Method, system and equipment for automatically generating activity network diagram

Abstract

The invention provides an active network diagram automatic generation method, system and device, and relates to the technical field of network diagram automatic generation. The method comprises the following steps: firstly, obtaining to-be-processed project data; secondly, performing feature mining on the to-be-processed project data by utilizing a semantic feature mining model included in the target analysis network, and outputting target project semantic features corresponding to the to-be-processed project data; then, analyzing semantic features of the target project by utilizing a semantic feature analysis model included in the target analysis network, and outputting a plurality of monitoring activities and a relationship among the plurality of monitoring activities; and finally, based on the plurality of monitoring activities and the relationship among the plurality of monitoring activities, generating a corresponding target activity network graph. Based on the above content, the problem of low reliability of active network graph generation in the prior art can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of network diagram automatic generation, in particular to an activity network diagram automatic generation method, system and device. BACKGROUND

[0002] With the development of project management and business process management, activity network diagram as an important visualization tool plays an indispensable role in project planning, progress tracking and risk assessment. Activity network diagram makes project planning more intuitive and easy to understand by clearly defining the start and end time of each task and the dependency relationship between tasks. This helps project managers and team members quickly grasp the overall progress and critical path of the project, so as to arrange the work order and schedule more effectively; with the help of activity network diagram, project team can easily monitor the difference between actual progress and planned progress. By discovering deviations in time, the team can quickly take measures to adjust to ensure the project is completed on schedule; activity network diagram can also help identify potential risk points in the early stages of the project. For example, by identifying tasks on the critical path, the project team can better assess which task delays will have the greatest impact on the entire project, and then take appropriate preventive measures. It not only clearly shows the various components of the project and their logical relationships with each other, but also effectively helps project teams allocate resources, predict risks and make decisions.

[0003] Traditional manual drawing of activity network diagram is not only time-consuming and laborious, but also prone to errors, especially when dealing with complex projects, that is, there is a relatively low reliability problem. SUMMARY

[0004] Therefore, the purpose of the present application is to provide an activity network diagram automatic generation method, system and device to improve the low reliability problem of activity network diagram generation in the prior art.

[0005] To achieve the above purpose, the technical scheme adopted by the present application is as follows: An activity network diagram automatic generation method, comprising: Obtaining to-be-processed project data, wherein the to-be-processed project data is used at least to describe the activity content of each monitoring activity in a plurality of monitoring activities and the relationship between monitoring activities; Using a semantic feature mining model included in a target analysis network to perform feature mining on the to-be-processed project data, and outputting a target project semantic feature corresponding to the to-be-processed project data, wherein the target project semantic feature is used to reflect semantic information of the plurality of monitoring activities and the relationship between the plurality of monitoring activities, and the target analysis network belongs to a trained neural network model; analyzing the semantic features of the target item by using a semantic feature analysis model included in the target analysis network, and outputting the plurality of monitoring activities and the relationship between the plurality of monitoring activities; generating a target activity network graph corresponding to the to-be-processed item data based on the plurality of monitoring activities and the relationship between the plurality of monitoring activities.

[0006] In a preferred selection of the present application, in the above-mentioned activity network graph automatic generation method, the step of performing feature mining on the to-be-processed item data by using a semantic feature mining model included in the target analysis network and outputting the target item semantic features corresponding to the to-be-processed item data comprises: segmenting the to-be-processed item data to form a plurality of item sub-data corresponding to the to-be-processed item data, and mapping each of the item sub-data to a feature space by using a feature space mapping unit in the semantic feature mining model included in the target analysis network to form item sub-data features corresponding to each of the item sub-data; performing association processing on the item data features corresponding to each of the item sub-data based on the association relationship between the plurality of item sub-data to form an initial feature distribution graph; updating the initial feature distribution graph by using a feature updating unit in the semantic feature mining model to form a target feature distribution graph; determining the target item semantic features corresponding to the to-be-processed item data based on the updated item data features in the target feature distribution graph.

[0007] In a preferred selection of the present application, in the above-mentioned activity network graph automatic generation method, the step of performing association processing on the item data features corresponding to each of the item sub-data based on the association relationship between the plurality of item sub-data to form an initial feature distribution graph comprises: determining, for each of the plurality of item sub-data, whether the two item sub-data have an adjacency relationship in the to-be-processed item data; determining two item sub-data having an adjacency relationship as having an association relationship; for each of the two item sub-data not having an adjacency relationship, determining whether the two item sub-data have an association relationship based on the degree of coincidence between keywords in the two item sub-data, wherein the keywords in the item sub-data refer to words appearing more than a preset number of times in the corresponding item sub-data and not appearing more than the preset number of times in each item sub-data; performing connection processing on the item data features corresponding to each of the two item sub-data having an association relationship to form an initial feature distribution graph.

