Patent information processing method, device, electronic device and storage medium

Abstract

This application provides a patent information processing method, device, electronic device, and storage medium. The method includes: obtaining target patent content information and target patent description information of a target patent, and reference patent content information and reference patent description information of reference patents that have a citation relationship with the target patent; constructing a first node using the target patent content information and target patent description information, constructing a second node using the reference patent content information and reference patent description information, and constructing a node relationship corresponding to a preset time point using the patent citation relationship corresponding to the preset time point; constructing a patent network information graph with time sequence information based on the first node, the second node, and the node relationship corresponding to the preset time point; and determining target patent evaluation information with time sequence information based on the patent network information graph with time sequence information. This application can obtain patent evaluation information that changes over time, and this information can reflect the characteristics of patent evolution.

Description

Technical Field

[0001] The present application relates to the field of information processing technology, and specifically to a patented information processing method, device, electronic device and storage medium. Background Art

[0002] In order to reflect the value of a patent, it is necessary to evaluate the patent. In the current patent evaluation process, patent evaluators mainly conduct evaluations based on patent materials. This method relies on the manual processing capabilities of patent personnel. However, due to the limited ability of humans to analyze patent materials, faced with a large amount of patent materials, it is impossible to analyze the changes in the evaluation information of the patent at different times from the materials. Therefore, it can only obtain static evaluation information of the patent, and static evaluation information is often one-sided and can only represent the status of the patent at a certain point in time. In practical applications, it is often unable to accurately reflect the evolving value of the patent, and has poor effectiveness and low information content. Therefore, the manual patent evaluation method cannot match the evolution characteristics of patent information, and the evaluated information does not conform to the characteristics of patent evolution. Summary of the Invention

[0003] One purpose of this application is to propose a patent information processing method, device, electronic device and storage medium, which can obtain patent evaluation information that changes over time, so that the evaluation of patents can reflect the characteristics of patent evolution.

[0004] According to one aspect of an embodiment of the present application, a patent information processing method is disclosed, the method comprising:

[0005] Obtain target patent content information and target patent description information of the target patent, and reference patent content information and reference patent description information of reference patents that have a citation relationship with the target patent;

[0006] A first node is constructed using the target patent content information and the target patent description information, a second node is constructed using the reference patent content information and the reference patent description information, and a node relationship corresponding to a preset time point is constructed using a patent citation relationship corresponding to a preset time point, wherein the patent citation relationship is used to describe the citation relationship between the target patent and the reference patent, and the node relationship is used to describe the relationship between the first node and the second node;

[0007] Constructing a patent network information graph with time sequence information based on the first node, the second node, and the node relationship corresponding to the preset time point;

[0008] Based on the patent network information graph with time sequence information, target patent evaluation information with time sequence information is determined.

[0009] According to one aspect of an embodiment of the present application, a patent information processing device is disclosed, the patent information processing device comprising:

[0010] An acquisition module, configured to acquire target patent content information and target patent description information of a target patent, and reference patent content information and reference patent description information of reference patents that have a citation relationship with the target patent;

[0011] a first determining module configured to construct a first node using the target patent content information and the target patent description information, construct a second node using the reference patent content information and the reference patent description information, and construct a node relationship corresponding to a preset time point using a patent citation relationship corresponding to a preset time point, wherein the patent citation relationship is used to describe the citation relationship between the target patent and the reference patent, and the node relationship is used to describe the relationship between the first node and the second node;

[0012] A construction module, configured to construct a patent network information graph having time sequence information based on the first node, the second node, and the node relationship corresponding to the preset time point;

[0013] The second determination module is used to determine the target patent evaluation information with time sequence information based on the patent network information graph with time sequence information.

[0014] In an exemplary embodiment of the present application, the device is configured as follows:

[0015] Using a linear layer of a preset time-varying graph neural network model to perform data conversion on the first node, the second node, and the node relationship corresponding to the preset time point to obtain target data;

[0016] Using a time-varying spatial convolution module and a multi-scale temporal convolution module of a preset time-varying graph neural network model to alternately extract the spatiotemporal features of the target data;

[0017] The readout module of the preset time-varying graph neural network model is used to process the spatiotemporal features to construct a patent network information graph with time series information.

[0018] In an exemplary embodiment of the present application, the device is configured as follows:

[0019] Using the time-varying spatial convolution module to learn the embedded features of the target data to obtain a stable adjacency matrix;

[0020] Using the time-varying spatial convolution module to perform convolution processing on the target data to obtain a dynamic adjacency matrix;

[0021] Constructing an attention vector between patent nodes, and based on the attention vector, normalizing the dynamic adjacency matrix to obtain normalized data, wherein the patent nodes include the first node and the second node;

[0022] Determining a time-varying graph convolution node based on the standardized data and the stable adjacency matrix;

[0023] For the time-varying graph convolution node, the multi-scale temporal convolution module is used to concatenate the outputs of multiple layers in an extended causal convolution to extract the spatiotemporal features.

[0024] In an exemplary embodiment of the present application, the device is configured as follows:

[0025] Determine, based on the patent network information graph having time series information, the information network centrality of the target patent corresponding to the time series information, where the information network centrality is closeness centrality, betweenness centrality, or eigenvector centrality;

[0026] Determining the importance of the target patent and time information corresponding to the importance based on the information network centrality corresponding to the time series information;

[0027] The target patent evaluation information is determined based on the importance of the target patent and time information corresponding to the importance.

[0028] In an exemplary embodiment of the present application, the device is configured as follows:

[0029] Based on the importance of the target patent and the time information corresponding to the importance, a time-varying importance curve of the target patent is constructed, and the time-varying importance curve of the target patent is used as the target patent evaluation information.

[0030] In an exemplary embodiment of the present application, the device is configured as follows:

[0031] Using a word embedding model to encode the target patent content information to obtain a high-dimensional dense vector;

[0032] Encoding the target patent description information to obtain a first sparse feature vector;

[0033] The first node is constructed based on the high-dimensional dense vector and the first sparse feature vector.

