequence include .

[0075] In each of the above relationship network sequences, Sort from 1 to m according to the time of acquisition.

[0076] In one possible implementation, for any Before executing the following steps, the edges and corresponding resource anchors with weight values ​​less than the preset weight value are removed from the second relationship network of the user's anchor point to obtain a new For subsequent prediction steps. Exemplarily, in the embodiment of the present invention, the preset weight value is 0.1.

[0077] S1042, calculating the relationship network similarity between the second relationship network most recently obtained in the first relationship network sequence and each second relationship network in each second relationship network sequence, and obtaining a calculation result.

[0078] Calculate the first relationship network sequence The most recent acquisition , and the relationship network similarity of each second relationship network in each second relationship network sequence to obtain a calculation result.

[0079] For example, the second relationship network sequence of the nth other user , that is, calculate and The relationship network similarity is calculated to obtain the result.

[0080] In the embodiment of the present invention, any second relationship network is a one-dimensional data including the association weights between the user and each resource anchor point. =( ); =( ), k is the one-dimensional data length of the second relationship network. When the one-dimensional data lengths of the two second relationship networks are different, the shorter data length is padded to the longer data length by null value filling, and then the relationship network similarity is calculated; the relationship network similarity calculation formula is as follows:

[0081] S1043: According to the calculation result, at least one similar user is selected from other users to obtain a third relationship network sequence between the similar users and the resource anchor points.

[0082] Get all the calculation results from the first other user to the nth other user. For the nth other user. In the second relationship network sequence middle, For The second relationship network with the greatest similarity to the relationship network.

[0083] Get the second relationship network sequence of each other user in turn The second most similar relationship network ; Set the preset similarity threshold , for the second relationship network Filter. Get the values ​​that are greater than or equal to the preset similarity threshold. The second relationship network.

[0084]

[0085] The similar users are other user 1, other user 2, other user 3, other user 4 and other user 5. The third relationship network sequence is the relationship network sequence corresponding to other users 1-5 .

[0086] S1044: Obtain a prediction result of a relationship change between the user and the resource anchor point according to the first relationship network sequence and the third relationship network sequence.

[0087] In an embodiment of the present invention, based on the first relationship network sequence of a user and the third relationship network sequences of similar users, the change of the first relationship network sequence of the user in a subsequent period can be predicted through the change of the third relationship network sequence.

[0088] In a possible implementation, obtaining a prediction result of a relationship change between a user and a resource anchor point according to the first relationship network sequence and the third relationship network sequence includes S1044A-S1044C.

[0089] S1044A, in each third relationship network sequence, sort the second relationship networks from large to small according to the relationship network similarity, and obtain the similar relationship network ranked first.

[0090] Specifically, in each third relationship network sequence, the second relationship networks are sorted from large to small according to the relationship network similarity, and the similar relationship network ranked first is obtained. Specifically, for each third relationship network sequence , the similarity relationship networks ranked first are ,like Figure 4 In the relational network sequence The first relationship network sequence The most recent acquisition (in this embodiment, m is 10), and Alignment, see Figure 4 .

[0091] S1044B, in each third relationship network sequence, obtain a second relationship network obtained after a preset time interval for each similar relationship network to obtain a predicted relationship network.

[0092] like Figure 4 As shown, the dotted arrows correspond to the predicted relationship network sequence , where the mth (10) is the predicted relationship network .

[0093] In an embodiment of the present invention, the predicted relationship network can be obtained in the following manner: For the first relation network of the predicted relation network sequence For example, the relationship network corresponding to the dotted arrows includes and ,according to and According to the preset weight superposition, we get ; Similarly, we get the relationship network .

[0094] In an embodiment of the present invention, the predicted relationship network can also be obtained in the following manner: The prediction model structure is as follows Figure 5As shown in the figure, the UNet graph structure based on RNN is used as the time series. The time series length is fixed to 10. Before input, the input sequence graph structure is adjusted to the same structure. Since the second relationship network structure of the user-anchor point is relatively simple, two layers of graph downsampling are used for aggregation, and RNN is used to obtain the time series relationship features. Finally, a layer of graph upsampling is used to obtain the weight change prediction result of the second relationship network of the current user-anchor point.

[0095] S1044C, obtaining the predicted result of the relationship change between the user and the resource anchor point according to the predicted relationship network.

[0096] In a possible implementation, obtaining the predicted result of the relationship change between the user and the resource anchor point according to the first relationship network sequence and the third relationship network sequence further includes step S1044D.

[0097] Step S1044D: when the similar relationship network does not have a second relationship network after a preset time interval, the user corresponding to the similar relationship network is excluded from the similar users.