[0008] In the preferable selection of the present application, in the above-mentioned activity network graph automatic generation method, the step of updating the initial feature distribution graph by using the feature updating unit in the semantic feature mining model to form a target feature distribution graph comprises: For each item data feature in the initial feature distribution graph, focus feature mining processing is performed on the item data feature to form a focused item data feature corresponding to the item data feature, and the item data feature is updated to the corresponding focused item data feature; For each focused item data feature in the initial feature distribution graph, a local feature distribution graph centered on the focused item data feature is determined according to the connection relationship in the initial feature distribution graph, wherein the focused item data feature at the center and each focused item data feature not at the center have a connection relationship in the local feature distribution graph; The focused item data feature at the center in the local feature distribution graph is updated to form an updated item data feature; Based on each of the updated item data features, a target feature distribution graph is determined.

[0009] In the preferable selection of the present application, in the above-mentioned activity network graph automatic generation method, the step of updating the focused item data feature at the center in the local feature distribution graph to form an updated item data feature comprises: When the focused item data features not at the center included in the local feature distribution graph are multiple, the feature matching degree between the focused item data feature at the center and each focused item data feature not at the center is determined in the local feature distribution graph; According to the multiple focused item data features not at the center and the feature matching degrees, the focused item data feature at the center is updated in multiple levels to form an updated item data feature.

[0010] In the preferable selection of the present application, in the above-mentioned activity network graph automatic generation method, the step of updating the focused item data feature at the center in the local feature distribution graph to form an updated item data feature comprises: According to the size relationship between the corresponding feature matching degrees, the multiple focused item data features not at the center are iterated in sequence to form a focused item data feature currently iterated, wherein the feature matching degree corresponding to the first iterated focused item data feature has a maximum value; when the currently traversed focus item data feature belongs to the first traversed focus item data feature, performing focus feature mining processing on the focus item data feature belonging to the center based on the currently traversed focus item data feature, and outputting the corresponding center focus item feature; when the currently traversed focus item data feature does not belong to the first traversed focus item data feature, performing n-1 times of pooling processing on the currently traversed focus item data feature to form a corresponding first pooled item feature, performing 1 time of pooling processing on the center focus item feature corresponding to the previously traversed focus item data feature to form a corresponding second pooled item feature, and performing focus feature mining processing on the second pooled item feature based on the first pooled item feature to output the corresponding center focus item feature, wherein n is the number of current traversals; after obtaining the center focus item feature corresponding to the last traversed focus item data feature, determining the corresponding updated item data feature based on the center focus item feature.

[0011] In a preferred selection of the present application, in the above-mentioned activity network graph automatic generation method, the step of determining the target project semantic feature corresponding to the to-be-processed project data based on the updated item data feature in the target feature distribution graph comprises: splicing each updated item data feature in the target feature distribution graph according to the order of the corresponding project sub-data in the to-be-processed project data to form a project splicing feature corresponding to the target feature distribution graph; performing convolution processing on the project splicing feature to form a project convolution feature; performing pooling processing on the project convolution feature to form the target project semantic feature corresponding to the to-be-processed project data.

[0012] In a preferred selection of the present application, in the above-mentioned activity network graph automatic generation method, before the step of performing feature mining on the to-be-processed project data by using the semantic feature mining model included in the target analysis network to output the target project semantic feature corresponding to the to-be-processed project data, the activity network graph automatic generation method further comprises a step of training the target analysis network, which comprises: obtaining training project data and a training project label corresponding to the training project data, wherein the training project data is used at least to describe the activity content of each training monitoring activity in a plurality of training monitoring activities and the relationship between the training monitoring activities; Using the semantic feature mining model included in the candidate analysis network, feature mining is performed on the training project data, and the semantic features of the training project corresponding to the training project data are output. The semantic features of the training project are used to reflect the semantic information of the multiple training monitoring activities and the relationship between the multiple training monitoring activities. Using the semantic feature analysis model included in the candidate analysis network, the semantic features of the training items are analyzed, the multiple training monitoring activities and the relationship between the multiple training monitoring activities are output, and the network parameters of the candidate analysis network are updated based on the error between the multiple training monitoring activities and the relationship between the multiple training monitoring activities and the training item labels, thus forming the target analysis network.

[0013] This application also provides an automatic activity network diagram generation system, including: The project data acquisition module is used to acquire project data to be processed, wherein the project data to be processed is used at least to describe the activity content of each of the multiple monitoring activities and the relationship between the monitoring activities; The project feature mining module is used to perform feature mining on the project data to be processed using the semantic feature mining model included in the target analysis network, and output the target project semantic features corresponding to the project data to be processed. The target project semantic features are used to reflect the semantic information of the multiple monitoring activities and the relationship between the multiple monitoring activities. The target analysis network is a trained neural network model. The project feature analysis module is used to analyze the semantic features of the target project using the semantic feature analysis model included in the target analysis network, and output the multiple monitoring activities and the relationship between the multiple monitoring activities. The network graph generation module is used to generate a target activity network graph corresponding to the project data to be processed, based on the multiple monitoring activities and the relationships between the multiple monitoring activities.