[0034] In an exemplary embodiment of the present application, the device is configured as follows:

[0035] Encoding the target patent citation relationship to obtain a second sparse feature vector;

[0036] Generate a target node feature including the high-dimensional dense vector, the first sparse feature vector, and the second sparse feature vector;

[0037] Processing the target node feature and the preset node feature using a deep layer of the deep network to obtain first output information;

[0038] Processing the target node feature and the preset node feature using a cross layer of the cross network to obtain second output information;

[0039] Splicing the first output information and the second output information to obtain spliced ​​information;

[0040] Determining the similarity between the target patent and the reference patent based on the splicing information;

[0041] If the similarity meets the preset similarity condition, the recommendation information corresponding to the reference patent is determined.

[0042] According to one aspect of an embodiment of the present application, an electronic device is disclosed, comprising: one or more processors; a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the electronic device implements the methods provided in the various optional implementations described above.

[0043] According to one aspect of an embodiment of the present application, a computer program medium is disclosed, on which computer-readable instructions are stored. When the computer-readable instructions are executed by a processor of a computer, the computer executes the methods provided in the various optional implementations described above.

[0044] According to one aspect of an embodiment of the present application, a computer program product or computer program is provided, comprising computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the methods provided in the various optional implementations described above.

[0045] Technical effects:

[0046] In the embodiments of the present application, a patent network information graph with time-series information is constructed by analyzing the citation relationship between the target patent and the reference patent over time. Based on the patent network information graph with time-series information, patent evaluation information with time-series information is determined. This avoids the problem of missing evaluation information and the resulting low information validity and poor reference value that exists in manual analysis of patent materials. In addition, patent evaluation information that conforms to the characteristics of time evolution can be obtained by combining the patent citation relationship that evolves over time. This allows the patent evaluation to conform to the characteristics of patent evolution.

[0047] Other features and advantages of the present application will become apparent from the following detailed description, or may be learned in part by practice of the present application.

[0048] It should be understood that the foregoing general description and the following detailed description are merely illustrative and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] The above and other objects, features and advantages of the present application will become more apparent by describing in detail example embodiments thereof with reference to the attached drawings.

[0050] Figure 1 The figure is a flowchart of a patent information processing method according to an embodiment of the present application.

[0051] Figure 2 A schematic diagram comparing static patent networks and dynamic patent networks.

[0052] Figure 3 Schematic diagram of a process for processing patent information according to another embodiment of the present application.

[0053] Figure 4 Schematic diagram of the structure of the preset time-varying graph neural network model.

[0054] Figure 5 Schematic diagram of a process for processing patent information according to another embodiment of the present application.

[0055] Figure 6 2 is a flow chart of a method for processing patent information according to another embodiment of the present application.

[0056] Figure 7 A schematic diagram of the software architecture involved in the patented information processing method.

[0057] Figure 8 Schematic diagram of the recommended model.

[0058] Figure 9 Schematic diagram of the structure of a patent information processing device according to an embodiment of the present application.

[0059] Figure 10 FIG. 1 is a schematic structural diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0060] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that the description of this application will be more comprehensive and complete and will fully convey the concepts of the example embodiments to those skilled in the art. The accompanying drawings are merely schematic illustrations of the present application and are not necessarily drawn to scale. Identical reference numerals in the figures indicate identical or similar parts, and thus repeated descriptions thereof will be omitted.

[0061] In addition, the described features, structures or characteristics can be combined in one or more example embodiments in any suitable manner. In the following description, many specific details are provided to provide a full understanding of the example embodiments of the present application. However, those skilled in the art will appreciate that the technical solutions of the present application can be practiced while omitting one or more of the specific details, or other methods, components, steps, etc. can be adopted. In other cases, known structures, methods, implementations or operations are not shown or described in detail to avoid obscuring the main content and making various aspects of the present application vague.

[0062] Some of the blocks shown in the accompanying drawings are functional entities that do not necessarily correspond to physically or logically independent entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0063] Figure 1 The flowchart of the patent information processing method provided in this application is shown, and the method includes:

[0064] Step S101: Obtain target patent content information and target patent description information of the target patent, and reference patent content information and reference patent description information of reference patents that have a citation relationship with the target patent.

[0065] The target patent is the patent that needs to be evaluated. There can be one or more target patents. If there are multiple target patents, a patent network information graph with time-series information can be constructed for multiple target patents at the same time to reflect the trend of the citation relationship between multiple target patents over time, as well as the evaluation information of multiple target patents over time. The target patent content information is the content part of the target patent, mainly including the part of the application document of the target patent that records the substantive content of the patent, such as the abstract and the main text. The main text can include the claims and description. The target patent description information is used to describe the relevant information of the patent, such as: classification number, authorization information, etc.

[0066] The citation relationship between patents is a major factor in reflecting the value of a patent. The more times a patent is cited, the higher its centrality in the patent network, and the more important the patent is likely to be. For example, if a patent is cited as a comparative document by multiple other patents during the examination stage, it indicates that the content described in the patent includes important technical solutions in the field, making it a relatively important patent. This can also objectively reflect the value of the patent and avoid the technical problems of high subjectivity, one-sidedness, and uneven accuracy that arise from evaluating using various artificial evaluation rules.

[0067] Reference patents are patents other than the target patent. In this embodiment, the patent network information graph is primarily constructed based on the relevant information of reference patents that have a citation relationship with the target patent. Reference patent content information is the content portion of the reference patent, mainly including the portion of the reference patent that records the substantive content of the patent, such as the abstract and the main text, which may include the claims and the specification. Reference patent description information is used to describe the relevant information of the reference patent, such as: classification number, authorization information, etc.

[0068] It is understandable that in order to avoid an excessive number of reference patents, patents from the most recent period may be selected as reference patents. Furthermore, in combination with the technical field of the patent to be evaluated, only patents in specific technical fields may be selected as reference patents.

