News credibility evaluation system guided by time features

Abstract

The invention discloses a time feature-guided news credibility evaluation system, which comprises a data acquisition module for inputting a demand text and the release time of the text, and inputting a timestamp extractor; the data processing module is used for extracting timestamps according to the requirements with the timestamps and sorting the timestamps to generate a first data packet, and the data packet comprises news articles; the data analysis module is used for analyzing each piece of sub-data in the first data packet, decomposing the sub-data into a plurality of paragraph-level features capable of being connected in series, and identifying the paragraph-level features so as to screen proper data to generate a second data packet; and the data optimization module optimizes the features of each article through an inductive graph neural network and then outputs the features. According to the method, accurate and objective description and information related to actual events can be captured, so that the obtained reliable news is more accurate.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a news credibility evaluation system guided by time characteristics. Background Art

[0002] As social media becomes more and more popular, people are increasingly relying on it as a source of information.

[0003] At present, the methods for detecting false information generally adopt research based on the time dimension, that is, a specific time point, such as the release time of a news article. However, using a single time point as time information cannot fully capture the time dynamics. In the medical field, medical knowledge is crucial to evaluating the effectiveness of news. There are currently a variety of graph neural network models for knowledge fusion. Some studies combine the conduction graph neural network model with the knowledge graph for the task of false information detection. However, in many cases, the graph evolves as the network structure changes and new nodes appear. These methods usually require loading the entire graph to update the information, which may introduce noise and have a negative impact on the accuracy of the prediction. Summary of the invention

[0004] The purpose of the present invention is to provide a news credibility evaluation system guided by time characteristics.

[0005] In order to achieve the above object, the present invention provides the following technical solution: a news credibility evaluation system guided by time characteristics, comprising:

[0006] Data acquisition module, inputs the requirement text and the release time of the text, and inputs the timestamp extractor;

[0007] A data processing module extracts and sorts the timestamps according to the requirements with the timestamps to generate a first data packet, wherein the data packet includes the news article;

[0008] A data analysis module analyzes each sub-data in the first data packet, decomposes the sub-data into a number of paragraph-level features that can be connected in series, and identifies them to select suitable data to generate a second data packet;

[0009] The data optimization module optimizes the features of each article through the inductive graph neural network and then outputs it.

[0010] Preferably, the method for the data processing module to perform data matching on the requirements with timestamps includes:

[0011] S01, obtaining the requirement text to determine the input time;

[0012] S02, extracting time interval information from the article based on the input time:

[0013] If the article has a publication date, then:

[0014] t_interval i =Interval(t_post i , min{t1...t n});

[0015] Among them, Interval represents the time interval information of the extracted articles, t_post i represents the time information of article release, and {t1,...,tn} represents the time set of events mentioned in the article;

[0016] If the article has no publication date, t_interval i If it is a null value, the obtained time interval information is encoded with the original text and normalized to obtain the time interval representation:

[0017] t_rep i =tanh(Sigmoid(Encode(t_interval i )));

[0018] Wherein, Encode is the time interval information.

[0019] Preferably, the acquisition of paragraph-level features is based on the ALBERT pre-trained model on huggingface to obtain paragraph-level article features, and the paragraph-level feature representation a is encoded by using a bidirectional gated recurrent unit. i .

[0020] Preferably, the data analysis module further includes a knowledge embedding unit, which obtains the knowledge representation in the article based on the expert knowledge graph. i , knowledge representation k_rep i ,but:

[0021] k_rep i =getKnowledge(S i , Knowledge Graph);

[0022] The final article represents f i , expressed as:

[0023] f i =concat(a i , k_rep i , t_rep i );

[0024] Among them, a iIndicates article features, t_rep i Represents knowledge embedding, t_rep i Represents the time interval characteristics.

[0025] Preferably, the method for optimizing each article feature by inductive graph neural network in the data optimization module includes:

[0026] S03. Output generated by embedding: Output = graphSAGE(G, f, K, W k ), where G represents the input graph structure G(V,ε)f represents the input feature set K represents the search neighborhood depth of each node, W k Represents the weight matrix

[0027] S04. Operations for each aggregated neighbor: in, Indicates that through the aggregator function The result of each layer aggregation is Represents the neighborhood function N: i→2 V The final result of aggregating the nodes.