[0014] Based on the above, this application also provides an electronic device, including: Memory, used to store computer programs; A processor connected to the memory is used to execute the computer program stored in the memory to implement the above-described method for automatically generating active network graphs.

[0015] The method, system, and device for automatically generating activity network diagrams provided in this application first acquire the project data to be processed; second, using the semantic feature mining model included in the target analysis network, feature mining is performed on the project data to be processed, outputting the target project semantic features corresponding to the project data to be processed; then, using the semantic feature analysis model included in the target analysis network, the semantic features of the target project are analyzed, outputting multiple monitoring activities and the relationships between multiple monitoring activities; finally, based on the multiple monitoring activities and the relationships between multiple monitoring activities, the corresponding target activity network diagram is generated. Based on the above, since a trained neural network model can be used, multiple monitoring activities and the relationships between multiple monitoring activities can be reliably analyzed from the project data to be processed, thereby ensuring the reliability of the network diagram generated accordingly. Therefore, compared with the existing technology of manually drawing network diagrams, both efficiency and accuracy can be improved, thus addressing the problem of low reliability in the generation of activity network diagrams in the prior art. Attached Figure Description

[0016] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings.

[0017] Figure 1 A structural block diagram of an electronic device provided in an embodiment of this application.

[0018] Figure 2 This is a flowchart illustrating the method for automatically generating activity network graphs provided in an embodiment of this application.

[0019] Figure 3 This is a schematic diagram of the target activity network diagram provided in the embodiments of this application.

[0020] Figure 4 This is a schematic diagram of the initial feature distribution map provided in the embodiments of this application.

[0021] Figure 5 A block diagram illustrating the automatic generation system of activity network diagrams provided in this application embodiment. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0023] Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0024] like Figure 1 As shown in the illustration, this application provides an electronic device. The electronic device may include a memory, a processor, and an automatic active network graph generation system.

[0025] Specifically, the memory and the processor are electrically connected directly or indirectly to enable data transmission or interaction. For example, the memory and the processor can be electrically connected via one or more communication buses or signal lines. The automatic active network graph generation system includes at least one software functional module stored in the memory in the form of software or firmware. The processor is used to execute executable computer programs stored in the memory, such as the software functional modules and computer programs included in the automatic active network graph generation system, to implement the automatic active network graph generation method provided in this application embodiment.

[0026] Optionally, the memory may be, but is not limited to, random access memory (RAM), read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), etc.

[0027] Furthermore, the processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), a system on chip (SoC), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0028] Understandable. Figure 1The structure shown is for illustrative purposes only; the electronic device may also include components that are more advanced than those shown. Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown may include, for example, a communication unit for exchanging information with other devices.

[0029] Combination Figure 2 This application also provides a method for automatically generating an active network graph applicable to the aforementioned electronic device. The method steps defined in the process of the automatic generation method for the active network graph can be implemented by the electronic device. The following will describe... Figure 2 The specific process shown will be explained in detail.

[0030] Step S110: Obtain the data of the project to be processed.

[0031] In this embodiment, the electronic device can acquire project data to be processed. This project data is used at least to describe the content of each monitoring activity and the relationship between the monitoring activities. For example, the project data to be processed can be the acquired raw project data, or it can be project data formed by preprocessing the raw project data. Preprocessing of the raw project data includes, but is not limited to, data cleaning, anomaly detection, and missing value imputation, etc., which are not specifically limited here.

[0032] Step S120: Using the semantic feature mining model included in the target analysis network, feature mining is performed on the project data to be processed, and the target project semantic features corresponding to the project data to be processed are output.

[0033] In this embodiment, after acquiring the project data to be processed, the electronic device can utilize the semantic feature mining model included in the target analysis network to perform feature mining on the project data to be processed, and output the target project semantic features corresponding to the project data to be processed. The target project semantic features are used to reflect the semantic information of the multiple monitoring activities and the relationships between them; that is, by mining the potential semantic information and representing it in vector form, the target analysis network is a trained neural network model.

[0034] Step S130: Utilize the semantic feature analysis model included in the target analysis network to analyze the semantic features of the target project and output the multiple monitoring activities and the relationship between the multiple monitoring activities.

[0035] In this embodiment, after outputting the semantic features of the target item, the electronic device can utilize the semantic feature analysis model included in the target analysis network to analyze the semantic features of the target item and output the multiple monitoring activities and the relationships between them. For example, the semantic feature analysis model can be a decoder, thus generating corresponding text data that describes the multiple monitoring activities and the relationships between them. For example, in the first step, the decoder outputs the probability parameter distribution of each word in the dictionary based on the semantic features of the target item, and then selects the word with the largest probability parameter as the first word in the text data. In the second step, the embedding features of the first word and the semantic features of the target item are subjected to cross-attention processing. Then, the decoder outputs the probability parameter distribution of each word in the dictionary based on the result of the cross-attention processing, thereby determining the second word in the text data. In this way, each word in the text data can be output.

