A knowledge base-based network information extension pre-judgment method and system
By constructing a regulatory map and guiding computational model, combined with sensitivity assessment, the problem of insufficient prediction of online information dissemination was solved, enabling precise supervision of abnormal information spread and improving the quality of the information dissemination environment on social media.
Patent Information
- Application Number
- CN202611190235.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-06
- Publication Date
- 2026-11-10
AI Technical Summary
Existing technologies lack effective prediction of the spread of online information, resulting in poor social media information regulation and difficulty in dealing with the abnormal spread of misinformation, hate speech and extremist ideologies.
A regulatory graph is constructed based on a knowledge base. Multiple guidance characteristic values are calculated through a guidance calculation model to predict the abnormal spread of forwarding nodes. The sensitivity calculation model is combined with the sensitivity of text data to evaluate the sensitivity of text data and adjust the abnormal label level of forwarding nodes to improve the regulatory effect.
It has enabled accurate prediction of abnormal information dissemination, improved the effectiveness of social media information supervision, and ensured a clean and stable cyberspace.
Smart Images

Figure CN123069059A_ABST
Abstract
Description
Technical Field
[0001] This application belongs to the field of information dissemination technology, and in particular relates to a method and system for predicting the extension of network information based on a knowledge base. Background Technology
[0002] The importance of regulating the forwarding of information on social media lies in its direct relation to the healthy operation of public safety, social order, and individual rights. Currently, social media has become a core channel for information dissemination. Its "viral" dissemination characteristics allow a single message to reach millions of users within minutes. Without effective constraints, this efficiency makes it highly susceptible to exploitation by negative content such as misinformation, hate speech, and extremist ideologies.
[0003] In the digital age, social media has become deeply integrated into social governance, and the level of regulation in its information dissemination directly determines the cleanliness of cyberspace. Only by building a "traceable, verifiable, and accountable" communication environment through scientific regulation can we safeguard freedom of speech while upholding the bottom line of social stability and public interest, allowing social media to truly become a public platform for disseminating knowledge and promoting progress.
[0004] However, existing technologies lack effective prediction of the spread of online information, resulting in poor regulatory effectiveness. Summary of the Invention
[0005] This application provides a knowledge base-based method for predicting the spread of network information. This solution constructs a regulatory graph based on the knowledge base, predicts the forwarding nodes of forwarded content through the regulatory graph, and calculates multiple guidance characteristic values through a preset guidance calculation model to predict abnormal information spread and improve the regulatory effect.
[0006] In a first aspect, embodiments of this application provide a method for predicting network information extensions based on a knowledge base, the method comprising the following steps: Based on the number of times each user is forwarded in the knowledge base, forwarded nodes are constructed; based on the forwarding status of the user corresponding to each forwarded node, forwarding nodes for that user are constructed, and edges between forwarded nodes and forwarding nodes are constructed to obtain the regulatory graph. The historical forwarding data of each forwarding node is classified, and a forwarding tag is constructed for each forwarding node, wherein the forwarding tag includes multiple forwarding categories; Each forwarded node in the monitoring graph is monitored. When the forwarded node sends text data, the text data is classified. The forwarding node is matched based on the classification of the data. The forwarding node is monitored in advance based on the matching. When the forwarding node under pre-monitoring forwards text data, it obtains the supplementary text added when forwarding the text data and inputs the supplementary text into a preset guidance calculation model. The guidance calculation model is set with a preset number of calculation channels, each calculation channel corresponds to a guidance characteristic, and each calculation channel is sequentially set with an attention layer and a normalization layer. The guidance weight value is calculated based on multiple guidance characteristic values output by the guidance calculation model, and the abnormal diffusion value is calculated based on the guidance weight value.
[0007] The above scheme first determines an initial regulatory map based on the content collected in the knowledge base. This map includes forwarding nodes and forwarded nodes. By monitoring forwarding nodes with a large number of forwarded messages and matching the forwarding tags of forwarding nodes with the categories of the actual sent content, the forwarding nodes of the forwarded content can be predicted. In the actual forwarding monitoring process, multiple guidance characteristic values are calculated through a preset guidance calculation model. By comprehensively calculating the abnormal diffusion value through multiple guidance characteristics, this scheme can further predict the abnormal diffusion of information and improve the monitoring effect.