[0069] The citation relationship between the reference patent and the target patent can be obtained based on the comparative document information recorded in the authorization text of the reference patent or the target patent, or based on the comparative document information involved in the examination process, or based on the comparative document information involved in the invalidation stage or the litigation stage.

[0070] For example, during the examination phase, if patent A cites patents B and C as references, this indicates a citation relationship between patent A and both patents B and C. Furthermore, patent A cites patents B and C, and both patents B and C are cited by patent A. In other words, patent citation relationships can include both active citations of other patents and citations by other patents.

[0071] It is understood that patent citation relationships can be primarily derived from publicly available patent data at the examination, grant, invalidation, and litigation stages. However, citation relationships between target patents and reference patents can also be derived from other data sources, as long as the patent citation relationships can reflect the correlation between the target patent and the reference patent.

[0072] Step S102: construct a first node with the target patent content information and the target patent description information, construct a second node with the reference patent content information and the reference patent description information, and construct a node relationship corresponding to a preset time point with the patent citation relationship corresponding to the preset time point.

[0073] The patent citation relationship is used to describe the citation relationship between the target patent and the reference patent, and the node relationship is used to describe the relationship between the first node and the second node.

[0074] The first node corresponds to the target patent, and the second node corresponds to the reference patent. These first and second nodes are used to construct a patent network information graph. Patent citation relationships change over time. For example, patent A is cited by n patents at time T1, but at time T2, due to factors such as examination status updates, it is cited by n+1 patents, causing the citation relationship between A and other patents to change. Patent citation relationships at preset time points can be obtained to reflect the temporal evolution of patent citation relationships.

[0075] Patent citation relationships can, to a certain extent, reflect the interconnectedness between patents and, from a certain perspective, the inherent connections between them. Compared to methods that evaluate only the information of the patents themselves, this embodiment introduces patent citation relationships to avoid the problem of isolated patents that fail to form a complete patent network. It can mine key network information for more accurate patent evaluation.

[0076] Step S103: Based on the first node, the second node, and the node relationship corresponding to the preset time point, a patent network information graph with time sequence information is constructed.

[0077] like Figure 2 As shown, Figure 2The following is a schematic diagram comparing static and dynamic patent networks. A through G represent the nodes corresponding to patents A through G, respectively. In the static patent network, based on the relationships between nodes, it can be determined that at time T, Patent E has a citation relationship with Patents B, D, G, and F; Patent B has a citation relationship with Patent A; and Patent D has a citation relationship with Patent G. This static patent network only identifies specific patent citation relationships and node relationships at time T, failing to capture the evolution of patent citation relationships over time. In contrast, the dynamic patent network shows the dynamic patent network from time T to time T+P-1 and time T+P. It can be seen that from time T to time T+P-1, the citation relationship between Patent E and Patent C changes from zero to one, while from time T+P-1 to time T+P, the patent citation relationship remains unchanged. This dynamic patent network demonstrates that Patent E is cited by more patents over time, whereas a static patent network alone cannot capture this evolution.

[0078] Step S104: Based on the patent network information graph with time sequence information, determine target patent evaluation information with time sequence information.

[0079] Target patent evaluation information is information about the target patent. This information can include the importance of the target patent. By quantitatively evaluating the centrality of each node in the patent network graph, the importance of the patent corresponding to that node can be derived based on the centrality.

[0080] The patent evaluation method of the embodiment of the present application can avoid the problem that static graphs cannot process changes in patent citation relationships. It can mine the citation relationships of patents that evolve over time based on dynamic patent network information graphs, and further obtain patent evaluation information that evolves over time.

[0081] In addition, the above method can also avoid the problem of subjective factors in manual patent evaluation interfering with the stability and accuracy of patent evaluation, and can evaluate patents intelligently, efficiently and accurately.

[0082] In the embodiments of this application, a patent network information graph with time-series information is constructed by analyzing the citation relationships between the target patent and the reference patents over time. Patent evaluation information with time-series information is then determined based on the patent network information graph with time-series information. This allows the patent evaluation information to be derived that conforms to the characteristics of time-series evolution, combining the patent citation relationships that evolve over time. This allows the evaluation information to reflect the changes in the patent over time, ensuring that the patent evaluation conforms to the characteristics of patent evolution.

[0083] Figure 3 This is a flowchart of a patent information processing method according to another embodiment of the present application. The patent information processing method includes:

[0084] Step S201: Obtain target patent content information and target patent description information of the target patent, and reference patent content information and reference patent description information of reference patents that have a citation relationship with the target patent.

[0085] Step S202: construct a first node with the target patent content information and the target patent description information, construct a second node with the reference patent content information and the reference patent description information, and construct a node relationship corresponding to a preset time point with the patent citation relationship corresponding to the preset time point.

[0086] The patent citation relationship is used to describe the citation relationship between the target patent and the reference patent, and the node relationship is used to describe the relationship between the first node and the second node.

[0087] Step S203: using the linear layer of the preset time-varying graph neural network model to perform data conversion on the first node, the second node, and the node relationship corresponding to the preset time point to obtain target data;

[0088] The preset time-varying graph neural network model is a pre-obtained neural network model that can be used to construct a patent network information graph with time series information. Graphs are generally used to represent objects, such as images and social networks, for pattern classification. In a patent network, the citation relationship between the target patent and other patents can evolve over time, thereby having certain time series differences. Therefore, this embodiment uses a time-varying graph neural network TV-GCN to perform time series analysis on the node relationships of the patent relationship network. On the basis of the traditional graph neural network, time series information is added to form dynamic patent graph data to more accurately understand the patent network of customers at different time nodes.