[0028] Preferably, the data optimization module includes sampling and aggregating multiple nodes, constructing a two-layer network with the node as the center, selecting up to 20 neighbors for sampling and aggregation to update the characteristics of the central node, and using a mean aggregator: Pick The element mean of the vector in , converts the unordered vector set into a vector;

[0029] Where N(i) represents the set of neighboring nodes of node i, and the mean aggregator function algorithm is as follows:

[0030]

[0031] Preferably, after obtaining the set of adjacent nodes based on i, the data optimization module updates the current node i using a combination of the previous representation of node i and the aggregated representation.

[0032] Then send the embedding result to the softmax layer for classification to obtain the prediction structure

[0033] In the above technical scheme, the news credibility assessment system guided by time features provided by the present invention has the following beneficial effects: using time intervals as feature representations, effectively capturing time dimension information, the inductive graph neural network improves the efficiency of graph loading during training, and domain expertise is mainly able to provide additional information, allowing the model to more fully judge the authenticity of news in the field. The entities extracted from the knowledge graph are used to enhance the semantic information of the article, and the inductive graph neural network is used to optimize the feature representation of each article by using a subset of adjacent articles in the article-article binary graph, which can more effectively learn article features while reducing the noise caused by unnecessary data. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0035] Figure 1 A schematic diagram of the module structure provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0036] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0037] Embodiment 1

[0038] like Figure 1 As shown, a news credibility evaluation system guided by time characteristics includes:

[0039] Data acquisition module, inputs the requirement text and the release time of the text, and inputs the timestamp extractor;

[0040] The data processing module matches data according to the requirements with the timestamp, and sorts the data before the timestamp to generate a first data packet, which includes the news article;

[0041] The data analysis module analyzes each sub-data in the first data packet, decomposes the sub-data into a number of paragraph-level features that can be connected in series, and identifies them to select suitable data to generate a second data packet;

[0042] The data optimization module optimizes the features of each article through the inductive graph neural network and then outputs it.

[0043] Specifically, the embodiment integrates the temporal encoder and textual knowledge into an embedding, learns information from the neighbors of an article in a graph, and then samples and aggregates the neighbors' information to update its feature representation, uses each updated article feature representation as input data, and outputs a predicted label.

[0044] The data processing module performs data matching on the requirements with timestamps in the following ways:

[0045] S01. Obtain the requirement text to determine the input time;

[0046] S02. Extracting time interval information from the article based on input time:

[0047] If the article has a publication date, then:

[0048] t_interval i =Interval(t_post i , min{t1...t n});

[0049] Among them, Interval represents the time interval information of the extracted articles, t_post i represents the time information of article release, and {t1,...,tn} represents the time set of events mentioned in the article;

[0050] If the article has no publication date, t_interval i If it is a null value, the obtained time interval information is encoded with the original text and normalized to obtain the time interval representation:

[0051] t_rep i =tanh(Sigmoid(Encode(t_interval i )));

[0052] Wherein, Encode is the time interval information.

[0053] In the above technology, the time interval is used as the feature representation to effectively capture the time dimension. The inductive graph neural network improves the efficiency of graph loading during training. The domain expertise can mainly provide additional information so that the model can more fully judge the authenticity of news in the field. The entities extracted from the knowledge graph are used to enhance the semantic information of the article, and the inductive graph neural network is used to optimize the feature representation of each article by using a subset of adjacent articles in the article-article binary graph, which can more effectively learn article features while reducing the noise caused by unnecessary data.

[0054] Embodiment 2

[0055] The acquisition of paragraph-level features is based on the ALBERT pre-trained model on huggingface to obtain paragraph-level article features, and the paragraph-level feature representation is encoded by using a bidirectional gated recurrent unit. i .

[0056] Furthermore, the data analysis module also includes a knowledge embedding unit, which obtains the knowledge representation in the article based on the expert knowledge graph. i , knowledge representation k_rep i ,but:

[0057] k_rep i =getKnowledge(S i , Knowledge Graph);

[0058] The final article represents f i , expressed as:

[0059] f i =concat(a i , k_rep i , t_rep i );

[0060] Among them, a i Indicates article features, k_rep i Represents knowledge embedding, t_rep i Represents the time interval feature representation.