[0098] Specifically, when there is no second relationship network in the similar relationship network after a preset time interval, it means that the relationship network sequence of the similar user lacks subsequent reference value, and the user corresponding to the similar relationship network is excluded from the similar users to avoid interference with the relationship change prediction result.

[0099] Step S105 , obtaining a recommendation result of a public cultural resource according to the relationship change prediction result and the first relationship network, so as to recommend the public cultural resource to the user according to the recommendation result.

[0100] For example, based on the prediction results obtained For each museum, obtain the first relationship network between its cultural relics and anchor points, and predict the relationship change results according to the corresponding weights. Multiply and add them to the corresponding cultural relics. The five cultural relics with the highest scores are the recommendation results for the current museum. Count the recommendation results and scores of all museums, and select the top 10 cultural relics with the highest scores as the final recommendation results for users.

[0101] By adopting the above method, the present invention can achieve at least one of the following beneficial effects: 1. By constructing a first relationship network for characterizing the resource similarities between multiple cultural resource points and resource anchor points in each public cultural resource data, and constructing a second relationship network between users and resource anchor points, the resource anchor points that users are interested in in the future are predicted, so that the public cultural resources that users are most likely to be interested in can be recommended to users based on the resource anchor points that users are interested in and further based on the first relationship network, which solves the problem that there are too many public cultural resources and a large number of registered users, making it difficult to associate recommendations and unable to meet users' recommendation needs.

[0102] 2. By calculating the feature similarity between cultural resource points and resource anchor points to construct the first relationship network, it can accurately reflect the intrinsic connection and similarity between cultural resources, and provide a reliable basis for subsequent recommendations.

[0103] 3. Build a second relationship network between users and resource anchors based on the user's historical browsing history, and introduce association weights to measure the degree of association between users and resource anchors, making the recommendation results more personalized and accurate.

[0104] 4. When a user generates a new browsing history, the second relationship network can be updated in real time to ensure the timeliness and accuracy of the recommendation results and to better adapt to changes in user interests.

[0105] 5. By acquiring and analyzing the relationship network sequence between users and other users, it is possible to predict the changing trend of the relationship between users and resource anchors, perceive the changes in users' interests in advance, and provide more timely recommendation services.

[0106] The public cultural resource recommendation device based on graph neural network provided by the present invention is described below. The public cultural resource recommendation device based on graph neural network described below and the public cultural resource recommendation method based on graph neural network described above can refer to each other.

[0107] like Figure 6 As shown, the present invention also provides a public cultural resource recommendation device based on graph neural network, including the following modules: The data processing module 610 is used to process a plurality of public cultural resource data to obtain resource characteristics and resource anchor points of each public cultural resource data, wherein each public cultural resource data includes a plurality of cultural resource points; A first relationship building module 620, for building a first relationship network between multiple cultural resource points and resource anchor points according to the resource characteristics and resource anchor points of each public cultural resource data, wherein the first relationship network is used to characterize the resource similarity between multiple cultural resource points and resource anchor points in each public cultural resource data; A second relationship building module 630, configured to build a second relationship network between users and resource anchors based on the first relationship network; The relationship prediction module 640 is used to obtain the relationship change prediction result between the user and the resource anchor point according to the second relationship network; The resource recommendation module 650 is used to obtain the recommendation results of the public cultural resources according to the relationship change prediction results and the first relationship network, so as to recommend the public cultural resources to the user according to the recommendation results.

[0108] In a possible implementation manner, the first relationship building module 620 is specifically configured to: According to the resource characteristics of each public cultural resource data, the characteristic similarity between each cultural resource point and the resource anchor point is calculated; according to the characteristic similarity between each cultural resource point and the resource anchor point, a first relationship network is obtained.

[0109] In a possible implementation manner, the second relationship building module 630 is specifically configured to: Obtain cultural resource points browsed by the user in a historical period to obtain at least one first cultural resource point; query the first resource anchor point corresponding to the first cultural resource point according to the first relationship network; at the same time, obtain the resource similarity between the first cultural resource point and the first resource anchor point; obtain the association weight between the user and the first resource anchor point according to the resource similarity between the first cultural resource point and the first resource anchor point; and construct a second relationship network according to the association weight between the user and the first resource anchor point.

[0110] In a possible implementation manner, the second relationship building module 630 is further configured to: When a user generates a browsing record of a cultural resource point after a historical period, obtain a second cultural resource point browsed by the user; obtain a second resource anchor point corresponding to the second cultural resource point based on the first relationship network, and at the same time, obtain the resource similarity between the second cultural resource point and the second resource anchor point; and update the second relationship network based on the resource similarity between the second cultural resource point and the second resource anchor point.