[0036] Step S140: Based on the multiple monitoring activities and the relationship between the multiple monitoring activities, generate a target activity network diagram corresponding to the project data to be processed.

[0037] In this embodiment of the application, after outputting the plurality of monitoring activities and the relationships between the plurality of monitoring activities, the electronic device can generate a target activity network diagram (e.g., based on the plurality of monitoring activities and the relationships between the plurality of monitoring activities) corresponding to the project data to be processed. Figure 3 (As shown). For example, a target activity network graph corresponding to the project data to be processed can be generated based on the multiple monitoring activities and the relationships between the multiple monitoring activities using graph layout algorithms (such as force-directed layout, hierarchical layout, etc.). The specific process can be referred to relevant existing technologies.

[0038] Based on the above, since a trained neural network model can be used, it is possible to reliably analyze multiple monitoring activities and the relationships between them from the data of the project to be processed, thereby ensuring the reliability of the network diagram generated. Therefore, compared with the existing technology of manually drawing network diagrams, both efficiency and accuracy can be improved, thus addressing the problem of low reliability in the generation of activity network diagrams in the existing technology.

[0039] It should be noted that for step S120, the specific method of feature mining on the project data to be processed is not limited and can be selected according to actual needs.

[0040] For example, in an alternative implementation, in order to improve the reliability of feature mining and make the semantic representation accuracy of the mined target item semantic features higher, the above step S120 may further include steps S121, S122, S123 and S124, the specific contents of each step are as follows.

[0041] Step S121: The project data to be processed is segmented to form multiple project sub-data corresponding to the project data to be processed. Then, the feature space mapping unit in the semantic feature mining model included in the target analysis network is used to map each project sub-data to the feature space to form the project sub-data feature corresponding to each project sub-data.

[0042] In this embodiment, the project data to be processed can be segmented to form multiple project sub-data corresponding to the project data to be processed. Then, using the feature space mapping unit in the semantic feature mining model included in the target analysis network, each project sub-data is mapped to a feature space, forming a project sub-data feature corresponding to each project sub-data. For example, segmentation can be based on a preset data volume, such as the data volume of each project sub-data being less than or equal to a preset data volume. Alternatively, segmentation can be based on a preset number, such as the number of project sub-data being equal to a preset number. Furthermore, the feature space mapping unit can refer to a word embedding model; mapping the project sub-data to the feature space can refer to word embedding processing using a word embedding model. The specific processing procedure can refer to relevant existing technologies.

[0043] Step S122: Based on the correlation between the multiple project sub-data, perform correlation processing on the project data features corresponding to each project sub-data to form an initial feature distribution map.

[0044] In this embodiment of the application, after obtaining the project data features, the project data features corresponding to each project sub-data can be associated based on the correlation between the multiple project sub-data to form an initial feature distribution map. In this way, the project data features can be associated, making the reliability of subsequent processing higher.

[0045] Step S123: Use the feature update unit in the semantic feature mining model to update the initial feature distribution map to form the target feature distribution map.

[0046] In this embodiment, after obtaining the initial feature distribution map, the feature update unit in the semantic feature mining model can be used to update the initial feature distribution map to form the target feature distribution map. That is, since the project data features in the initial feature distribution map are local semantic features of the corresponding project sub-data, contextual semantic information may be lost, making it difficult to guarantee the accuracy of semantic representation. Therefore, it is necessary to update the initial feature distribution map to capture the corresponding semantic information.

[0047] Step S124: Based on the updated project data features in the target feature distribution map, determine the target project semantic features corresponding to the project data to be processed.

[0048] In this embodiment of the application, after obtaining the target feature distribution map, the target project semantic features corresponding to the project data to be processed can be determined based on the updated project data features in the target feature distribution map (i.e., the updated features of the project data features in the initial feature distribution map).

[0049] It is understood that in step S122 above, the specific method of associating the project data features corresponding to each of the project sub-data is not limited. For example, in an alternative implementation, in order to effectively associate the project data features with correlation relationships, so that the associated project data features can be fully mined in subsequent processing, step S122 above may further include the following: First, for each pair of project sub-data in the plurality of project sub-data, determine whether the two project sub-data have an adjacency relationship in the project data to be processed. For example, the first project sub-data and the second project sub-data have an adjacency relationship, while the first project sub-data and the third project sub-data do not have an adjacency relationship. Secondly, two adjacent sub-data items can be identified as having an associated relationship; Then, for each pair of item sub-data that are not adjacent, the correlation between the two item sub-data is determined based on the overlap between the keywords in the two item sub-data. Here, keywords in item sub-data refer to words that appear more than a preset number of times in the corresponding item sub-data, but not more than a preset number of times in every item sub-data. The preset number of times can be selected according to actual needs, for example, it can be a value such as 3, 4, 5, 6, 7, etc. In addition, when the overlap between keywords is greater than the preset overlap, a correlation can be determined, and when the overlap between keywords is not greater than the preset overlap, a correlation can be determined, for example, it can be a value such as 0.4, 0.5, 0.6, etc. Finally, the features of the project data corresponding to every two related project sub-data can be concatenated to form an initial feature distribution map, such as... Figure 4 As shown.