[0008] In some embodiments of the present invention, the guidance characteristics include emotional attachment, cognitive reinforcement, interactive guidance, and contextual adaptation. The guidance calculation model is provided with four calculation channels, which correspond to emotional attachment, cognitive reinforcement, interactive guidance, and contextual adaptation, respectively. Each calculation channel is further provided with a prediction layer after the normalization layer. The prediction layer is provided with a fully connected layer, a ReLU activation function layer, and a sigmoid activation function layer in sequence, and the guidance characteristic value is output through the sigmoid activation function layer.
[0009] In some embodiments of the present invention, in the step of calculating the guidance weight value based on multiple guidance characteristic values output by the guidance calculation model, the guidance weight value is calculated using the following formula:
[0010] in, Indicates the guiding weight value. , , and These represent the guidance characteristic values corresponding to emotional attachment, cognitive reinforcement, interactive guidance, and contextual adaptability, respectively. , , and These represent the calculation weights corresponding to emotional attachment, cognitive reinforcement, interactive guidance, and contextual adaptability, respectively.
[0011] Using the above scheme, this scheme sets up four calculation channels, which correspond to emotional attachment, cognitive reinforcement, interactive guidance, and contextual adaptability, respectively. Among them, emotional attachment is the degree to which the text evokes emotional resonance in readers, cognitive reinforcement is the degree to which the text enhances readers' cognition, interactive guidance is the degree to which the text guides readers to participate in interaction, and contextual adaptability is the degree to which the text matches the current context. By combining the above four points to calculate the guidance weight value, it is possible to calculate the possibility of guiding readers' emotional fluctuations and easily leading to further abnormal spread.
[0012] In some embodiments of the present invention, in the step of calculating the abnormal diffusion value based on the guiding weight value, the text data sent by the forwarded node is input into a preset sensitivity calculation model, which is sequentially configured with an input layer, an LSTM hidden layer, a fully connected layer and an output layer, and outputs a sensitivity value through the output layer; the abnormal diffusion value is calculated based on the sensitivity value and the guiding weight value.
[0013] Using the above scheme, this scheme calculates the sensitivity of the text data sent by the forwarding node through a sensitivity calculation model. For some highly sensitive texts, they are prone to abnormal propagation. This scheme further calculates the abnormal diffusion value by combining the sensitivity value and the guiding weight value. The abnormal diffusion value corresponds to the possibility of abnormal propagation of the text itself and the text added by the forwarder. The abnormal diffusion value can more accurately determine the occurrence of abnormal propagation of the text.
[0014] In some embodiments of the present invention, the method further includes determining whether the forwarding is likely to cause abnormal spread based on the abnormal spread value; if it is determined that it is likely to cause abnormal spread, then modifying the abnormal label of the corresponding forwarding node. During the process of modifying the abnormal label of the corresponding forwarding node, it is determined whether the forwarding node has an abnormal label; If it does not exist, add an exception label and mark the exception level as 1; If it exists, determine whether the anomaly level of the anomaly label has reached the upper limit threshold level. If it has, the anomaly level will not be increased. If it has not, the anomaly level will be increased by one level.
[0015] In some embodiments of the present invention, the LSTM hidden layer of the sensitivity calculation model is provided with an LSTM layer and an attention layer in sequence, the attention layer is provided with multiple attention heads, and the LSTM layer is used to capture long-distance semantic dependencies.
[0016] Using the above scheme, the attention layer with multiple attention heads is used to extract key sentiment words and topic cores in a weighted manner.
[0017] In some embodiments of the present invention, in the steps of calculating a guidance weight value based on multiple guidance characteristic values output by the guidance calculation model and calculating anomaly diffusion value based on the guidance weight value, the anomaly label of the corresponding forwarding node is obtained. If the anomaly level in the anomaly label is greater than the anomaly threshold level, the label weight is calculated based on the current anomaly threshold level, and the anomaly diffusion value is calculated based on the label weight value, the guidance weight value, and the sensitivity value.
[0018] Using the above scheme, this scheme further introduces a tag weight value, which can provide corresponding bonuses to certain network promoter users. When there are frequent abnormal forwarding attempts, the abnormality level of the node increases, thereby increasing the tag weight value and ensuring accurate positioning of the user's forwarding.