[0089] The structure of the preset time-varying graph neural network model is as follows Figure 4 As shown. Figure 4 In the proposed method, the time-varying graph neural network (TV-GCN) first uses a linear layer to transform the input data. It then alternately uses multiple multi-scale temporal convolution modules (GMS-TCMs) and multiple time-varying spatial convolution modules (TV-SCMs) to extract the spatiotemporal features of the patent data. The readout module then establishes temporal relationships and outputs them. The TV-SCM is composed of a diffuse convolutional network with a stable and dynamic adjacency matrix generation mechanism. This allows the TV-SCM to improve the stability of capturing patent data and generate a wider range of dynamic spatial patterns.

[0090] Step S204: using the time-varying spatial convolution module and the multi-scale temporal convolution module of the preset time-varying graph neural network model to alternately extract the spatiotemporal features of the target data;

[0091] As an optional implementation, the spatiotemporal features include time features and space features.

[0092] As an optional implementation, the spatial feature can be characterized by the technical field, which can use the first four digits of the IPC classification number. The technical field can better describe the spatial correlation between patents.

[0093] As an optional implementation, a time-varying spatial convolution module and a multi-scale temporal convolution module of a preset time-varying graph neural network model are used to alternately extract the temporal and spatial features of the target data, including: using the time-varying spatial convolution module to learn the embedded features of the target data to obtain a stable adjacency matrix; using the time-varying spatial convolution module to perform convolution processing on the target data to obtain a dynamic adjacency matrix; constructing an attention vector between patent nodes, and based on the attention vector, normalizing the dynamic adjacency matrix to obtain standardized data, where the patent node includes a first node and a second node; determining the time-varying graph convolution node based on the standardized data and the stable adjacency matrix; for the time-varying graph convolution node, using the multi-scale temporal convolution module to concatenate the outputs of multiple layers in the extended causal convolution to extract the temporal and spatial features.

[0094] Step S205: Using a readout module of a preset time-varying graph neural network model to process the spatiotemporal features to construct a patent network information graph with temporal information.

[0095] Step S206: Based on the patent network information graph with time sequence information, determine the target patent evaluation information with time sequence information.

[0096] The following is a detailed description of how to construct a patent network information graph.

[0097] In order to effectively capture the stable and dynamic spatial correlations between patent nodes, the time-varying spatial convolution module TV-SCM generates stable and dynamic adjacency matrices without any prior knowledge. The specific algorithm is as follows.

[0098] The stable adjacency matrix is ​​estimated by two relatively stable learnable node embedding vectors, and the formula is:

[0099]

[0100] Among them, E s 、E t Represent the embedded features of the source node and the target node respectively, which are parameters that can be learned. In addition, It has nothing to do with the input features, but only with the input relationship. Once the relationship information is confirmed, It remains stable in this dataset.

[0101] In order to further capture the dynamic spatial correlation between patents, the dynamic adjacency matrix is ​​determined by the input features. The formula of dynamic features is:

[0102]

[0103] Among them, d is a parameter, which represents the impact expansion rate of a single patent in the time series. The impact expansion rate is used to describe the degree of influence on other patents. t is the node feature input at time t, TC d (x) is the output feature at time t after convolution. In order to obtain as much embedded information as possible, the final result incorporates the influence of multi-layer expansion rate, as shown in the following formula:

[0104]

[0105] H represents the embedded features of node X that contain historical time series information. The model adds an attention mechanism to construct a common attention vector a between nodes:

[0106] e ij =Tanh(a|h i ||h j | T );

[0107] The dynamic adjacency matrix can be used as model input data after normalization:

[0108]

[0109] Where [x] + =max{0,x},τ satisfies ∑[e ij -τ(e i,: )] + = 1 threshold function. The traditional graph convolution output can be represented by the adjacency matrix, eigenvector and transfer matrix:

[0110]

[0111] This embodiment adds the processing of K-step time series on this basis, so it can be rewritten as:

[0112]

[0113] Combining the previously obtained adjacency matrix at each time point, the output at time point K is:

[0114]

[0115] The time-varying graph convolution node can be expressed as:

[0116] Xscm ={Z t-T+1 ,…,Z t-1 ,Z t};

[0117] The Multi-Scale Temporal Convolution (GMS-TCM) module addresses the problem of information loss in the temporal dimension often seen in existing temporal feature extraction algorithms. By concatenating the outputs of several layers in an extended causal convolution, both short-term and long-term temporal features can be effectively extracted. After processing through these two modules, the network extracts input feature information for the complete time period.

[0118] Specifically, extended causal convolution extracts the temporal features of patent sequences by stacking multiple layers of one-dimensional convolution with increasing expansion rates. l Node v on the channel i The output of the lth layer in the extended causal convolution is calculated as follows

[0119]

[0120] in, l=0,1,…,L. d is the expansion rate of time series. c l-1 and c l Indicates the number of input and output channels respectively.

[0121] The readout module establishes the time dependency between multiple moments based on the above information and outputs a patent network information graph with time sequence differences.

[0122] In an embodiment of the present application, the linear layer of a preset time-varying graph neural network model is used to perform data conversion on the first node, the second node, and the node relationship corresponding to the preset time point to obtain target data. The time-varying spatial convolution module and the multi-scale temporal convolution module of the preset time-varying graph neural network model are used to alternately extract the spatiotemporal features of the target data. The spatiotemporal features are then processed using the readout module of the preset time-varying graph neural network model to construct a patent network information graph with temporal information. This improves the stability of captured patent data, generates a wider range of dynamic spatial patterns, and outputs a patent network information graph with temporal differences.

[0123] Figure 5 1 is a flow chart of a patent information processing method according to another embodiment of the present application. The patent information processing method includes:

[0124] Step S301: Obtain target patent content information and target patent description information of the target patent, and reference patent content information and reference patent description information of reference patents that have a citation relationship with the target patent.

[0125] Step S302: construct a first node with the target patent content information and the target patent description information, construct a second node with the reference patent content information and the reference patent description information, and construct a node relationship corresponding to a preset time point with the patent citation relationship corresponding to the preset time point.

[0126] The patent citation relationship is used to describe the citation relationship between the target patent and the reference patent, and the node relationship is used to describe the relationship between the first node and the second node.