[0050] It is understood that the specific method of updating the initial feature distribution map in step S123 above is not limited. For example, in an alternative implementation, in order to make the semantic representation accuracy of the updated item data features more accurate by updating, step S123 above may further include steps S123a, S123b, S123c and S123d, the specific contents of each step are as follows.

[0051] Step S123a: For each item data feature in the initial feature distribution map, perform focused feature mining processing within the item data feature to form a focused item data feature corresponding to the item data feature, and update the item data feature to the corresponding focused item data feature.

[0052] In this embodiment of the application, for each item data feature in the initial feature distribution map, a focused feature mining process is performed inside the item data feature (for example, self-attention processing can be performed on the item data feature) to form a focused item data feature corresponding to the item data feature, and the item data feature is updated to the corresponding focused item data feature.

[0053] Step S123b: For each focused item data feature in the initial feature distribution map, a local feature distribution map centered on the focused item data feature is determined based on the connection relationship in the initial feature distribution map.

[0054] In this embodiment, after forming focused project data features, for each focused project data feature in the initial feature distribution map, a local feature distribution map centered on that focused project data feature is determined based on the connection relationships in the initial feature distribution map. In this local feature distribution map, the focused project data feature belonging to the center has a connection relationship (e.g., a direct connection) with each focused project data feature not belonging to the center, such as... Figure 4 When the project data feature 2 is taken as the center, the corresponding local feature distribution map includes the project data feature 1, the project data feature 2, and the project data feature 3.

[0055] Step S123c: Within the local feature distribution map, the focus project data features belonging to the center are updated to form updated project data features.

[0056] In this embodiment, after determining the local feature distribution map, the focused item data features belonging to the center can be updated within the local feature distribution map to form updated item data features. That is, the focused item data features belonging to the center can be updated based on the focused item data features that do not belong to the center, so that the focused vector data features representing local semantic features can also incorporate relevant semantic features from the context.

[0057] Step S123d: Based on the data features of each of the updated projects, determine the target feature distribution map.

[0058] In this embodiment of the application, after obtaining the updated project data features, a target feature distribution map can be determined based on each of the updated project data features. For example, each focused project data feature in the initial feature distribution map can be replaced with the corresponding updated project data feature to form the target feature distribution map.

[0059] It is understood that in step S123c above, the specific method of updating the data features of the focused project belonging to the center is not limited. For example, in an alternative implementation, in order to achieve deep fusion of the data features of the focused project, step S123c above may further include steps c1 and c2, the specific contents of each step are as follows.

[0060] Step c1: When the local feature distribution map includes multiple non-center focused project data features, within the local feature distribution map, the feature matching degree between the center focused project data feature and each non-center focused project data feature is determined.

[0061] In this embodiment of the application, when there are multiple non-centering focused project data features included in the local feature distribution map (when there is only one non-centering focused project data feature, focused feature mining processing can be performed directly), the feature matching degree (such as cosine similarity) between the centering focused project data feature and each non-centering focused project data feature can be determined within the local feature distribution map.

[0062] Step c2: Based on the feature matching degree of multiple non-center focused project data features, perform multi-level update processing on the feature of the center focused project data features to form updated project data features.

[0063] In this embodiment, after obtaining the feature similarity, the centered focused project data features can be updated in multiple levels based on multiple non-centered focused project data features and the feature matching degree, forming updated project data features. For example, one non-centered focused project data feature corresponds to one level of update processing. Thus, multiple levels of update processing can be performed for multiple non-centered focused project data features. Based on this, high-level, abstract features can be captured during the update process.