[0019] In some embodiments of the present invention, in the step of classifying the historical forwarding data of each forwarding node and constructing the forwarding label of each forwarding node, each piece of historical forwarding data of the forwarding node is classified based on a preset classification model, and the number of data entries in each category is recorded. The name of the forwarding category with a larger number of data entries is used as the forwarding label of the forwarding node.
[0020] In some embodiments of the present invention, when the forwarded node sends text data, the text data is classified, and the forwarding node is matched based on the classification of the data. The text data is input into a preset classification model, the classification model outputs the classification of the text data, and the classification is matched with the forwarding category name in the forwarding tag of the forwarding node. If the match is successful, the forwarding node is taken as the matched forwarding node.
[0021] Another aspect of the present invention relates to a knowledge base-based network information extension prediction system. The system includes a computer device, which includes a processor and a memory. The memory stores computer instructions, and the processor is used to execute the computer instructions stored in the memory. When the computer instructions are executed by the processor, the system implements the steps of the method.
[0022] Additional advantages, objects, and features of the invention will be set forth in part in the description which follows, and will also become apparent in part to those skilled in the art upon studying the text, or may be learned by practice of the invention. The objects and other advantages of the invention will become apparent from the description and the accompanying drawings.
[0023] Those skilled in the art will understand that the objectives and advantages achievable with the present invention are not limited to those specifically described above, and that the above and other objectives achievable with the present invention will become clearer from the following detailed description. Attached Figure Description
[0024] The accompanying drawings are provided to further understand the present disclosure and form part of the specification. They are used together with the following detailed description to explain the present disclosure, but do not constitute a limitation thereof.
[0025] In the attached diagram: Figure 1 This is a schematic diagram of one implementation of the network information extension prediction method based on a knowledge base in this application; Figure 2 This is a schematic diagram of another implementation of the knowledge base-based network information extension prediction method of this application.
[0026] Figure 3 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation
[0027] To better understand the above-mentioned objectives, features, and advantages of this disclosure, the solutions disclosed herein will be further described below. It should be noted that, unless otherwise specified, the embodiments and features described herein can be combined with each other.
[0028] Numerous specific details are set forth in the following description in order to provide a full understanding of this disclosure, but this disclosure may also be implemented in other ways different from those described herein; obviously, the embodiments in the specification are only some, and not all, of the embodiments of this disclosure.
[0029] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the term "comprising" or any other variations thereof is 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. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.
[0030] Generally speaking, existing technologies lack effective prediction of network information dissemination. This solution constructs a regulatory graph based on a knowledge base, predicts the forwarding nodes of forwarded content through the regulatory graph, and calculates multiple guidance characteristic values through a preset guidance calculation model to predict abnormal information spread and improve the regulatory effect.
[0031] Therefore, in order to effectively predict the spread of network information, this application provides a network information extension prediction method based on a knowledge base.
[0032] Figure 1 This is a schematic diagram of a knowledge base-based network information extension prediction method provided in an embodiment of this application.
[0033] Firstly, such as Figure 1 As shown, Figure 1 This application provides a knowledge-based network information extension prediction method, which includes the following steps: A flowchart of the core steps is presented, clearly showing the complete technical chain of 'monitoring graph construction → forwarding tag generation → monitoring of forwarded nodes and matching of forwarding nodes → supplementary text analysis → abnormal diffusion value calculation'. Step S100: Filter based on the number of times each user is forwarded in the knowledge base and construct forwarded nodes; construct forwarding nodes for each user corresponding to each forwarded node, and construct edges between forwarded nodes and forwarding nodes to obtain the regulatory graph. In the specific implementation process, the number of forwarded requests for each user is compared with the preset forwarded request threshold. If it is greater than the preset forwarded request threshold, a forwarded node is constructed accordingly.
[0034] In the specific implementation process, in the step of constructing the forwarding node for each user based on the forwarding status of the user corresponding to each forwarded node, among the users who forward the user corresponding to the forwarded node, users whose forwarding volume is greater than the preset forwarding threshold are constructed as forwarding nodes. The forwarding threshold is set according to the influence of the forwarded node. For example, when the forwarded node is an account with millions of followers, the forwarding threshold is set to 50 times (that is, users who forward the content of the node more than 50 times are listed as forwarding nodes); when the forwarded node is an ordinary user, the forwarding threshold is set to 10 times.