[0061] Specifically, the embodiment focuses on the paragraph-level features of the article, thereby dividing the article into multiple mutually connected paragraph-level features, namely, the following characterizations, to ensure the accuracy of the analysis. The knowledge graph in the above embodiment can effectively organize and represent knowledge, so that it can be effectively used in advanced applications. The purpose of using knowledge embedding is to obtain entities in the knowledge graph. In order to obtain the knowledge representation in the article, the knowledge graph and the entities in the article are matched to obtain the modification of the article.

[0062] Embodiment 3

[0063] The methods for optimizing the features of each article through the inductive graph neural network in the data optimization module include:

[0064] S03. Output generated by embedding: Output = graphSAGE(G, f, K, W k ), where G represents the input graph structure G(V,ε)f represents the input feature set K represents the search neighborhood depth of each node, W k Represents the weight matrix

[0065] S04. Operations for each aggregated neighbor: in, Indicates that through the aggregator function The result of each layer aggregation is Represents the neighborhood function N: i→2 V The final result of aggregating the nodes.

[0066] Furthermore, the data optimization module includes sampling and aggregation of multiple nodes. It builds a two-layer network with the node as the center, selects up to 20 neighbors for sampling and aggregation to update the characteristics of the central node, and uses the mean aggregator: Pick The element mean of the vector in , converts the unordered vector set into a vector;

[0067] Where N(i) represents the set of neighboring nodes of node i, and the mean aggregator function algorithm is as follows:

[0068]

[0069] Secondly, after obtaining the set of adjacent nodes based on i in the data optimization module, the current node i is updated using the combination of the previous representation of node i and the aggregated representation.

[0070] Then send the embedding result to the softmax layer for classification to obtain the prediction structure

[0071] Specifically, the inductive graph neural network generates embeddings by sampling and aggregating features of vertex neighborhoods. Instead of learning the embeddings of all vertices on the graph, it learns a mapping to generate an embedding for each vertex.

[0072] The pseudo code of the graphSAGE embedding generation forward propagation algorithm is as follows:

[0073]

[0074] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0075] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, 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, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0076] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0077] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0078] The present invention uses specific embodiments to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

[0079] The embodiments of the present application also provide a specific implementation of an electronic device capable of implementing all steps of the method in the above embodiments, and the electronic device specifically includes the following contents:

[0080] Processor, memory, communications interface and bus;

[0081] Wherein, the processor, memory, and communication interface communicate with each other via the bus;

[0082] The processor is used to call the computer program in the memory, and when the processor executes the computer program, all the steps in the method in the above embodiment are implemented.

[0083] The embodiments of the present application also provide a computer-readable storage medium capable of implementing all the steps of the method in the above embodiments, wherein a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, all the steps of the method in the above embodiments are implemented.

[0084] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the hardware + program embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. Although the embodiment of this specification provides the method operation steps described in the embodiment or flow chart, more or less operation steps can be included based on conventional or non-creative means. The order of steps listed in the embodiment is only one way of executing the order of many steps, and does not represent the only execution order. When the device or terminal product in practice is executed, it can be executed in sequence or in parallel according to the method shown in the embodiment or the accompanying drawings (for example, a parallel processor or a multi-threaded processing environment, or even a distributed data processing environment). The term "include", "include" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method, product or device including a series of elements includes not only those elements, but also includes other elements that are not explicitly listed, or also includes elements inherent to such a process, method, product or device. In the absence of more restrictions, it is not excluded that there are other identical or equivalent elements in the process, method, product or device including the elements. For the convenience of description, the above device is described by dividing it into various modules according to its functions. Of course, when implementing the embodiments of this specification, the functions of each module can be implemented in the same one or more software and / or hardware, or the module implementing the same function can be implemented by a combination of multiple sub-modules or sub-units. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the coupling or direct coupling or communication connection between each other shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms. The present invention is described with reference to the flowchart and / or block diagram of the method, device (system) and computer program product according to the embodiments of the present invention. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the process and / or box in the flowchart and / or block diagram 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, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the process Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0085] Those skilled in the art will appreciate that the embodiments of this specification may be provided as methods, systems or computer program products. Therefore, the embodiments of this specification may be in the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, the embodiments of this specification may be in the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes. Each embodiment in this specification is described in a progressive manner, and the same and similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts refer to the partial description of the method embodiment. In the description of this specification, the description of the reference term "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the embodiment of this specification.