[0064] It is understood that in step c2 above, the specific method of performing multi-level update processing on the data features of the focused project belonging to the center is not limited. For example, in an alternative implementation, in order to achieve matching of fused information during the multi-level update process and make the reliability of fused focus relatively higher, step c2 above can further include the following: First, based on the relationship between the corresponding feature matching degrees, multiple non-center focused project data features can be traversed sequentially to form the currently traversed focused project data features. The feature matching degree corresponding to the first traversed focused project data feature has the maximum value. That is, the traversal order is sorted according to the feature matching degree from large to small. Secondly, when the currently traversed focused project data feature belongs to the first traversed focused project data feature, based on the currently traversed focused project data feature, focused feature mining processing can be performed on the focused project data feature belonging to the center, and the corresponding center focused project feature can be output. For example, based on the currently traversed focused project data feature, cross-attention processing can be performed on the focused project data feature belonging to the center, and the feature obtained by cross-attention processing can be summed with the focused project data feature belonging to the center. Furthermore, the feature obtained by summing can be normalized to obtain the corresponding center focused project feature. Then, when the currently traversed focused project data feature does not belong to the first traversed focused project data feature (such as the second traversed focused project data feature, the third traversed focused project data feature, etc.), n-1 pooling processes are performed on the currently traversed focused project data feature. (Since the traversal order is based on the feature matching degree from largest to smallest, the importance of the focused project data feature traversed later is lower. Therefore, pooling can extract important features while losing unimportant features, thus improving the accuracy of further focused feature mining.) This forms the corresponding first pooled project feature. The central focused project feature corresponding to the previously traversed focused project data feature is then pooled once to form the corresponding second pooled project feature. Based on this first pooled project feature, the second pooled project feature is... The system performs focused feature mining on the data features and outputs the corresponding center focused item features, where n refers to the current traversal number. For example, for the second traversed focused item data feature, it can perform one pooling process to obtain the first pooled item feature. Then, it performs pooling on the center focused item feature corresponding to the first traversed focused item data feature to obtain the second pooled item feature. Then, it performs focused feature mining and outputs the center focused item feature corresponding to the second traversed focused item feature. For the third traversed focused item data feature, it can perform two pooling processes to obtain the first pooled item feature. Then, it performs pooling on the center focused item feature corresponding to the second traversed focused item data feature to obtain the second pooled item feature. Then, it performs focused feature mining and outputs the center focused item feature corresponding to the third traversed focused item feature. Finally, after obtaining the center focus project feature corresponding to the last traversed focus project data feature, the corresponding updated project data feature is determined based on the center focus project feature. For example, the center focus project feature corresponding to the last traversed focus project data feature can be directly used as the updated project data feature, or the center focus project features corresponding to each traversed focus project data feature can be concatenated to obtain the updated project data feature.

[0065] It is understood that the specific method for determining the target project semantic features corresponding to the project data to be processed in step S124 above is not limited. For example, in an alternative implementation, in order to mine more advanced and abstract semantic features and thereby improve the semantic representation accuracy of the obtained target project semantic features, step S124 above may include the following: First, each updated item data feature in the target feature distribution map can be spliced ​​together according to the order of the corresponding item sub-data in the item data to be processed, to form the item splicing feature corresponding to the target feature distribution map. Secondly, the project splicing features can be processed by convolution to form project convolution features; Then, the project convolutional features can be pooled to form the target project semantic features corresponding to the project data to be processed.

[0066] Regarding steps S120 and S130 above, it should also be noted that, in order to reliably execute steps S120 and S130 through the target analysis network, the automatic generation method of the active network graph further includes a step of training to form the target analysis network. In an alternative implementation, this step may include the following: First, training project data and corresponding training project tags can be obtained. The training project data is used at least to describe the activity content of each training monitoring activity and the relationship between the training monitoring activities in multiple training monitoring activities. Secondly, the semantic feature mining model included in the candidate analysis network can be used to perform feature mining on the training project data and output the training project semantic features corresponding to the training project data. The training project semantic features are used to reflect the semantic information of the relationship between the multiple training monitoring activities and the multiple training monitoring activities. Then, the semantic feature analysis model included in the candidate analysis network can be used to analyze the semantic features of the training items, output the multiple training monitoring activities and the relationship between the multiple training monitoring activities, and, based on the error between the multiple training monitoring activities and the relationship between the multiple training monitoring activities and the training item labels (such as the mean square error between the corresponding embedded features), the network parameters of the candidate analysis network can be updated (e.g., updated along the direction of reducing error until the error converges), forming the target analysis network. In this way, the target analysis network can fully learn the semantic correspondence between the training item data and the corresponding training item labels, thereby ensuring the reliable execution of steps S120 and S130.

[0067] Combination Figure 5 This application also provides an automatic activity network diagram generation system applicable to the aforementioned electronic devices. The automatic activity network diagram generation system may include a project data acquisition module, a project feature mining module, a project feature analysis module, and a network diagram generation module.

[0068] The project data acquisition module is used to acquire project data to be processed, wherein the project data to be processed is used at least to describe the activity content of each of the multiple monitoring activities and the relationship between the monitoring activities. In this embodiment of the application, the project data acquisition module can be used to execute... Figure 2 For details regarding the project data acquisition module shown in step S110, please refer to the previous description of step S110.

[0069] The project feature mining module is used to perform feature mining on the project data to be processed using the semantic feature mining model included in the target analysis network, and output the target project semantic features corresponding to the project data to be processed. The target project semantic features are used to reflect the semantic information of the multiple monitoring activities and the relationships between the multiple monitoring activities. The target analysis network is a trained neural network model. In this embodiment, the project feature mining module can be used to execute... Figure 2 The relevant content regarding the project feature mining module in step S120 shown can be found in the previous description of step S120.