[0035] Step S200: Classify the historical forwarding data of each forwarding node and construct a forwarding tag for each forwarding node, wherein the forwarding tag includes multiple forwarding categories; Step S300: Monitor each forwarded node in the monitoring graph. When the forwarded node sends text data, classify the text data, match the forwarding node based on the classification of the data, and monitor the forwarding node in advance based on the matching. In the specific implementation process, the text data sent by the forwarded node is constructed into a text vector, the text vector is input into a preset classification model, and the classification model outputs the classification of the text. The classification model can be a convolutional neural network model.
[0036] Step S400: When the forwarding node under pre-monitoring forwards text data, obtain the supplementary text added when forwarding the text data, and input the supplementary text into the preset guidance calculation model. The guidance calculation model is set with a preset number of calculation channels, each calculation channel corresponds to a guidance characteristic, and each calculation channel is sequentially set with an attention layer and a normalization layer. In the specific implementation process, the guided calculation model is also equipped with a preprocessing module. The preprocessing module adopts the BERT model to construct the text data into text vectors, and inputs the text vectors into each calculation channel respectively.
[0037] Step S500: Calculate the guidance weight value based on the multiple guidance characteristic values output by the guidance calculation model, and calculate the abnormal diffusion value based on the guidance weight value.
[0038] The above scheme first determines an initial regulatory map based on the content collected in the knowledge base. This map includes forwarding nodes and forwarded nodes. By monitoring forwarding nodes with a large number of forwarded messages and matching the forwarding tags of forwarding nodes with the categories of the actual sent content, the forwarding nodes of the forwarded content can be predicted. In the actual forwarding monitoring process, multiple guidance characteristic values are calculated through a preset guidance calculation model. By comprehensively calculating the abnormal diffusion value through multiple guidance characteristics, this scheme can further predict the abnormal diffusion of information and improve the monitoring effect.
[0039] In some embodiments of the present invention, the guidance characteristics include emotional attachment, cognitive reinforcement, interactive guidance, and contextual adaptation. The guidance calculation model is provided with four calculation channels, which correspond to emotional attachment, cognitive reinforcement, interactive guidance, and contextual adaptation, respectively. Each calculation channel is further provided with a prediction layer after the normalization layer. The prediction layer is provided with a fully connected layer, a ReLU activation function layer, and a sigmoid activation function layer in sequence, and the guidance characteristic value is output through the sigmoid activation function layer.
[0040] In some embodiments of the present invention, in the step of calculating the guidance weight value based on multiple guidance characteristic values output by the guidance calculation model, the guidance weight value is calculated using the following formula:
[0041] in, Indicates the guiding weight value. , , and These represent the guidance characteristic values corresponding to emotional attachment, cognitive reinforcement, interactive guidance, and contextual adaptability, respectively. , , and These represent the computational weights corresponding to emotional attachment, cognitive reinforcement, interactive guidance, and contextual adaptability, respectively. , , and The value range of is 0-1, and satisfies + + + =1, which can be dynamically adjusted according to the priority requirements of the regulatory scenario (such as public safety, corporate reputation). For example, in the public safety scenario, the weight of 'emotional attachment' can be increased. Up to 0.4.
[0042] Using the above scheme, this scheme sets up four calculation channels, corresponding to emotional attachment, cognitive reinforcement, interactive guidance, and contextual adaptability. Emotional attachment refers to the degree to which the text evokes emotional resonance in readers; cognitive reinforcement refers to the degree to which the text enhances readers' cognition; interactive guidance refers to the degree to which the text guides readers to participate in interaction; and contextual adaptability refers to the degree to which the text matches the current context. Combining these four points to calculate the guidance weight value can determine the likelihood of guiding readers' emotional fluctuations and potentially leading to further abnormal dissemination. Emotional attachment refers to the degree to which supplementary text evokes emotional resonance such as anger, anxiety, or identification in readers; cognitive reinforcement refers to the degree to which supplementary text reinforces readers' existing viewpoints or shapes new cognitions; interactive guidance refers to the strength of supplementary text prompts readers to interact through phrases such as "Welcome to comment" and "Forward reminder"; and contextual adaptability refers to the degree to which supplementary text fits the theme of the original text and the current online discussion scenario.
[0043] In some embodiments of the present invention, in the step of calculating the abnormal diffusion value based on the guiding weight value, the text data sent by the forwarded node is input into a preset sensitivity calculation model, which is sequentially configured with an input layer, an LSTM hidden layer, a fully connected layer and an output layer, and outputs a sensitivity value through the output layer; the abnormal diffusion value is calculated based on the sensitivity value and the guiding weight value.