[0111] In a possible implementation, the relationship prediction module 640 includes: The relationship acquisition unit is used to acquire the second relationship network multiple times at a preset time interval to obtain a first relationship network sequence between the user and the resource anchor point; and acquire the second relationship network sequences between other users and the resource anchor points to obtain multiple second relationship network sequences; A similarity calculation unit is used to calculate the relationship network similarity between the second relationship network most recently obtained in the first relationship network sequence and each second relationship network in each second relationship network sequence to obtain a calculation result; A user screening unit, configured to screen out at least one similar user from other users according to the calculation result, and obtain a third relationship network sequence between the similar users and the resource anchor points; The result prediction unit is used to obtain the relationship change prediction result between the user and the resource anchor point according to the first relationship network sequence and the third relationship network sequence.

[0112] In a possible implementation manner, the result prediction unit is specifically configured to: In each third relationship network sequence, the second relationship networks are sorted from large to small according to the relationship network similarity to obtain the similar relationship network ranked first; in each third relationship network sequence, the second relationship network obtained after a preset time interval for each similar relationship network is obtained to obtain a predicted relationship network; based on the predicted relationship network, the relationship change prediction result between the user and the resource anchor point is obtained.

[0113] In a possible implementation, the user screening unit is further configured to exclude the user corresponding to the similar relationship network from the similar users when there is no second relationship network in the similar relationship network after a preset time interval.

[0114] Figure 7 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 7 As shown, the electronic device may include: a processor 710, a communications interface 720, a memory 730 and a communication bus 740, wherein the processor 710, the communications interface 720 and the memory 730 communicate with each other through the communication bus 740. The processor 710 may call the logic instructions in the memory 730 to execute the public cultural resource recommendation method based on the graph neural network, and the method includes: Processing a plurality of public cultural resource data to obtain resource characteristics and resource anchor points of each public cultural resource data, wherein each public cultural resource data includes a plurality of cultural resource points; According to the resource characteristics and resource anchor points of each public cultural resource data, a first relationship network between multiple cultural resource points and resource anchor points is constructed, and the first relationship network is used to characterize the resource similarity between multiple cultural resource points and resource anchor points in each public cultural resource data; Based on the first relationship network, construct a second relationship network of users and resource anchors; According to the second relationship network, obtaining the relationship change prediction result between the user and the resource anchor point; According to the relationship change prediction result and the first relationship network, a recommendation result of the public cultural resource is obtained to recommend the public cultural resource to the user according to the recommendation result.

[0115] In addition, the logic instructions in the above-mentioned memory 730 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present invention can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0116] On the other hand, the present invention further provides a computer program product, the computer program product includes a computer program, the computer program can be stored on a non-transitory computer-readable storage medium, and when the computer program is executed by a processor, the computer can execute the public cultural resource recommendation method based on graph neural network provided by the above methods, the method includes: Processing a plurality of public cultural resource data to obtain resource characteristics and resource anchor points of each public cultural resource data, wherein each public cultural resource data includes a plurality of cultural resource points; According to the resource characteristics and resource anchor points of each public cultural resource data, a first relationship network between multiple cultural resource points and resource anchor points is constructed, and the first relationship network is used to characterize the resource similarity between multiple cultural resource points and resource anchor points in each public cultural resource data; Based on the first relationship network, construct a second relationship network of users and resource anchors; According to the second relationship network, obtaining the relationship change prediction result between the user and the resource anchor point; According to the relationship change prediction result and the first relationship network, a recommendation result of the public cultural resource is obtained to recommend the public cultural resource to the user according to the recommendation result.

[0117] In another aspect, the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon, which is implemented when the computer program is executed by a processor to execute the public cultural resource recommendation method based on graph neural network provided by the above methods, the method comprising: Processing a plurality of public cultural resource data to obtain resource characteristics and resource anchor points of each public cultural resource data, wherein each public cultural resource data includes a plurality of cultural resource points; According to the resource characteristics and resource anchor points of each public cultural resource data, a first relationship network between multiple cultural resource points and resource anchor points is constructed, and the first relationship network is used to characterize the resource similarity between multiple cultural resource points and resource anchor points in each public cultural resource data; Based on the first relationship network, construct a second relationship network of users and resource anchors; According to the second relationship network, obtaining the relationship change prediction result between the user and the resource anchor point; According to the relationship change prediction result and the first relationship network, a recommendation result of the public cultural resource is obtained to recommend the public cultural resource to the user according to the recommendation result.