[0070] The project feature analysis module is used to analyze the semantic features of the target project using the semantic feature analysis model included in the target analysis network, and output the multiple monitoring activities and the relationship between the multiple monitoring activities. In this embodiment of the application, the project feature analysis module can be used to execute... Figure 2 The relevant content regarding the project feature analysis module in step S130 shown can be found in the previous description of step S130.

[0071] The network graph generation module is used to generate a target activity network graph corresponding to the project data to be processed, based on the multiple monitoring activities and the relationships between the multiple monitoring activities. In this embodiment, the network graph generation module can be used to execute... Figure 2 The details of step S140, and the relevant content regarding the network graph generation module, can be found in the preceding description of step S140.

[0072] In this embodiment of the application, corresponding to the above-described method for automatically generating an active network graph applied to the electronic device, a computer-readable storage medium is also provided, which stores a computer program that executes the various steps of the method for automatically generating an active network graph when the computer program is run.

[0073] The steps executed by the aforementioned computer program during runtime will not be described in detail here, but can be found in the explanation of the automatic generation method of the activity network diagram described above.

[0074] In summary, the method, system, and device for automatically generating activity network diagrams provided in this application first acquire the project data to be processed; second, using the semantic feature mining model included in the target analysis network, feature mining is performed on the project data to be processed, outputting the target project semantic features corresponding to the project data to be processed; then, using the semantic feature analysis model included in the target analysis network, the semantic features of the target project are analyzed, outputting multiple monitoring activities and the relationships between multiple monitoring activities; finally, based on the multiple monitoring activities and the relationships between multiple monitoring activities, the corresponding target activity network diagram is generated. Based on the above, since a trained neural network model can be used, multiple monitoring activities and the relationships between multiple monitoring activities can be reliably analyzed from the project data to be processed, thereby ensuring the reliability of the network diagram generated accordingly. Therefore, compared with the existing technology of manually drawing network diagrams, both efficiency and accuracy can be improved, thus addressing the problem of low reliability in the generation of activity network diagrams in the prior art.

[0075] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can also be implemented in other ways. The apparatus and method embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0076] In addition, the functional modules in the various embodiments of this application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0077] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, electronic device, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks. It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. In the absence of further restrictions, an element defined by the phrase "comprising a..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0078] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. An active network graph automatic generation method, characterized in that, The method comprises the following steps: acquiring to-be-processed project data, wherein the to-be-processed project data is used at least for describing activity content of each monitoring activity in a plurality of monitoring activities and a relationship between the monitoring activities; performing feature mining on the to-be-processed project data by using a semantic feature mining model included in a target analysis network, and outputting target project semantic features corresponding to the to-be-processed project data, wherein the target project semantic features are used for reflecting semantic information of the plurality of monitoring activities and the relationship between the plurality of monitoring activities, and the target analysis network belongs to a trained neural network model; analyzing the target project semantic features by using a semantic feature analysis model included in the target analysis network, and outputting the plurality of monitoring activities and the relationship between the plurality of monitoring activities; generating a target activity network graph corresponding to the to-be-processed project data based on the plurality of monitoring activities and the relationship between the plurality of monitoring activities.

2. The method of claim 1, wherein, The step of performing feature mining on the to-be-processed project data by using the semantic feature mining model included in the target analysis network and outputting the target project semantic features corresponding to the to-be-processed project data comprises the following steps: segmenting the to-be-processed project data to form a plurality of project sub-data corresponding to the to-be-processed project data, and respectively mapping each project sub-data to a feature space by using a feature space mapping unit in the semantic feature mining model included in the target analysis network to form project sub-data features corresponding to each project sub-data; performing association processing on the project sub-data features corresponding to each project sub-data based on an association relationship between the plurality of project sub-data to form an initial feature distribution graph; updating the initial feature distribution graph by using a feature updating unit in the semantic feature mining model to form a target feature distribution graph; determining the target project semantic features corresponding to the to-be-processed project data based on updated project data features in the target feature distribution graph.

3. The method of claim 2, wherein, The step of performing association processing on the project sub-data features corresponding to each project sub-data based on an association relationship between the plurality of project sub-data to form an initial feature distribution graph comprises the following steps: for each two project sub-data in the plurality of project sub-data, determining whether the two project sub-data have an adjacency relationship in the to-be-processed project data; determining two project sub-data having an adjacency relationship as having an association relationship; for each two project sub-data not having an adjacency relationship, determining whether the two project sub-data have an association relationship based on a coincidence degree between keywords in the two project sub-data, wherein the keywords in the project sub-data refer to words appearing more than a preset number of times in the corresponding project sub-data and not appearing more than the preset number of times in each project sub-data; performing connection processing on the project sub-data features corresponding to each two project sub-data having an association relationship to form an initial feature distribution graph.