[0044] In some embodiments of the present invention, in the step of calculating the abnormal diffusion value based on the sensitivity value and the guiding weight value, the product of the sensitivity value and the guiding weight value is calculated as the abnormal diffusion value.
[0045] Using the above scheme, this scheme calculates the sensitivity of the text data sent by the forwarding node through a sensitivity calculation model. For some highly sensitive texts, they are prone to abnormal propagation. This scheme further calculates the abnormal diffusion value by combining the sensitivity value and the guiding weight value. The abnormal diffusion value corresponds to the possibility of abnormal propagation of the text itself and the text added by the forwarder. The abnormal diffusion value can more accurately determine the occurrence of abnormal propagation of the text.
[0046] like Figure 2 As shown, Figure 2 The flowchart for the extended implementation of the method adds an 'abnormal label modification' step to the basic process, specifically demonstrating the logic of 'abnormal propagation value determination → abnormal label existence detection → abnormal level adjustment (addition / upgrade / upper limit locking)'; in some embodiments of the present invention, the steps of the method further include step S600, determining whether the forwarding is likely to cause abnormal propagation based on the abnormal propagation value, and if it is determined that it is likely to cause abnormal propagation, then modifying the abnormal label of the corresponding forwarding node.
[0047] During the process of modifying the abnormal label of the corresponding forwarding node, it is determined whether the forwarding node has an abnormal label; If it does not exist, add an exception label and mark the exception level as 1; If it exists, determine whether the anomaly level of the anomaly label has reached the upper limit threshold level. If it has, the anomaly level will not be increased. If it has not, the anomaly level will be increased by one level.
[0048] In some embodiments of the present invention, the LSTM hidden layer of the sensitivity calculation model is provided with an LSTM layer and an attention layer in sequence, the attention layer is provided with multiple attention heads, and the LSTM layer is used to capture long-distance semantic dependencies.
[0049] Using the above scheme, the attention layer with multiple attention heads is used to extract key sentiment words and topic cores in a weighted manner.
[0050] In some embodiments of the present invention, in the steps of calculating a guidance weight value based on multiple guidance characteristic values output by the guidance calculation model and calculating anomaly diffusion value based on the guidance weight value, the anomaly label of the corresponding forwarding node is obtained. If the anomaly level in the anomaly label is greater than the anomaly threshold level, the label weight is calculated based on the current anomaly threshold level, and the anomaly diffusion value is calculated based on the label weight value, the guidance weight value, and the sensitivity value.
[0051] In some embodiments of the present invention, each of the anomaly levels corresponds to a pre-set label weight value. In the step of calculating the anomaly diffusion value based on the label weight value, the guiding weight value, and the sensitivity value, the product of the label weight value, the guiding weight value, and the sensitivity value is calculated as the anomaly diffusion value.
[0052] Using the above scheme, this scheme further introduces a tag weight value, which can provide corresponding bonuses to certain network promoter users. When frequent abnormal forwarding is carried out, the abnormality level of the node increases, thereby increasing the tag weight value, ensuring accurate positioning of the user's forwarding. The preset correspondence between tag weight and abnormality level is as follows: Level 1 abnormality (first abnormality) corresponds to tag weight 1.0, Level 2 abnormality corresponds to 1.2, Level 3 abnormality corresponds to 1.4, and Level 4 abnormality (upper threshold level) corresponds to 1.6. The increment of weight can be adjusted according to the strictness of supervision. If it is necessary to strengthen the control of high-frequency abnormal nodes, the increment can be increased to 0.3.
[0053] In some embodiments of the present invention, in the step of classifying the historical forwarding data of each forwarding node and constructing the forwarding label of each forwarding node, each piece of historical forwarding data of the forwarding node is classified based on a preset classification model, and the number of data entries in each category is recorded. The name of the forwarding category with a larger number of data entries is used as the forwarding label of the forwarding node.
[0054] In the specific implementation process, 1. The classification model can be a convolutional neural network (CNN), a BERT-based model, or a TextCNN model, among which the BERT model is suitable for complex semantic classification scenarios; 2. The threshold for the number of times a message can be forwarded can be set to 100 (this can be adjusted to 500 for platforms with more than 10 million users). 3. Set the preset number of forwarding categories to 3-5. For example, if the number of data entries for 'Social Hot Topics', 'Technology News', and 'People's Livelihood Topics' ranks highest in the historical forwarding of a certain forwarding node, then use these three categories as forwarding tags.
