Microblog unreal information detection method and system based on double difference characteristics

Abstract

The invention provides a microblog unreal information detection method and system based on double difference characteristics, and relates to the technical field of natural language processing, and the method comprises the steps: obtaining to-be-detected microblog data; dividing the microblog comments to be detected into content-related comments and other comments; the difference between the microblog text to be detected and the content-related comments is extracted, and content difference features are obtained; extracting local emotion difference features between the to-be-detected microblog text and the content-related comments and global emotion difference features between the to-be-detected microblog text and other comments; splicing the content difference feature, the local emotion difference feature and the global emotion difference feature to obtain a content emotion dual difference feature; and inputting the content emotion double difference characteristics into a multi-layer perceptron model, and outputting the authenticity category of the unreal information of the microblog to be detected. According to the method, the accuracy of microblog unreal information detection can be effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of natural language processing, and in particular to a method and system for detecting false information on microblogs based on dual difference features. Background Art

[0002] Weibo false information usually refers to false or misleading information posted on the Weibo platform, which may be deliberately fabricated or caused by misunderstanding or miscommunication. Weibo false information detection based on dual difference features is a method to identify false information by analyzing the content and sentiment differences between Weibo texts and comments.

[0003] With the rapid development of Internet technology, more and more people choose to obtain information through social media platforms (such as Sina Weibo, Kuaishou, Douyin, Twitter and Facebook). Social media has gradually replaced traditional media as the mainstream channel for information dissemination. However, the information quality of social media is often not as good as that of traditional media, and it is full of a large amount of false information. The widespread dissemination of false information on Weibo not only confuses and manipulates public opinion, but may also cause the public to accept wrong prejudices and concepts, causing serious negative social impact. Therefore, effective identification of false information on Weibo is of great significance to maintaining social order, improving information quality, protecting the interests of individuals and organizations, and the sustainable development of the Weibo platform.

[0004] However, in the current false information detection technology, false information discrimination mainly relies on the text and comment data of Weibo. These methods only extract manual features of Weibo text or use deep learning technology to obtain its implicit features to achieve false information detection. Although the existing methods combining Weibo text and comments design an adaptive interactive fusion network to capture the similar and conflicting semantics between Weibo text and comments, the connection between the content and sentiment of the two is still insufficient, and there is a lack of comprehensive analysis of the relationship between the two. At the same time, when extracting content and sentiment features, most of them are associated through simple splicing or collaborative attention mechanism, without in-depth consideration of the event correlation between comments and Weibo text. In addition, in sentiment analysis, it is also unable to effectively distinguish content-related comments from other comments, resulting in low accuracy and poor false information detection effect. Summary of the invention

[0005] In order to solve the problem that the existing methods for detecting false information by combining microblog text and comments are still insufficient in mining the connection between the content and sentiment of the two, and lack a comprehensive analysis of the relationship between the two. At the same time, when extracting content and sentiment features, most of them are associated through simple splicing or collaborative attention mechanism, without in-depth consideration of the event correlation between comments and microblog text. In addition, in sentiment analysis, it is also unable to effectively distinguish between content-related comments and other comments, resulting in low accuracy and poor effect of false information detection. The present invention provides a method and system for detecting false information on microblogs based on dual difference features.

[0006] The technical solution provided by the embodiment of the present invention is as follows:

[0007] First aspect

[0008] The embodiment of the present invention provides a method for detecting false information on microblogs based on dual difference features, comprising:

[0009] S1: Acquire microblog data to be detected, wherein the microblog data to be detected includes microblog text to be detected and microblog comments to be detected;

[0010] S2: Divide the microblog comments to be detected into content-related comments and other comments;

[0011] S3: Extract the difference between the microblog text to be detected and the content-related comments to obtain the content difference features;

[0012] S4: extract the local sentiment difference features between the microblog text to be detected and the content-related comments, and the global sentiment difference features between the microblog text to be detected and other comments respectively;

[0013] S5: Concatenate content difference features, local emotion difference features and global emotion difference features to obtain content emotion dual difference features;

[0014] S6: Input the dual difference features of content and emotion into the multi-layer perceptron model to output the authenticity category of the false information of the microblog to be detected, where the authenticity category of the false information includes false information and non-false information.