[0127] Step S303: Based on the first node, the second node, and the node relationship corresponding to the preset time point, a patent network information graph with time sequence information is constructed.

[0128] Step S304: Based on the patent network information graph with time series information, determine the information network centrality of the target patent corresponding to the time series information, where the information network centrality is closeness centrality, betweenness centrality, or eigenvector centrality.

[0129] Information network centrality, also known as degree centrality, is a direct measure of node centrality in network analysis. The greater a node's degree, the higher its degree centrality and the more important the node is in the network.

[0130] As an optional implementation, by determining the closeness centrality, closeness centrality is used as the information network centrality. Closeness centrality measures the average distance between a vertex and other vertices. The higher the value, the shorter the distance between the vertex and other vertices. The formula is:

[0131]

[0132] Among them, l(v i ,v j ) represents the vertex v i With vertex v j The shortest path length between .

[0133] As an optional implementation, betweenness centrality is determined and used as the centrality of the information network. Betweenness centrality measures the ratio of the shortest path connecting two points in the network to the total number of shortest paths between the two points. The formula is:

[0134]

[0135] Among them, σ st (u) represents the number of shortest paths from s to t passing through node u, σ stIndicates the number of shortest paths from s to t.

[0136] As an optional implementation, by determining the eigenvector centrality, the eigenvector centrality is used as the information network centrality. For the eigenvector centrality, the importance of a node depends on both the number of its neighbor nodes (i.e., the degree of the node) and the importance of its neighbor nodes. The formula is:

[0137]

[0138] Among them, A=(a i,j ) is the adjacency matrix, x is the eigenvector corresponding to the eigenvalue λ of matrix A, and the i-th value of x is the eigenvector centrality of the i-th node.

[0139] Step S305: Based on the information network centrality corresponding to the time series information, determine the importance of the target patent and the time information corresponding to the importance.

[0140] Step S306: Determine target patent evaluation information based on the importance of the target patent and time information corresponding to the importance.

[0141] As an optional implementation, target patent evaluation information is determined based on the importance of the target patent and the time information corresponding to the importance, including: constructing an importance curve of the target patent that changes over time based on the importance of the target patent and the time information corresponding to the importance, and using the importance curve of the target patent that changes over time as the target patent evaluation information.

[0142] Among them, the importance curve can intuitively show the changing trend of patent evaluation information, but the embodiment of the present application is not limited to using only the importance curve as the target patent evaluation information, and other data representation forms can also be used to present the target patent evaluation information, such as a table form.

[0143] In one embodiment, graph data is first constructed using patent information as nodes, with citation-to-citation relationships as node relationships, and IPC classification numbers and constructed patent factors as node features. Time series information is then incorporated to construct a time-varying patent technology network across different time layers. Information network centrality analysis is then applied to the graph data to quantitatively assess the importance of each patent technology node. By incorporating time series information, a time curve of the importance of each patent can be generated, providing a more comprehensive and detailed depiction of the patent technology.

[0144] In this embodiment, the information network centrality of the target patent corresponding to the time series information is determined based on a patent network information graph with time series information. Based on the information network centrality corresponding to the time series information, the target patent's importance and the time information corresponding to the importance are determined. Furthermore, based on the target patent's importance and the time information corresponding to the importance, target patent evaluation information is determined. This allows for an assessment of the importance of patents over time, providing a more comprehensive and detailed representation of the evolving trends in patent importance.

[0145] Figure 6 The following is a flowchart of a method for processing patent information according to another embodiment of the present application. The method for processing patent information includes:

[0146] Step S401: Obtain target patent content information and target patent description information of the target patent, and reference patent content information and reference patent description information of reference patents that have a citation relationship with the target patent.

[0147] Step S402: Use a word embedding model to encode the target patent content information to obtain a high-dimensional dense vector.

[0148] The target patent content information mainly includes the main text and abstract parts recorded in the patent application documents. The main text and abstract parts are mainly text information, and text information cannot be directly modeled by the neural network. In order to enable the neural network to model, this embodiment adopts a word embedding model to discretize the text features and convert the unstructured text data into discrete features so that it can be input into the model for processing. Compared with the method of using one-hot encoding for processing, this embodiment uses a word embedding model for processing, which can avoid the problem that as the size of the dictionary increases, the dimension of the bag-of-words model composed of sentences becomes larger and larger, and the matrix becomes sparser, resulting in low processing efficiency. High-dimensional dense vectors can be obtained efficiently.

[0149] Step S403: Encode the target patent description information to obtain a first sparse feature vector.

[0150] Step S404: Based on the high-dimensional dense vector and the first sparse feature vector, construct a first node to construct a second node with reference to the patent content information and the reference to the patent description information, and construct a node relationship corresponding to the preset time point with the patent citation relationship corresponding to the preset time point.

[0151] The patent citation relationship is used to describe the citation relationship between the target patent and the reference patent, and the node relationship is used to describe the relationship between the first node and the second node.

[0152] Step S405: Based on the first node, the second node, and the node relationship corresponding to the preset time point, a patent network information graph with time sequence information is constructed.

[0153] Step S406: Based on the patent network information graph with time sequence information, determine the target patent evaluation information with time sequence information.

[0154] In one embodiment, if Figure 7 As shown, Figure 7 The figure is a schematic diagram of the software architecture involved in the patent information processing method. The software architecture includes a data layer, a model layer, and an application layer. The data layer includes a public patent database, which can include parameters such as the patent title, text, abstract, citations, citations, IPC classification, and authorization information. The model layer includes a text processing module and a relationship processing module. The text processing module includes a recommendation algorithm that combines word embedding and deep cross network (DCN). The text information of the patent text and abstract is processed and incorporated into the model training, and similar patents are recommended based on the output patent similarity matrix.

[0155] Furthermore, the relationship processing module consists of a time-varying graph neural network (TV-GCN) and information network centrality. The TV-GCN is primarily responsible for constructing a time-varying patent information network based on the citation relationships and temporal information of patent information. Information network centrality is then used to assess the importance of different patents within the time-varying patent information network.

