A microblog false information detection method and system based on double difference features

Abstract

The application provides a microblog false information detection method and system based on double difference features, and relates to the technical field of natural language processing.The method comprises the following steps: obtaining microblog data to be detected; dividing microblog comments to be detected into content-related comments and other comments; extracting the difference between the microblog text to be detected and the content-related comments to obtain content difference features; respectively 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; splicing the content difference features, the local sentiment difference features and the global sentiment difference features to obtain content-sentiment double difference features; and inputting the content-sentiment double difference features into a multilayer perception machine model to output the true or false category of the microblog to be detected.The application can effectively improve the accuracy of microblog false information detection.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of natural language processing, and particularly relates to a microblog false information detection method and system based on double difference features. BACKGROUND

[0002] Microblog false information generally refers to false or misleading information published on microblog platforms. These information may be intentionally fabricated or caused by misunderstanding or misinformation, and microblog false information detection based on double difference features is a method of identifying false information by analyzing the content and emotional differences between microblog 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, TikTok, Twitter and Facebook), and social media gradually replaces traditional media as the mainstream channel of information dissemination. However, the quality of information on social media is often not as good as that of traditional media, and there are a large number of false information. The widespread dissemination of microblog false information not only confuses and manipulates public opinion, but also may make the public accept false prejudices and concepts, causing serious social negative effects. Therefore, effective identification of microblog false information is of great significance for maintaining social order, improving information quality, protecting the interests of individuals and organizations and the sustainable development of microblog platforms.

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

[0005] In order to solve the technical problems in the prior art that the existing combined microblog text and comment false information detection method still lacks mining of the relationship between the content and emotion of the two, and lacks comprehensive analysis of the relationship between the two, and when extracting the content and emotion features, most of them are associated through simple splicing or collaborative attention mechanism, without deeply considering the event correlation of the comments and the microblog text, and in the emotion analysis, the content-related comments and other comments cannot be effectively distinguished, resulting in low false information detection accuracy and poor false information detection effect, the present application provides a microblog false information detection method and system based on double difference features.

[0006] The technical scheme provided by the embodiment of the present application is as follows:

[0007] The first aspect

[0008] The microblog false information detection method based on double difference features provided by the embodiment of the present application comprises:

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

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

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

[0012] S4: respectively extracting a local emotion difference feature between the microblog text to be detected and the content-related comments and a global emotion difference feature between the microblog text to be detected and the other comments;

[0013] S5: splicing the content difference feature, the local emotion difference feature and the global emotion difference feature to obtain a content emotion double difference feature;

[0014] S6: inputting the content emotion double difference feature into a multi-layer perception machine model to output a false information true or false category of the microblog to be detected, wherein the false information true or false category comprises false information and non-false information.

[0015] The second aspect

[0016] The microblog false information detection system based on double difference features provided by the embodiment of the present application comprises:

[0017] A processor;

[0018] A memory, the memory stores computer readable instructions, and the computer readable instructions are executed by the processor to realize the microblog false information detection method based on double difference features of the first aspect.

[0019] Third aspect

[0020] The computer readable storage medium provided by the embodiment of the present application stores a computer program, and the program is executed by a processor to realize the microblog false information detection method based on double difference features according to the first aspect.

[0021] The technical scheme provided by the embodiment of the present application has at least the following beneficial effects:

[0022] In the present application, the microblog data to be detected including the microblog text to be detected and the microblog comment to be detected is acquired to provide basic data for subsequent analysis. Then, the nature of the comment is further distinguished, and the content difference features between the microblog text and the content-related comment, the local sentiment difference features between the microblog text and the content-related comment, and the global sentiment difference features between the microblog text and other comments are extracted respectively. Then, the content difference features, the local sentiment difference features and the global sentiment difference features are spliced to obtain the content-sentiment double difference features. Finally, the comprehensive features are classified by the multilayer perception machine model to realize the automatic identification of false information. According to the present application, for each source post text and the corresponding comment text, the comment is divided into content-related comment and other comment, the content difference features between the microblog text and the content-related comment are extracted, and the sentiment difference features between the microblog text and the two types of comments are analyzed to obtain the double difference features, thereby improving the effect of false information detection and the quality of information guarantee, and improving the accuracy and robustness of microblog false information detection. BRIEF DESCRIPTION OF DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0024] Figure 1 The flowchart of the microblog false information detection method based on double difference features provided by the embodiment of the present application is shown in the figure.