[0015] Second aspect

[0016] An embodiment of the present invention provides a microblog false information detection system based on dual difference features, comprising:

[0017] processor;

[0018] A memory having computer-readable instructions stored thereon, which, when executed by a processor, implements the method for detecting false information on Weibo based on dual difference features as in the first aspect.

[0019] The third aspect

[0020] An embodiment of the present invention provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the method for detecting false information on Weibo based on dual difference features as in the first aspect is implemented.

[0021] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0022] In the present invention, by obtaining the microblog data to be detected including the microblog text to be detected and the microblog comments to be detected, basic data is provided for subsequent analysis, and then the nature of the comments is further distinguished, and the content difference features between the microblog text and the content-related comments, the local sentiment difference features between the microblog text and the content-related comments, and the global sentiment difference features between the microblog text and other comments are extracted respectively, and then the content difference features, the local sentiment difference features and the global sentiment difference features are spliced ​​to obtain the content-sentiment dual difference features, and finally, the comprehensive features are classified through the multi-layer perceptron model to realize the automatic identification of false information. The method is for each source post text and its corresponding comment text, by dividing the comments into content-related comments and other comments, extracting the content difference features between the microblog text and the content-related comments, and analyzing the sentiment difference features between the microblog text and the two types of comments, thereby obtaining the dual difference features, improving the effect of false information detection and the quality assurance of information, and improving the accuracy and robustness of false information detection on microblogs. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0024] Figure 1 A schematic diagram of a flow chart of a method for detecting false information on Weibo based on dual difference features provided by an embodiment of the present invention;

[0025] Figure 2 A schematic diagram of a detection framework of a method for detecting false information on Weibo based on dual difference features provided by an embodiment of the present invention;

[0026] Figure 3 A structural schematic diagram of a microblog false information detection system based on dual difference features provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0027] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0028] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "example" in the present invention should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of the word "example" is intended to present the concept in a specific way. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or it can be either of the two.

[0029] In order to make the technical problems, technical solutions and advantages to be solved by the present invention more clear, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0030] Reference Manual Attached Figure 1 , showing a flow chart of a method for detecting false information on Weibo based on dual difference features provided by an embodiment of the present invention.

[0031] The embodiment of the present invention provides a method for detecting false information on Weibo based on dual difference features, which can be implemented by a device for detecting false information on Weibo based on dual difference features, and the device for detecting false information on Weibo based on dual difference features can be a terminal or a server. The processing flow of the method for detecting false information on Weibo based on dual difference features can include the following steps:

[0032] S1: Obtaining microblog data to be detected, wherein the microblog data to be detected includes microblog text to be detected and microblog comments to be detected.

[0033] Among them, the microblog data to be tested refers to a microblog posted on the Weibo platform and all its associated user comments; the microblog text to be tested refers to the main content posted in the microblog, usually consisting of text and possible media (such as pictures, videos); the microblog comments to be tested refer to all comments associated with the microblog, which contain other users' views and reactions to the microblog content.

[0034] It should be noted that by fully acquiring the Weibo text and comment data, sufficient information is provided for multi-angle analysis. The Weibo text and related comments are input as overall data, ensuring that the context of the text and the user's feedback and reaction can be fully understood in subsequent steps, laying the foundation for accurate detection of false information.

[0035] Reference Manual Attached Figure 2 , showing a detection framework schematic diagram of a method for detecting false information on Weibo based on dual difference features provided by an embodiment of the present invention.