[0086] In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradicting each other. The above is only an embodiment of the embodiment of this specification and is not intended to limit the embodiment of this specification. For those skilled in the art, the embodiment of this specification may have various changes and variations. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the embodiment of this specification shall be included in the scope of the claims of the embodiment of this specification.

Claims

1. A news credibility evaluation system guided by time characteristics, characterized in that: include: Data acquisition module, inputs the requirement text and the release time of the text, and inputs the timestamp extractor; A data processing module extracts and sorts the timestamps according to the requirements with the timestamps to generate a first data packet, wherein the data packet includes the article; A data analysis module analyzes each sub-data in the first data packet, decomposes the sub-data into a plurality of paragraph-level features that can be connected in series, and identifies them to select suitable data to generate a second data packet; The data optimization module optimizes the features of each article through the inductive graph neural network and then outputs it.

2. A news credibility evaluation system guided by time characteristics according to claim 1, characterized in that: The method for the data processing module to perform data matching on the requirements with timestamps includes: S01, obtaining the requirement text to determine the input time; S02, extracting time interval information from the article based on the input time: If the article has a publication date, then: t_interval i =Interval(t_post i ,min{t1...t n }); Among them, Interval represents the time interval information of the extracted articles, t_post i represents the time information of article release, and {t1,...,tn} represents the time set of events mentioned in the article; If the article has no publication date, t_interval i If it is a null value, the obtained time interval information is encoded with the original text and normalized to obtain the time interval representation: t_rep i =tanh(Sigmoid(Encode(t_interval i ))); Wherein, Evcode is the time interval information.

3. A news credibility evaluation system guided by time characteristics according to claim 1, characterized in that: The acquisition of paragraph-level features is based on the ALBERT pre-trained model on huggingface to obtain paragraph-level article features, and the paragraph-level feature representation is encoded by using a bidirectional gated recurrent unit. i .

4. A news credibility evaluation system guided by time characteristics according to claim 3, characterized in that: The data analysis module also includes a knowledge embedding unit, which obtains the knowledge representation in the article based on the expert knowledge graph. i , knowledge representation k_rep i ,but: k_rep i =getKnowledge(S i ,Knowledge Graph); The final article represents f i , expressed as: f i =concat(a i ,k_rep i ,t_rep i ); Among them, a i Indicates article features, k_rep i Represents knowledge embedding, t_rep i Represents the time interval feature representation.

5. The news credibility evaluation system guided by time characteristics according to claim 1 is characterized in that: The method for optimizing the features of each article by inductive graph neural network in the data optimization module includes: S03. Output generated by embedding: Output = graphSAGE(G, f, K, W k ), where G represents the input graph structure G(V,ε)f represents the input feature set K represents the search neighborhood depth of each node, W k Represents the weight matrix S04. Operations for each aggregated neighbor: in, Indicates that the aggregator function AGGREGATE k , The result of each layer aggregation is Represents the neighborhood function N: i→2 V The final result of aggregating the nodes.

6. A news credibility evaluation system guided by time characteristics according to claim 5, characterized in that: The data optimization module includes sampling and aggregating multiple nodes. It builds a two-layer network with the node as the center, selects up to 20 neighbors for sampling and aggregation to update the characteristics of the central node, and uses a mean aggregator: Pick The element mean of the vector in the , converts the unordered vector set into a vector; Where N(i) represents the set of neighboring nodes of node i, and the mean aggregator function algorithm is as follows:

7. The news credibility evaluation system guided by time characteristics according to claim 1 is characterized in that: After obtaining the set of adjacent nodes based on i, the data optimization module updates the current node i using a combination of the previous representation of node i and the aggregated representation. Then send the embedding result to the softmax layer for classification to obtain the prediction structure 8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the visual positioning method based on local variance and posterior probability classifier according to any one of claims 1 to 7 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the visual positioning method based on local variance and posterior probability classifier described in any one of claims 1 to 7 are implemented.

Images