[0025] Figure 2 The detection framework diagram of the microblog false information detection method based on double difference features provided by the embodiment of the present application is shown in the figure.

[0026] Figure 3 The structure diagram of the microblog false information detection system based on double difference features provided by the embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0027] The technical solutions in the present application will be described below with reference to the drawings.

[0028] In the embodiments of the present application, the words such as "example", "for example" are used to represent an example, illustration, or description. Any embodiment or design scheme described as "example" in the present application should not be interpreted as more preferred or more advantageous than other embodiments or design schemes. Rather, the word "example" is intended to present the concept in a specific manner. In addition, in the embodiments of the present application, the meaning expressed by "and / or" can be both, or can be one of the two.

[0029] In order to make the technical problems, technical solutions and advantages to be solved by the present application more clear, the following will be described in detail in combination with the drawings and specific embodiments.

[0030] Referring to the drawings attached to the specification Figure 1 , a flowchart of a microblog false information detection method based on double difference features provided by an embodiment of the present application is shown.

[0031] An embodiment of the present application provides a microblog false information detection method based on double difference features, which can be realized by a microblog false information detection device based on double difference features. The microblog false information detection device based on double difference features can be a terminal or a server. The processing flow of the microblog false information detection method based on double difference features can include the following steps:

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

[0033] It should be noted that by completely obtaining the microblog text and comment data, sufficient information is provided for multi-angle analysis. The microblog text and related comments are input as a whole data, which ensures that the background of the text and the feedback and reaction of the user can be comprehensively understood in the subsequent steps, thereby laying a foundation for accurate detection of false information.

[0034] It should be noted that by completely obtaining the microblog text and comment data, sufficient information is provided for multi-angle analysis. The microblog text and related comments are input as a whole data, which ensures that the background of the text and the feedback and reaction of the user can be comprehensively understood in the subsequent steps, thereby laying a foundation for accurate detection of false information.

[0035] Referring to the drawings attached to the specification Figure 2 , a detection framework diagram of a microblog false information detection method based on double difference features provided by an embodiment of the present application is shown.

[0036] Figure 2In specific, based on the keywords of the micro-blog text to be detected, the micro-blog comments to be detected are divided into content-related comments and other comments, the content-related comments refer to the comments containing the keywords of the micro-blog text to be detected, and the other comments refer to the comments not containing the keywords of the micro-blog text to be detected, then the text content features and the text sentiment features of the micro-blog text to be detected are extracted, the comment content features and the comment sentiment features are extracted from the content-related comments, the other comment sentiment features are extracted from the other comments, the content difference features are obtained by calculating the difference between the text content features of the micro-blog text to be detected and the comment content features, the local sentiment difference features are obtained by calculating the difference between the text sentiment features of the micro-blog text to be detected and the comment sentiment features of the content-related comments, the global sentiment difference features are obtained by calculating the difference between the text sentiment features of the micro-blog text to be detected and the other comment sentiment features, finally, the content difference features, the local sentiment difference features and the global sentiment difference features are spliced together to form a content-sentiment dual difference feature, the MLP model processes the input content-sentiment dual difference feature, and finally outputs the classification result to distinguish whether the micro-blog to be detected is false information or non-false information.

[0037] S2: dividing the micro-blog comments to be detected into content-related comments and other comments.

[0038] In specific, the content-related comments refer to the comments closely related to the theme of the micro-blog text, these comments contain the same or similar theme keywords as the micro-blog text, and reflect the direct response or opinion of the user to the micro-blog content, and the other comments refer to the comments not directly related to the theme of the micro-blog text, which may contain unrelated discussions, casual conversations and other contents.

[0039] It should be noted that through the division of the comments, the detection process can distinguish different types of comments, so as to analyze the comment information related to the micro-blog content in a targeted manner, which can reduce the interference of irrelevant information, highlight the key comments, and enhance the accuracy of the analysis of the relationship between the micro-blog text and the comments.