[0036] Figure 2In the method, based on the keywords of the microblog text to be detected, the microblog comments to be detected are divided into content-related comments and other comments. Content-related comments refer to comments containing the keywords of the microblog text to be detected, and other comments are comments that do not contain the keywords of the microblog text to be detected. Then, the text content features and text sentiment features of the microblog text to be detected are extracted respectively, and the comment content features and comment sentiment features are extracted from the content-related comments, and the sentiment features of other comments in other comments are extracted. The content difference features are obtained by calculating the difference between the content features of the microblog text to be detected and the comment content features, the difference between the sentiment features of the microblog text to be detected and the sentiment features of the content-related comments is calculated to obtain the local sentiment difference features, the difference between the sentiment features of the microblog text to be detected and the sentiment features of the content-related comments is calculated to obtain the global sentiment difference features, and finally, the content difference features, the local sentiment difference features and the global sentiment difference features are spliced ​​together through the splicing operation to form a content-sentiment dual difference feature. The MLP model processes the input content-sentiment dual difference features and finally outputs the classification result to distinguish whether the microblog to be detected is false information or not.

[0037] S2: Divide the microblog comments to be detected into content-related comments and other comments.

[0038] Among them, content-related comments refer to comments that are closely related to the topic of the Weibo text. These comments contain the same or similar topic keywords as the Weibo text, reflecting the user's direct response or opinion on the Weibo content. Other comments refer to comments that are not directly related to the topic of the Weibo text. They may contain discussions, chats, etc. that are unrelated to the text topic.

[0039] It should be noted that by dividing the comments, the detection process can distinguish different types of comments, so as to analyze the comment information related to the Weibo content in a targeted manner. This division can reduce the interference of irrelevant information, highlight key comments, and enhance the accuracy of the connection analysis between Weibo text and comments.

[0040] In a possible implementation manner, S2 is specifically:

[0041] S201: extract the subject keywords of the microblog text to be detected, traverse the microblog comments to be detected, and determine whether the subject keywords exist in the microblog comments to be detected. If so, classify the microblog comments to be detected as content-related comments, otherwise, classify the microblog comments to be detected as other comments.

[0042] It should be noted that by extracting the topic keywords of the Weibo text and then judging whether the comments contain these keywords, it is possible to more accurately identify comments that are closely related to the topic of the Weibo text. This method can capture comments that are consistent with the main theme of the Weibo, thereby improving the accuracy of classification.

[0043] In a possible implementation, S201 specifically includes:

[0044] S2011: Extracting multiple topic keywords of different topics from the microblog text to be detected through the latent Dirichlet allocation topic model.

[0045] Among them, the Latent Dirichlet Allocation (LDA) topic model is an unsupervised learning method for text topic extraction. It discovers potential topics in documents through statistical analysis. Topic keywords are high-frequency words representing specific topics generated by the LDA model based on Weibo text. They represent the main content or topic of the Weibo text.

[0046] S2012: De-duplicate the obtained subject keywords.

[0047] S2013: traverse the microblog comments to be detected to see if any of the removed topic keywords exists. If so, classify the microblog comments to be detected as content-related comments. Otherwise, classify the microblog comments to be detected as other comments.

[0048] It should be noted that by extracting multiple keywords of different topics through the LDA model, the main content of the Weibo text can be covered more comprehensively. The LDA model can identify potential topics in the Weibo text and generate representative high-frequency words based on each topic. The extracted keywords can more comprehensively and accurately represent the core content of the Weibo, making the comment classification more accurate.

[0049] S3: Extract the difference between the microblog text to be detected and the content-related comments to obtain content difference features.

[0050] Among them, the content difference feature is a feature representation obtained by calculating the difference in content features between the microblog text to be detected and its content-related comments. In this way, the similarities or contradictions in the information content between the comments and the text can be revealed. Extracting the content difference feature can effectively capture the possible questioning, supplementary or refutation information about the text content in the comments, so as to better judge the credibility of the text. By focusing on analyzing content-related comments, the interference caused by other comments is reduced, making the feature extraction more accurate.

[0051] In a possible implementation manner, the calculation formula of the content difference feature is specifically:

[0052]

[0053] Among them, SD i Represents the i-th microblog text S to be detected iContent difference features between content-related comments. Sem() represents content features extracted based on the BERT pre-training model. represents the jth content-related comment in the content-related comment set, j = 1, 2, ..., R, R represents the content-related comment set of the microblog data to be detected The number of content-related comments included in , Σ represents the summation symbol, and | | represents the absolute value symbol.