[0055] In some embodiments of the present invention, when the forwarded node sends text data, the text data is classified, and the forwarding node is matched based on the classification of the data. The text data is input into a preset classification model, the classification model outputs the classification of the text data, and the classification is matched with the forwarding category name in the forwarding tag of the forwarding node. If the match is successful, the forwarding node is taken as the matched forwarding node.
[0056] Another aspect of the present invention relates to a knowledge base-based network information extension prediction system. The system includes a computer device, which includes a processor and a memory. The memory stores computer instructions, and the processor is used to execute the computer instructions stored in the memory. When the computer instructions are executed by the processor, the system implements the steps of the method.
[0057] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.
[0058] like Figure 3 As shown, Figure 3 This is a schematic diagram of the electronic device structure corresponding to the system. The processor 1201 executes computer instructions in the memory 1202, the communication interface 1203 realizes data interaction, and the bus 1210 connects various components to support the operation of the method. This application embodiment provides an electronic device, which includes: a processor and a memory storing computer program instructions; when the processor executes the computer program instructions, it implements the above-mentioned knowledge base-based network information extension prediction method.
[0059] The electronic device may include a processor 1201 and a memory 1202 storing computer program instructions.
[0060] Specifically, the processor 1201 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.
[0061] Memory 1202 may include mass storage for data or instructions. For example, and not limitingly, memory 1202 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 1202 may include removable or non-removable (or fixed) media. Where appropriate, memory 1202 may be internal or external to the integrated gateway disaster recovery device. In a particular embodiment, memory 1202 is non-volatile solid-state memory.
[0062] Memory may include read-only memory (ROM), random access memory (RAM), disk storage media devices, optical storage media devices, flash memory devices, and electrical, optical, or other physical / tangible memory storage devices. Therefore, typically, memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the methods according to one aspect of this disclosure.
[0063] The processor 1201 reads and executes computer program instructions stored in the memory 1202 to implement any of the battery thermal runaway parameter determination methods in the above embodiments.
[0064] In one example, the electronic device may also include a communication interface 1203 and a bus 1210. For example, Figure 3 As shown, the processor 1201, memory 1202, and communication interface 1203 are connected through bus 1210 and complete communication with each other.
[0065] The communication interface 1203 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.
[0066] Bus 1210 includes hardware, software, or both, that couples components of an electronic device together. For example, and not limitingly, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 1210 may include one or more buses. Although specific buses are described and illustrated in embodiments of this application, any suitable bus or interconnect is contemplated herein.
[0067] It should be clarified that this application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of this application is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.
[0068] The functional blocks shown in the above-described structural diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. Programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried on a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.
[0069] It should also be noted that the exemplary embodiments mentioned in this application describe methods or systems based on a series of steps or apparatus. However, this application is not limited to the order of the above steps; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.
[0070] The aspects of this disclosure have been described above with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It should be understood that each block in the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that these instructions, executable via the processor of the computer or other programmable data processing apparatus, enable the implementation of the functions / actions specified in one or more blocks of the flowchart illustrations and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can also be implemented by special-purpose hardware performing the specified functions or actions, or can be implemented by a combination of special-purpose hardware and computer instructions.
[0071] The above description is merely a specific implementation of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the protection scope of this application.
Claims
1. A method for predicting the extension of network information based on a knowledge base, characterized in that, The method includes the following steps: Based on the number of times each user is forwarded in the knowledge base, forwarded nodes are constructed; based on the forwarding status of the user corresponding to each forwarded node, forwarding nodes for that user are constructed, and edges between forwarded nodes and forwarding nodes are constructed to obtain the regulatory graph. The historical forwarding data of each forwarding node is classified, and a forwarding tag is constructed for each forwarding node, wherein the forwarding tag includes multiple forwarding categories; Each forwarded node in the monitoring graph is monitored. When the forwarded node sends text data, the text data is classified. The forwarding node is matched based on the classification of the data. The forwarding node is monitored in advance based on the matching. When the forwarding node under pre-monitoring forwards text data, it obtains the supplementary text added when forwarding the text data and inputs the supplementary text into a preset guidance calculation model. The guidance calculation model is set with a preset number of calculation channels, each calculation channel corresponds to a guidance characteristic, and each calculation channel is sequentially set with an attention layer and a normalization layer. The guidance weight value is calculated based on multiple guidance characteristic values output by the guidance calculation model, and the abnormal diffusion value is calculated based on the guidance weight value.