[0156] After processing through the above two modules, the trained model is packaged as an offline model for user use. The application layer has evaluation and recommendation functions. When a user enters a patent in the database, the model automatically searches the database, retrieves the offline model, and evaluates the patent based on the patent similarity matrix and time-varying network importance, recommending similar patents for user reference.

[0157] The text processing module primarily provides patent recommendation functionality, and some of the information processed in the module can be used for patent evaluation. To achieve patent recommendation, word embedding is performed on the patent content of each patent, and the text information of the patent content is processed into a high-dimensional dense vector. This vector is then combined with other sparse features in the patent information, such as grant information, IPC classification number, and citation status, and then input into the DCN algorithm training model. The resulting output is a patent similarity matrix, forming an offline patent recommendation model.

[0158] In one embodiment, in order to achieve similar patent recommendation, after encoding the target patent description information to obtain a first sparse feature vector, the method also includes: encoding the target patent citation relationship to obtain a second sparse feature vector; generating target node features including a high-dimensional dense vector, a first sparse feature vector and a second sparse feature vector; using the deep layer of the deep network to process the target node features and the preset node features to obtain first output information; using the cross layer of the cross network to process the target node features and the preset node features to obtain second output information; splicing the first output information and the second output information to obtain spliced ​​information; determining the similarity between the target patent and the reference patent based on the spliced ​​information; if the similarity meets the preset similarity condition, determining the recommendation information corresponding to the reference patent.

[0159] The preset similarity condition may be, for example, that the similarity is greater than a preset threshold; or that the similarity is a preset number of similarities that are the highest among all similarities. If the similarity satisfies the preset similarity condition, the reference patent corresponding to the similarity that satisfies the preset similarity condition is selected as the patent to be recommended, and the relevant information of the patent to be recommended is used as the recommendation information.

[0160] The following, combined Figure 8 Explain the recommendation model. Figure 8 This is a structural diagram of the recommendation model. Figure 8 The Embedding and Stacking layer is the input layer, the CrossNetwork is the cross network, the DeepNetwork is the deep network, and the CombinationOutputLayer is the output layer. The input features include the processed dense feature vector and the corresponding sparse features of the patent description information. The sparse features are directly used as model input after embedding. All features are then input into the CrossNetwork and DeepNetwork networks. This process is called feature extraction. Finally, the results of the CrossNetwork and DeepNetwork networks are spliced ​​together to form the output of whether to recommend or not, finally completing the recommendation algorithm model.

[0161] For the embedding and stacking layers, the input data includes both sparse and dense features. In prediction tasks for web-scale recommendation systems, the input is primarily feature-rich, such as "Applicant's nationality = XX." These features are typically encoded as one-hot vectors such as [0, 1, 0]. However, this approach creates an extremely high-dimensional feature space. To reduce the number of dimensions, this embodiment uses a word embedding model to convert the data into real-valued dense vectors, and then superimposes the embedded vectors and normalized dense features to form a single vector.

[0162] For the cross network, the idea of ​​the cross network is to apply explicit feature cross in an efficient way. The cross network consists of cross layers, each layer has the following formula:

[0163]

[0164] Where: x l+1 is a column vector, representing the output from the lth and (l+1)th layer cross network respectively; w l , b l ∈R d are the weight and bias parameters of the lth layer. After completing a feature cross f, each cross layer will add its input back to iterate.

[0165] For deep networks, the small number of parameters in the cross network limits the model's capabilities. Therefore, DCN (Deep & Cross Network) introduces a deep network in parallel to capture high-order nonlinear crosses. The formula for each deep layer is as follows:

[0166] h l+1 =f(W l h l +b l );

[0167] Among them, h l ∈R nl ,h l+1 ∈R nl+1 are the hidden layers of layer l and layer l+1 respectively; W l 、b l is the parameter of the lth layer. f is the ReLU equation.

[0168] In this embodiment, the word embedding model is used for processing, which can avoid the problem of low processing efficiency caused by one-hot encoding, where the dimension of the bag-of-words model composed of sentences becomes larger and larger as the dictionary size increases, and the matrix becomes sparser. This can reduce the number of dimensions and complete data preprocessing efficiently.

[0169] Figure 9 The following is a schematic diagram of the structure of a patent information processing device according to an embodiment of the present application. The patent information processing device includes:

[0170] An acquisition module 501 is used to acquire target patent content information and target patent description information of a target patent, and reference patent content information and reference patent description information of reference patents that have a citation relationship with the target patent;

[0171] A first determining module 502 is configured to construct a first node using the target patent content information and the target patent description information, construct a second node using the reference patent content information and the reference patent description information, and construct a node relationship corresponding to a preset time point using the patent citation relationship corresponding to a preset time point, wherein the patent citation relationship is used to describe the citation relationship between the target patent and the reference patent, and the node relationship is used to describe the relationship between the first node and the second node;

[0172] A construction module 503 is used to construct a patent network information graph with time sequence information based on the first node, the second node, and the node relationship corresponding to the preset time point;

[0173] The second determining module 504 is configured to determine target patent evaluation information having time sequence information based on the patent network information graph having time sequence information.

[0174] In an exemplary embodiment of the present application, the patent information processing device is configured as follows:

[0175] Using the linear layer of a preset time-varying graph neural network model, data conversion is performed on the first node, the second node, and the node relationship corresponding to the preset time point to obtain target data;

[0176] The time-varying spatial convolution module and the multi-scale temporal convolution module of the preset time-varying graph neural network model are used to alternately extract the spatiotemporal features of the target data;

[0177] The readout module of the preset time-varying graph neural network model is used to process the spatiotemporal features to construct a patent network information graph with temporal information.