[0054] It should be noted that by utilizing the powerful semantic understanding ability of the BERT model, it is possible to accurately extract and analyze the content differences between Weibo text and content-related comments. At the same time, by averaging the absolute values ​​of the comment differences, the robustness and reliability of the features are improved. Overall, this method can more accurately measure the consistency or contradiction between the text and the comments, providing effective content feature support for false information detection.

[0055] S4: Extract the local sentiment difference features between the microblog text to be detected and the content-related comments, and the global sentiment difference features between the microblog text to be detected and other comments respectively.

[0056] Among them, the local sentiment difference feature refers to the difference in sentiment expression between Weibo text and content-related comments. These comments usually reflect the user's direct emotional response to specific content in Weibo. The global sentiment difference feature refers to the sentiment difference between Weibo text and other comments. These comments usually express the user's attitude towards the overall Weibo content.

[0057] It should be noted that by refining sentiment analysis, comments are subdivided into content-related and other comments, so as to extract local and global sentiment difference features respectively. Local sentiment differences can reveal users' reactions and positions to specific content, while global sentiment differences reflect users' attitudes towards the overall Weibo. This detailed sentiment feature extraction helps to capture the sentiment relationship between comments and text more comprehensively, avoid the limitations of single sentiment analysis, and thus improve the accuracy and sensitivity of false information detection.

[0058] In a possible implementation, the calculation formula of the local emotion difference feature is specifically:

[0059]

[0060] Among them, LED i Represents the i-th microblog text S to be detected i The local sentiment difference features between comments related to the content, emo() means to extract sentiment features using sentiment dictionary, represents the jth content-related comment in the set of content-related comments, Σ represents the summation symbol, and || represents the absolute value symbol.

[0061] It should be noted that by focusing on the sentiment difference analysis of content-related comments, it is possible to more specifically reflect the emotional consistency or conflict between the Weibo text and these comments. By calculating the fine-grained sentiment difference, the interference of extreme comments can be reduced. In addition, the sentiment dictionary can be used to improve the accuracy of sentiment feature extraction, providing richer and more reliable sentiment information for false information detection.

[0062] In a possible implementation, the calculation formula of the global sentiment difference feature is specifically:

[0063]

[0064] in, represents the j'th other comment in the other comment set, j'=1,2,…,O, O represents the other comment set of the microblog data to be detected The number of other reviews included in GED i Represents the i-th microblog text S to be detected i With other comments The global emotional difference characteristics between them.

[0065] It should be noted that by averaging the sentiment differences of other comments, the global sentiment deviation between Weibo text and comments can be captured comprehensively and robustly, providing a global perspective of sentiment features for false information detection, thereby further enhancing the accuracy and robustness of detection.

[0066] S5: Concatenate content difference features, local emotion difference features and global emotion difference features to obtain content emotion dual difference features.

[0067] Among them, the content-emotion dual difference feature is a comprehensive feature vector formed by splicing the above-mentioned content difference feature and emotion difference feature, which expresses the overall relationship between text and comments in terms of content and emotion.

[0068] It should be noted that by splicing together the features of the two dimensions of content difference and sentiment difference, a comprehensive feature vector containing multi-level information is formed. This feature splicing method enables the model to analyze the relationship between Weibo text and comments from the dual perspectives of content and sentiment, thereby capturing the characteristics of false information more comprehensively.

[0069] In a possible implementation, the calculation formula of the content-emotion dual difference feature is specifically:

[0070]

[0071] Among them, D i Represents the i-th microblog text S to be detected iThe dual difference characteristics of content and emotion between the microblog comments to be detected, SD i Represents the i-th microblog text S to be detected i Content difference features between content-related comments, LED i Represents the i-th microblog text S to be detected i Local sentiment difference feature between content-related comments, GED i Represents the i-th microblog text S to be detected i With other comments The global emotional difference characteristics between Represents a concatenation operation.