2. The knowledge base-based network information extension prediction method according to claim 1, characterized in that, The guidance characteristics include emotional attachment, cognitive reinforcement, interactive guidance, and contextual adaptation. The guidance calculation model has four calculation channels, which correspond to emotional attachment, cognitive reinforcement, interactive guidance, and contextual adaptation, respectively. Each calculation channel is connected to a prediction layer after the normalization layer. The prediction layer consists of a fully connected layer, a ReLU activation function layer, and a sigmoid activation function layer in sequence. The guidance characteristic value is output through the sigmoid activation function layer.
3. The knowledge base-based network information extension prediction method according to claim 2, characterized in that, In the step of calculating the guidance weight value based on multiple guidance characteristic values output by the guidance calculation model, the guidance weight value is calculated using the following formula: ; in, Indicates the guiding weight value. , , and These represent the guidance characteristic values corresponding to emotional attachment, cognitive reinforcement, interactive guidance, and contextual adaptability, respectively. , , and These represent the calculation weights corresponding to emotional attachment, cognitive reinforcement, interactive guidance, and contextual adaptability, respectively.
4. The knowledge base-based network information extension prediction method according to claim 1, characterized in that, In the step of calculating the abnormal diffusion value based on the guiding weight value, the text data sent by the forwarded node is input into a preset sensitivity calculation model. The sensitivity calculation model is sequentially set with an input layer, an LSTM hidden layer, a fully connected layer and an output layer, and the sensitivity value is output through the output layer. The abnormal diffusion value is calculated based on the sensitivity value and the guiding weight value.
5. The knowledge base-based network information extension prediction method according to any one of claims 1 to 4, characterized in that, The method further includes determining whether the forwarding is likely to cause abnormal spread based on the abnormal spread value; if it is determined that it is likely to cause abnormal spread, then modifying the abnormal label of the corresponding forwarding node. During the process of modifying the abnormal label of the corresponding forwarding node, it is determined whether the forwarding node has an abnormal label; If it does not exist, add an exception label and mark the exception level as 1; If it exists, determine whether the anomaly level of the anomaly label has reached the upper limit threshold level. If it has, the anomaly level will not be increased. If it has not, the anomaly level will be increased by one level.
6. The knowledge base-based network information extension prediction method according to claim 4, characterized in that, The sensitivity calculation model has an LSTM hidden layer and an attention layer set sequentially. The attention layer has multiple attention heads, and the LSTM layer is used to capture long-distance semantic dependencies.
7. The knowledge base-based network information extension prediction method according to claim 1, characterized in that, In the steps of calculating the guidance weight value based on multiple guidance characteristic values output by the guidance calculation model, and calculating the abnormal diffusion value based on the guidance weight value, the abnormal label of the corresponding forwarding node is obtained. If the abnormal level in the abnormal label is greater than the abnormal threshold level, the label weight is calculated based on the current abnormal threshold level, and the abnormal diffusion value is calculated based on the label weight value, the guidance weight value, and the sensitivity value.
8. The knowledge base-based network information extension prediction method according to claim 1, characterized in that, In the step of classifying the historical forwarding data of each forwarding node and constructing the forwarding label of each forwarding node, each piece of historical forwarding data of the forwarding node is classified based on a preset classification model, and the number of data entries in each category is recorded. The name of the forwarding category with the larger number of data entries is used as the forwarding label of the forwarding node.
9. The knowledge base-based network information extension prediction method according to claim 1, characterized in that, In the step of classifying the text data when the forwarded node sends text data and matching the forwarding node based on the classification of the data, the text data is input into a preset classification model, the classification model outputs the classification of the text data, and the classification is matched with the forwarding category name in the forwarding tag of the forwarding node. If the match is successful, the forwarding node is taken as the matched forwarding node.
10. A knowledge-based network information extension prediction system, characterized in that: The system includes a computer device, which includes a processor and a memory. The memory stores computer instructions, and the processor executes the computer instructions stored in the memory. When the computer instructions are executed by the processor, the system implements the steps of the method according to any one of claims 1 to 9.