4. The method of claim 2, wherein, The step of updating the initial feature distribution graph by using the feature updating unit in the semantic feature mining model to form a target feature distribution graph comprises the following steps: For each item data feature in the initial feature distribution graph, a focused feature mining process is performed on the item data feature to form a corresponding focused item data feature, and the item data feature is updated to the corresponding focused item data feature; For each focused item data feature in the initial feature distribution graph, a local feature distribution graph centered on the focused item data feature is determined according to the connection relationship in the initial feature distribution graph, wherein the focused item data feature at the center and each focused item data feature not at the center have a connection relationship in the local feature distribution graph; In the local feature distribution graph, the focused item data feature at the center is updated to form an updated item data feature; Based on each of the updated item data features, a target feature distribution graph is determined.

5. The method of claim 4, wherein, The step of updating the focused item data feature at the center in the local feature distribution graph to form an updated item data feature includes: When the focused item data features not at the center included in the local feature distribution graph are multiple, the feature matching degree between the focused item data feature at the center and each focused item data feature not at the center is determined in the local feature distribution graph; According to the multiple focused item data features not at the center and the feature matching degrees, the focused item data feature at the center is updated in multiple levels to form an updated item data feature.

6. The method of claim 5, wherein, The step of updating the focused item data feature at the center in the local feature distribution graph to form an updated item data feature includes: According to the size relationship between the corresponding feature matching degrees, the multiple focused item data features not at the center are iterated in turn to form a currently iterated focused item data feature, wherein the feature matching degree corresponding to the first iterated focused item data feature has a maximum value; When the currently iterated focused item data feature is the first iterated focused item data feature, the focused item data feature at the center is subjected to a focused feature mining process based on the currently iterated focused item data feature, and a corresponding center focused item feature is outputted; When the currently iterated focused item data feature is not the first iterated focused item data feature, the currently iterated focused item data feature is subjected to n-1 times of pooling processing to form a corresponding first pooled item feature, and the center focused item feature corresponding to the previously iterated focused item data feature is subjected to 1 time of pooling processing to form a corresponding second pooled item feature, and the second pooled item feature is subjected to a focused feature mining process based on the first pooled item feature, and a corresponding center focused item feature is outputted, wherein n is the number of iterations; After obtaining the center focused item feature corresponding to the last iterated focused item data feature, a corresponding updated item data feature is determined based on the center focused item feature.

7. The method of claim 2, wherein, The step of determining the target project semantic feature corresponding to the to-be-processed project data based on the updated project data features in the target feature distribution map comprises: Splicing each updated project data feature in the target feature distribution map according to the sequence of the corresponding project sub-data in the to-be-processed project data to form a project splicing feature corresponding to the target feature distribution map; Performing convolution processing on the project splicing feature to form a project convolution feature; Performing pooling processing on the project convolution feature to form the target project semantic feature corresponding to the to-be-processed project data.

8. The method of claim 1-7, wherein, Before the step of performing feature mining on the to-be-processed project data by using a semantic feature mining model included in the target analysis network to output the target project semantic feature corresponding to the to-be-processed project data, the activity network graph automatic generation method further comprises a step of training the target analysis network, which comprises: Obtaining training project data and training project labels corresponding to the training project data, wherein the training project data is used at least to describe the activity content of each training monitoring activity in a plurality of training monitoring activities and the relationship between the training monitoring activities; Performing feature mining on the training project data by using a semantic feature mining model included in a candidate analysis network to output training project semantic features corresponding to the training project data, wherein the training project semantic features are used to reflect semantic information of the plurality of training monitoring activities and the relationship between the plurality of training monitoring activities; Performing analysis on the training project semantic features by using a semantic feature analysis model included in the candidate analysis network to output the plurality of training monitoring activities and the relationship between the plurality of training monitoring activities, and performing network parameter updating operation on the candidate analysis network based on the error between the plurality of training monitoring activities, the relationship between the plurality of training monitoring activities, and the training project labels to form the target analysis network.

9. An active network graph automatic generation system, characterized by, Comprise: A project data acquisition module is configured to obtain to-be-processed project data, wherein the to-be-processed project data is used at least to describe the activity content of each monitoring activity in a plurality of monitoring activities and the relationship between the monitoring activities; A project feature mining module is configured to perform feature mining on the to-be-processed project data by using a semantic feature mining model included in a target analysis network to output target project semantic features corresponding to the to-be-processed project data, wherein the target project semantic features are used to reflect semantic information of the plurality of monitoring activities and the relationship between the plurality of monitoring activities, and the target analysis network is a trained neural network model; A project feature analysis module is configured to perform analysis on the target project semantic features by using a semantic feature analysis model included in the target analysis network to output the plurality of monitoring activities and the relationship between the plurality of monitoring activities; A network graph generation module is configured to generate a target activity network graph corresponding to the to-be-processed project data based on the plurality of monitoring activities and the relationship between the plurality of monitoring activities.

10. An electronic device, comprising: Comprise: a memory for storing a computer program; a processor connected with the memory for executing the computer program stored in the memory to realize the method for automatically generating an activity network diagram according to any one of claims 1-8.

Images