[0072] It should be noted that by concatenating multiple features into a comprehensive feature vector, the model input contains more information dimensions. This approach can improve the expressive power of deep learning models (such as MLP), enabling the model to better learn the complex relationship between text and comments, and thus more accurately distinguish between false information and non-false information during classification.

[0073] S6: Input the dual difference features of content and emotion into the multi-layer perceptron model to output the authenticity category of the false information of the microblog to be detected, where the authenticity category of the false information includes false information and non-false information.

[0074] Among them, the Multi-Layer Perceptron (MLP) model is a classic neural network model that can learn the complex relationship between input and output through linear and nonlinear combinations of multiple layers of neurons. The true and false category of false information refers to the result category of the final output of the model, which is used to distinguish false information from non-false information.

[0075] The main reasons for the prevalence of false information on Weibo include: (1) Users freely post and disseminate posts on the Weibo platform, resulting in a high degree of uncertainty in the credibility of information sources; (2) The content of false information posts is often inflammatory and concealed, making it difficult for users to distinguish between true and false, and they are easily gullible and further spread false information; (3) The Weibo platform's monitoring methods are relatively weak and the punishment is low, which makes the cost of spreading false information low, thus breeding false information.

[0076] It should be noted that the multi-layer perceptron model is used to classify and predict the dual difference features of the spliced ​​content and emotion, and the powerful learning ability of the neural network is used to capture the complex nonlinear relationship between features. The MLP model can accurately classify false information based on the comprehensive analysis of multiple features.

[0077] In a possible implementation manner, S6 specifically includes:

[0078] Output the authenticity category of the false information in the microblog to be detected through the following formula:

[0079]

[0080] in, represents the false information authenticity prediction result of the i-th microblog text to be detected, softmax represents the softmax activation function, and MLP represents the multi-layer perceptron model.

[0081] In the present invention, the IP address, port range, domain name, URL and / or TCP message that needs to be blocked is determined in an automated manner according to the abnormal category, so that the network abnormality can be responded to quickly and timely. No manual intervention is required, which can reduce the response time and improve the real-time performance and efficiency of the system. The target that needs to be blocked can be accurately determined according to the abnormal category, thereby preventing the spread of malicious attacks. The source of abnormal data and related information can be blocked in a timely manner to curb the spread of attack behaviors and protect network security. Corresponding bypass blocking based on the abnormal category of network transmission data can automatically respond to network abnormalities, prevent the spread of malicious attacks, protect system and data security, and improve network reliability. It also has flexibility and accuracy to effectively respond to various network security threats.

[0082] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0083] In the present invention, by obtaining the microblog data to be detected including the microblog text to be detected and the microblog comments to be detected, basic data is provided for subsequent analysis, and then the nature of the comments is further distinguished, and the content difference features between the microblog text and the content-related comments, the local sentiment difference features between the microblog text and the content-related comments, and the global sentiment difference features between the microblog text and other comments are extracted respectively, and then the content difference features, the local sentiment difference features and the global sentiment difference features are spliced ​​to obtain the content-sentiment dual difference features, and finally, the comprehensive features are classified through the multi-layer perceptron model to realize the automatic identification of false information. The method is for each source post text and its corresponding comment text, by dividing the comments into content-related comments and other comments, extracting the content difference features between the microblog text and the content-related comments, and analyzing the sentiment difference features between the microblog text and the two types of comments, thereby obtaining the dual difference features, improving the effect of false information detection and the quality assurance of information, and improving the accuracy and robustness of false information detection on microblogs.

[0084] Reference Manual Attached Figure 3 , showing a structural schematic diagram of a microblog false information detection system based on dual difference features provided by the present invention.

[0085] The present invention further provides a microblog false information detection system 20 based on dual difference features, which is applied to the above-mentioned microblog false information detection method based on dual difference features, and comprises:

[0086] Processor 201.

[0087] The memory 202 stores computer-readable instructions. When the computer-readable instructions are executed by the processor 201, a method for detecting false information on microblogs based on dual difference features as in the method embodiment is implemented.

[0088] The microblog false information detection system 20 based on dual difference features provided by the present invention can execute the above-mentioned microblog false information detection method based on dual difference features and achieve the same or similar technical effects. To avoid repetition, the present invention will not go into details.