[0178] In an exemplary embodiment of the present application, the patent information processing device is configured as follows:

[0179] A time-varying spatial convolution module is used to learn the embedded features of the target data and obtain a stable adjacency matrix;

[0180] The time-varying spatial convolution module is used to perform convolution processing on the target data to obtain a dynamic adjacency matrix;

[0181] Constructing an attention vector between patent nodes, and based on the attention vector, normalizing the dynamic adjacency matrix to obtain standardized data, where the patent nodes include the first node and the second node;

[0182] Determine the time-varying graph convolution nodes based on standardized data and stable adjacency matrix;

[0183] For time-varying graph convolution nodes, a multi-scale temporal convolution module is used to concatenate the outputs of multiple layers in an extended causal convolution to extract spatiotemporal features.

[0184] In an exemplary embodiment of the present application, the patent information processing device is configured as follows:

[0185] Based on the patent network information graph with time series information, determine the information network centrality of the target patent corresponding to the time series information, where the information network centrality is closeness centrality, betweenness centrality or eigenvector centrality;

[0186] Based on the information network centrality corresponding to the time series information, determine the importance of the target patent and the time information corresponding to the importance;

[0187] Based on the importance of the target patent and time information corresponding to the importance, target patent evaluation information is determined.

[0188] In an exemplary embodiment of the present application, the patent information processing device is configured as follows:

[0189] Based on the importance of the target patent and the time information corresponding to the importance, a time-varying importance curve of the target patent is constructed, and the time-varying importance curve of the target patent is used as the evaluation information of the target patent.

[0190] In an exemplary embodiment of the present application, the patent information processing device is configured as follows:

[0191] Use word embedding model to encode the target patent content information to obtain a high-dimensional dense vector;

[0192] Encoding the target patent description information to obtain a first sparse feature vector;

[0193] A first node is constructed based on the high-dimensional dense vector and the first sparse feature vector.

[0194] In an exemplary embodiment of the present application, the patent information processing device is configured as follows:

[0195] Encode the target patent citation relationship to obtain a second sparse feature vector;

[0196] Generate target node features including a high-dimensional dense vector, a first sparse feature vector, and a second sparse feature vector;

[0197] Processing target node features and preset node features using a deep layer of the deep network to obtain first output information;

[0198] Using the cross layer of the cross network to process the target node features and the preset node features to obtain second output information;

[0199] Splicing the first output information and the second output information to obtain spliced ​​information;

[0200] Determine the similarity between the target patent and the reference patent based on the splicing information;

[0201] If the similarity meets the preset similarity condition, the recommended information corresponding to the reference patent is determined.

[0202] Reference below Figure 10 The electronic device 60 according to the embodiment of the present application is described. Figure 10 The electronic device 60 shown is only an example and should not limit the functions and scope of use of the embodiments of the present application.

[0203] like Figure 10 As shown, electronic device 60 is implemented as a general-purpose computing device. Components of electronic device 60 may include, but are not limited to, at least one processing unit 610, at least one storage unit 620, and a bus 630 connecting various system components (including storage unit 620 and processing unit 610).

[0204] The storage unit stores program codes, which can be executed by the processing unit 610, so that the processing unit 610 performs the steps according to various exemplary embodiments of the present invention described in the description of the exemplary method described above. For example, the processing unit 610 can perform the following steps: Figure 1 The steps shown in .

[0205] The storage unit 620 may include a readable medium in the form of a volatile storage unit, such as a random access memory unit (RAM) 6201 and / or a cache memory unit 6202 , and may further include a read-only memory unit (ROM) 6203 .

[0206] The storage unit 620 may also include a program / utility 6204 having a set (at least one) of program modules 6205, such program modules 6205 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0207] Bus 630 may represent one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

[0208] The electronic device 60 can also communicate with one or more external devices 700 (e.g., a keyboard, a pointing device, a Bluetooth device, etc.), one or more devices that enable a user to interact with the electronic device 60, and / or any device that enables the electronic device 60 to communicate with one or more other computing devices (e.g., a router, a modem, etc.). This communication can occur via an input / output (I / O) interface 650. The I / O interface 650 is connected to the display unit 640. Furthermore, the electronic device 60 can communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network such as the Internet) via a network adapter 660. As shown, the network adapter 660 communicates with other modules of the electronic device 60 via the bus 660. It should be understood that, although not shown in the figure, other hardware and / or software modules can be used in conjunction with the electronic device 60, including but not limited to microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0209] Through the description of the above embodiments, it is easy for those skilled in the art to understand that the example embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solution according to the embodiments of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which can be a personal computer, a server, a terminal device, or a network device, etc.) to execute the method according to the embodiments of the present application.

[0210] In an exemplary embodiment of the present application, a computer-readable storage medium is further provided, on which computer-readable instructions are stored. When the computer-readable instructions are executed by a processor of a computer, the computer is caused to execute the method described in the above method embodiment.

[0211] According to one embodiment of the present application, a program product for implementing the method in the above method embodiment is also provided. The program product may be a portable compact disc read-only memory (CD-ROM) and includes program code, and can be run on a terminal device, such as a personal computer. However, the program product of the present invention is not limited thereto. In this document, a readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0212] The program product may employ any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or component, or any combination thereof. More specific examples (a non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0213] A computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable signal medium may also be any readable medium other than a readable storage medium that can transmit, propagate, or transfer a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0214] The program code embodied on the readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0215] The program code for performing the operations of the present invention may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as JAVA, C++, and the like, as well as conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a stand-alone software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0216] It should be noted that, although several modules or units of the device for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiment of the application, the features and functions of two or more modules or units described above can be concretized in one module or unit. On the contrary, the features and functions of one module or unit described above can be further divided into multiple modules or units to be concretized.

[0217] Furthermore, although the steps of the method of the present application are described in a particular order in the accompanying drawings, this does not require or imply that the steps must be performed in this particular order, or that all steps shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps.

[0218] Through the description of the above embodiments, it is easy for those skilled in the art to understand that the example embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solution according to the embodiments of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which can be a personal computer, a server, a mobile terminal, or a network device, etc.) to execute the method according to the embodiments of the present application.