[0118] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0119] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0120] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for recommending public cultural resources based on graph neural network, characterized in that: include: Processing a plurality of public cultural resource data to obtain resource characteristics and resource anchor points of each public cultural resource data, wherein each public cultural resource data includes a plurality of cultural resource points; According to the resource characteristics and resource anchor points of each public cultural resource data, a first relationship network between multiple cultural resource points and resource anchor points is constructed, and the first relationship network is used to characterize the resource similarity between multiple cultural resource points and resource anchor points in each public cultural resource data; Based on the first relationship network, construct a second relationship network of users and resource anchors; According to the second relationship network, obtaining the relationship change prediction result between the user and the resource anchor point; According to the relationship change prediction result and the first relationship network, a recommendation result of the public cultural resource is obtained to recommend the public cultural resource to the user according to the recommendation result.

2. The method for recommending public cultural resources based on graph neural network according to claim 1 is characterized in that: The step of constructing a first relationship network between cultural resource points and resource anchor points according to the resource characteristics and resource anchor points of each public cultural resource data includes: According to the resource characteristics of each public cultural resource data, the characteristic similarity between each cultural resource point and the resource anchor point is calculated; According to the feature similarity between each cultural resource point and the resource anchor point, a first relationship network is obtained.

3. The method for recommending public cultural resources based on graph neural network according to claim 1 is characterized in that: The step of constructing a second relationship network between users and resource anchors based on the first relationship network includes: Acquire cultural resource points browsed by the user in a historical period to obtain at least one first cultural resource point; According to the first relationship network, query the first resource anchor point corresponding to the first cultural resource point; at the same time, obtain the resource similarity between the first cultural resource point and the first resource anchor point; Obtaining an association weight between the user and the first resource anchor point according to the resource similarity between the first cultural resource point and the first resource anchor point; A second relationship network is constructed according to the association weights between the user and the first resource anchor point.

4. The method for recommending public cultural resources based on graph neural network according to claim 3 is characterized in that: Also includes: When the user generates a browsing record of a cultural resource point after the historical period, obtaining a second cultural resource point browsed by the user; According to the first relationship network, a second resource anchor point corresponding to the second cultural resource point is obtained, and at the same time, a resource similarity between the second cultural resource point and the second resource anchor point is obtained; The second relationship network is updated according to the resource similarity between the second cultural resource point and the second resource anchor point.

5. The method for recommending public cultural resources based on graph neural network according to claim 1 is characterized in that: The obtaining, according to the second relationship network, a prediction result of a relationship change between the user and the resource anchor point includes: Acquire the second relationship network multiple times at preset time intervals to obtain a first relationship network sequence between the user and the resource anchor point; and acquire the second relationship network sequences between other users and the resource anchor points to obtain multiple second relationship network sequences; Calculate the relationship network similarity between the second relationship network most recently obtained in the first relationship network sequence and each second relationship network in each second relationship network sequence to obtain a calculation result; According to the calculation result, at least one similar user is selected from other users to obtain a third relationship network sequence between the similar users and the resource anchor points; According to the first relationship network sequence and the third relationship network sequence, a relationship change prediction result between the user and the resource anchor point is obtained.

6. The method for recommending public cultural resources based on graph neural network according to claim 5 is characterized in that: The step of obtaining the predicted result of the relationship change between the user and the resource anchor point according to the first relationship network sequence and the third relationship network sequence includes: In each third relationship network sequence, the second relationship networks are sorted from large to small according to the relationship network similarity, and the similar relationship network ranked first is obtained; In each third relationship network sequence, a second relationship network obtained after a preset time interval of each similar relationship network is obtained to obtain a predicted relationship network; According to the predicted relationship network, the predicted results of the relationship changes between users and resource anchors are obtained.

7. The method for recommending public cultural resources based on graph neural network according to claim 6 is characterized in that: Also includes: When the similar relationship network does not have a second relationship network after a preset time interval, the user corresponding to the similar relationship network is excluded from the similar users.

8. A public cultural resource recommendation device based on graph neural network, characterized in that: include: A data processing module is used to process a plurality of public cultural resource data to obtain resource characteristics and resource anchor points of each public cultural resource data, wherein each public cultural resource data includes a plurality of cultural resource points; A first relationship building module, used to build a first relationship network between multiple cultural resource points and resource anchor points according to the resource characteristics and resource anchor points of each public cultural resource data, wherein the first relationship network is used to characterize the resource similarity between multiple cultural resource points and resource anchor points in each public cultural resource data; A second relationship building module, used to build a second relationship network between users and resource anchors based on the first relationship network; A relationship prediction module, used to obtain a prediction result of a relationship change between a user and a resource anchor point according to the second relationship network; The resource recommendation module is used to obtain the recommendation results of public cultural resources according to the relationship change prediction results and the first relationship network, so as to recommend the public cultural resources to the user according to the recommendation results.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, it implements the public cultural resource recommendation device method based on graph neural network as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, it implements the public cultural resource recommendation device method based on graph neural network as described in any one of claims 1 to 7.

Images