[0089] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0090] In the present invention, by obtaining the microblog data to be detected including the microblog text to be detected and the microblog comments to be detected, basic data is provided for subsequent analysis, and then the nature of the comments is further distinguished, and the content difference features between the microblog text and the content-related comments, the local sentiment difference features between the microblog text and the content-related comments, and the global sentiment difference features between the microblog text and other comments are extracted respectively, and then the content difference features, the local sentiment difference features and the global sentiment difference features are spliced ​​to obtain the content-sentiment dual difference features, and finally, the comprehensive features are classified through the multi-layer perceptron model to realize the automatic identification of false information. The method is for each source post text and its corresponding comment text, by dividing the comments into content-related comments and other comments, extracting the content difference features between the microblog text and the content-related comments, and analyzing the sentiment difference features between the microblog text and the two types of comments, thereby obtaining the dual difference features, improving the effect of false information detection and the quality assurance of information, and improving the accuracy and robustness of false information detection on microblogs.

[0091] It should be understood that the processor in the embodiment of the present invention may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0092] It should also be understood that the memory in the embodiments of the present invention may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic random access memory (DRAM), synchronous DRAM (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0093] The above embodiments can be implemented in whole or in part by software, hardware (such as circuits), firmware or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When a computer instruction or computer program is loaded or executed on a computer, a process or function according to an embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center by wired (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a tape), an optical medium (for example, a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0094] It should be understood that the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. A and B can be singular or plural. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship, but it may also indicate an "and / or" relationship. Please refer to the context for specific understanding.

[0095] In the present invention, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can be represented by: a, b, c, ab, ac, bc, or abc, where a, b, c can be single or multiple.

[0096] It should be understood that in various embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0097] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0098] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described equipment, devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0099] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection 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.

[0100] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0101] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0102] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0103] An embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon. When the program is executed by a processor, the method for detecting false information on Weibo based on dual difference features as described in the method embodiment is implemented.

[0104] A computer-readable storage medium provided by the present invention can implement the steps and effects of the method for detecting false information on Weibo based on dual difference features in the above method embodiment. To avoid repetition, the present invention will not go into details.

[0105] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0106] In the present invention, by obtaining the microblog data to be detected including the microblog text to be detected and the microblog comments to be detected, basic data is provided for subsequent analysis, and then the nature of the comments is further distinguished, and the content difference features between the microblog text and the content-related comments, the local sentiment difference features between the microblog text and the content-related comments, and the global sentiment difference features between the microblog text and other comments are extracted respectively, and then the content difference features, the local sentiment difference features and the global sentiment difference features are spliced ​​to obtain the content-sentiment dual difference features, and finally, the comprehensive features are classified through the multi-layer perceptron model to realize the automatic identification of false information. The method is for each source post text and its corresponding comment text, by dividing the comments into content-related comments and other comments, extracting the content difference features between the microblog text and the content-related comments, and analyzing the sentiment difference features between the microblog text and the two types of comments, thereby obtaining the dual difference features, improving the effect of false information detection and the quality assurance of information, and improving the accuracy and robustness of false information detection on microblogs.

[0107] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art who is familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

[0108] There are a few points to note:

[0109] (1) The drawings of the embodiments of the present invention only relate to the structures related to the embodiments of the present invention, and other structures may refer to the general design.

[0110] (2) For the sake of clarity, in the drawings used to describe the embodiments of the present invention, the thickness of the layers or regions is exaggerated or reduced, that is, these drawings are not drawn according to the actual scale. It is understood that when an element such as a layer, film, region or substrate is referred to as being "on" or "under" another element, the element may be "directly" "on" or "under" the other element or there may be intermediate elements.

[0111] (3) In the absence of conflict, the embodiments of the present invention and the features therein may be combined with each other to obtain new embodiments.