[0219] Those skilled in the art will readily appreciate other embodiments of the present invention after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present invention that follow the general principles of the present invention and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered merely as exemplary, and the true scope and spirit of the present application are indicated by the appended claims.

Claims

1. A patent information processing method, characterized in that: The method comprises: Obtain target patent content information and target patent description information of the target patent, and reference patent content information and reference patent description information of reference patents that have a citation relationship with the target patent; A first node is constructed using the target patent content information and the target patent description information, a second node is constructed using the reference patent content information and the reference patent description information, and a node relationship corresponding to a preset time point is constructed using a patent citation relationship corresponding to a preset time point, wherein the patent citation relationship is used to describe the citation relationship between the target patent and the reference patent, and the node relationship is used to describe the relationship between the first node and the second node; Based on the first node, the second node, and the node relationship corresponding to the preset time point, a patent network information graph with time sequence information is constructed, wherein the node relationship is used to describe the citation relationship between the target patent corresponding to the first node and the second node and the reference patent. The patent network information graph with time sequence information is used to describe the citation relationship between the target patent corresponding to the first node and the second node and the reference patent at the preset time point; Determine, based on the patent network information graph having time series information, the information network centrality of the target patent corresponding to the time series information, where the information network centrality is closeness centrality, betweenness centrality, or eigenvector centrality; Determining the importance of the target patent and time information corresponding to the importance based on the information network centrality corresponding to the time series information; The target patent evaluation information is determined based on the importance of the target patent and time information corresponding to the importance.

2. The method according to claim 1, wherein Based on the first node, the second node, and the node relationship corresponding to the preset time point, a patent network information graph with time sequence information is constructed, including: Using a linear layer of a preset time-varying graph neural network model to perform data conversion on the first node, the second node, and the node relationship corresponding to the preset time point to obtain target data; Using a time-varying spatial convolution module and a multi-scale temporal convolution module of a preset time-varying graph neural network model to alternately extract the spatiotemporal features of the target data; The readout module of the preset time-varying graph neural network model is used to process the spatiotemporal features to construct a patent network information graph with time series information.

3. The method according to claim 2, wherein The time-varying spatial convolution module and the multi-scale temporal convolution module of the preset time-varying graph neural network are used to alternately extract the spatiotemporal features of the target data, including: Using the time-varying spatial convolution module to learn the embedded features of the target data to obtain a stable adjacency matrix; Using the time-varying spatial convolution module to perform convolution processing on the target data to obtain a dynamic adjacency matrix; Constructing an attention vector between patent nodes, and based on the attention vector, normalizing the dynamic adjacency matrix to obtain normalized data, wherein the patent nodes include the first node and the second node; Determining a time-varying graph convolution node based on the standardized data and the stable adjacency matrix; For the time-varying graph convolution node, the multi-scale temporal convolution module is used to concatenate the outputs of multiple layers in an extended causal convolution to extract the spatiotemporal features.

4. The method according to claim 1, wherein Determining target patent evaluation information based on the importance of the target patent and time information corresponding to the importance includes: Based on the importance of the target patent and the time information corresponding to the importance, a time-varying importance curve of the target patent is constructed, and the time-varying importance curve of the target patent is used as the target patent evaluation information.

5. The method according to claim 1, wherein Constructing a first node with the target patent content information and the target patent description information includes: Using a word embedding model to encode the target patent content information to obtain a high-dimensional dense vector; Encoding the target patent description information to obtain a first sparse feature vector; The first node is constructed based on the high-dimensional dense vector and the first sparse feature vector.

6. The method according to claim 5, wherein After encoding the target patent description information to obtain a first sparse feature vector, the method further includes: Encoding the target patent citation relationship to obtain a second sparse feature vector; Generate a target node feature including the high-dimensional dense vector, the first sparse feature vector, and the second sparse feature vector; Processing the target node feature and the preset node feature using a deep layer of the deep network to obtain first output information; Processing the target node feature and the preset node feature using a cross layer of the cross network to obtain second output information; Splicing the first output information and the second output information to obtain spliced ​​information; Determining the similarity between the target patent and the reference patent based on the splicing information; If the similarity meets the preset similarity condition, the recommendation information corresponding to the reference patent is determined.

7. A patented information processing device, characterized in that: The patent information processing device includes: An acquisition module, configured to acquire target patent content information and target patent description information of a target patent, and reference patent content information and reference patent description information of reference patents that have a citation relationship with the target patent; a first determining module configured to construct a first node using the target patent content information and the target patent description information, construct a second node using the reference patent content information and the reference patent description information, and construct a node relationship corresponding to a preset time point using a patent citation relationship corresponding to a preset time point, wherein the patent citation relationship is used to describe the citation relationship between the target patent and the reference patent, and the node relationship is used to describe the relationship between the first node and the second node; A construction module for constructing a patent network information graph with time sequence information based on the first node, the second node, and the node relationship corresponding to the preset time point, wherein the node relationship is used to describe the citation relationship between the target patent corresponding to the first node and the second node and the reference patent, and the patent network information graph with time sequence information is used to describe the citation relationship between the target patent corresponding to the first node and the second node and the reference patent at the preset time point; The second determination module is used to determine the information network centrality of the target patent corresponding to the time series information based on the patent network information graph with time series information, where the information network centrality is closeness centrality, betweenness centrality or eigenvector centrality; determine the importance of the target patent and the time information corresponding to the importance based on the information network centrality corresponding to the time series information; and determine the target patent evaluation information based on the importance of the target patent and the time information corresponding to the importance.

8. An electronic device, characterized in that: include: one or more processors; A storage device for storing one or more programs, which, when executed by the one or more processors, enables the electronic device to implement the method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that Computer-readable instructions are stored thereon, and when the computer-readable instructions are executed by a processor of a computer, the computer is caused to execute the method according to any one of claims 1 to 6.

Images