[0112] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. The protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A method for detecting false information on microblogs based on dual difference features, characterized in that: include: S1: Acquire microblog data to be detected, wherein the microblog data to be detected includes microblog text to be detected and microblog comments to be detected; S2: Divide the microblog comments to be detected into content-related comments and other comments; S3: extracting the difference between the microblog text to be detected and the content-related comments to obtain content difference features; S4: extracting local sentiment difference features between the microblog text to be detected and the content-related comments and global sentiment difference features between the microblog text to be detected and the other comments respectively; S5: combining the content difference feature, the local emotion difference feature and the global emotion difference feature to obtain a content emotion dual difference feature; S6: Input the content-emotion dual difference features into a multi-layer perceptron model, and output the false information authenticity category of the microblog to be detected, wherein the false information authenticity category includes false information and non-false information.

2. The microblog false information detection method based on dual difference features according to claim 1 is characterized in that: The S2 is specifically: S201: extracting the subject keywords of the microblog text to be detected, traversing the microblog comments to be detected, and determining whether the subject keywords exist in the microblog comments to be detected; if so, classifying the microblog comments to be detected as content-related comments; otherwise, classifying the microblog comments to be detected as other comments.

3. The microblog false information detection method based on dual difference features according to claim 1 is characterized in that: The S201 specifically includes: S2011: extracting a plurality of topic keywords of different topics from the microblog text to be detected by using a latent Dirichlet allocation topic model; S2012: De-duplication of the obtained subject keywords; S2013: traverse the microblog comments to be detected to see whether any of the removed duplicate subject keywords exists. If so, classify the microblog comments to be detected as content-related comments. Otherwise, classify the microblog comments to be detected as other comments.

4. The microblog false information detection method based on dual difference features according to claim 1 is characterized in that: The calculation formula of the content difference feature is specifically: Among them, SD i Represents the i-th microblog text S to be detected i Content difference features between content-related comments. Sem() represents content features extracted based on the BERT pre-training model. represents the jth content-related comment in the content-related comment set, j = 1, 2, ..., R, R represents the content-related comment set of the microblog data to be detected The number of content-related comments included in , Σ represents the summation symbol, and || represents the absolute value symbol.

5. The microblog false information detection method based on dual difference features according to claim 1 is characterized in that: The calculation formula of the local emotional difference feature is specifically: Among them, LED i Represents the i-th microblog text S to be detected i The local sentiment difference features between comments related to the content, emo() means to extract sentiment features using sentiment dictionary, represents the jth content-related comment in the set of content-related comments, ∑ represents the summation symbol, and || represents the absolute value symbol.

6. The microblog false information detection method based on dual difference features according to claim 1 is characterized in that: The calculation formula of the global sentiment difference feature is specifically: in, represents the j'th other comment in the other comment set, j'=1,2,…,O, O represents the other comment set of the microblog data to be detected The number of other reviews included in GED i Represents the i-th microblog text S to be detected i With other comments The global emotional difference characteristics between them.

7. The microblog false information detection method based on dual difference features according to claim 1 is characterized in that: The calculation formula of the content emotion dual difference feature is specifically: Among them, D i Represents the i-th microblog text S to be detected i The dual difference characteristics of content and emotion between the microblog comments to be detected, SD i Represents the i-th microblog text S to be detected i Content difference features between content-related comments, LED i Represents the i-th microblog text S to be detected i Local sentiment difference feature between content-related comments, GED i Represents the i-th microblog text S to be detected i With other comments The global emotional difference characteristics between Represents a concatenation operation.

8. The microblog false information detection method based on dual difference features according to claim 7 is characterized in that: The S6 is specifically: The authenticity category of the false information of the microblog to be detected is output by the following formula: in, represents the false information authenticity prediction result of the i-th microblog text to be detected, softmax represents the softmax activation function, and MLP represents the multi-layer perceptron model.

9. An attack bypass blocking system based on network abnormal behavior analysis, characterized in that: include: processor; A memory having computer-readable instructions stored thereon, wherein when the computer-readable instructions are executed by the processor, the microblog false information detection method based on dual difference features as described in any one of claims 1 to 8 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method for detecting false information on Weibo based on dual difference features as described in any one of claims 1 to 8 is implemented.

Images