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

[0041] S201: extracting the theme keywords of the micro-blog text to be detected, traversing the micro-blog comments to be detected, judging whether the theme keywords exist in the micro-blog comments to be detected, if yes, dividing the micro-blog comments to be detected into content-related comments, otherwise, dividing the micro-blog comments to be detected into other comments.

[0042] It should be noted that by extracting the theme keywords of the micro-blog text, and then judging whether the keywords exist in the comments, the comments closely related to the theme of the micro-blog text can be accurately identified, and this method can capture the comments consistent with the theme of the micro-blog, thereby improving the accuracy of classification.

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

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

[0045] The latent Dirichlet allocation topic model (LDA) is an unsupervised learning method for text topic extraction, which discovers the latent topics in the document through statistical analysis. The topic keyword is a high-frequency word representing a specific topic generated by the LDA model according to the microblog text, which represents the main content or topic of the microblog text.

[0046] S2012: Removing the duplicate keywords from the obtained topic keywords.

[0047] S2013: Traversing the detected microblog comments to determine whether any of the removed topic keywords exist, if so, dividing the detected microblog comments into content-related comments, otherwise, dividing the detected microblog comments into other comments.

[0048] It should be noted that the LDA model can extract a plurality of keywords of different topics, which can more comprehensively cover the main content of the microblog text. The LDA model can identify the latent topics in the microblog text and generate representative high-frequency words according to each topic. The extracted keywords can more comprehensively and accurately represent the core content of the microblog, so that the accuracy of the comment classification is higher.

[0049] S3: Extracting the difference between the detected microblog text and the content-related comments to obtain a content difference feature.

[0050] The content difference feature is obtained by calculating the difference between the content features of the detected microblog text and its content-related comments. This method can reveal the similarities or contradictions between the comments and the text in the information content. Extracting the content difference feature can effectively capture the questioning, supplementing or refuting information in the comments about the text content, so as to better judge the credibility of the text. By focusing on analyzing the content-related comments, the interference of other comments is reduced, so that the feature extraction is more accurate.

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

[0052]

[0053] Wherein, SD i represents the i-th detected microblog text S icontent difference between the content-related comments and the content, Sem() represents the content features extracted based on the BERT pre-training model, represents the jth content-related comment in the set of content-related comments, j = 1, 2,..., R, and R represents the set of content-related comments of the microblog data to be detected the number of content-related comments included in the set of content-related comments, ∑ 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, the content difference between the microblog text and the content-related comments can be accurately extracted and analyzed, and meanwhile, by performing the absolute value average processing on the comment difference, the robustness and reliability of the features are improved, and overall, this method can more accurately measure the consistency or contradiction between the text and the comments in terms of content, thereby providing effective content feature support for the false information detection.

[0055] S4: Extracting 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 features refer to the difference in emotional expression between the microblog text and the content-related comments, and these comments usually reflect the direct emotional reaction of the user to the specific content in the microblog; the global sentiment difference features refer to the emotional difference between the microblog text and other comments, and these comments usually express the attitude of the user to the overall microblog content.

[0057] It should be noted that by refining the sentiment analysis, the comments are subdivided into content-related and other comments, so as to extract the local and global sentiment difference features, respectively. The local sentiment difference can reveal the reaction and position of the user to the specific content, while the global sentiment difference reflects the attitude of the user to the overall microblog. This detailed sentiment feature extraction is helpful to more comprehensively capture the emotional relationship between the comments and the text, avoid the limitations of single sentiment analysis, and thus improve the accuracy and sensitivity of the false information detection.

[0058] In one possible implementation, the calculation formula of the local sentiment difference features is specifically as follows:

[0059]

[0060] Among them, LED i represents the ith microblog text S i to be detected and the content-related comments, emo() represents the extraction of the sentiment features by using the sentiment dictionary, represents the jth content-related comment in the set of content-related comments, and ∑ represents the summation symbol.

[0061] It should be noted that by focusing on the sentiment difference analysis of content-related comments, the consistency or conflict of microblog text and the comments in sentiment can be more targetedly reflected, the interference of extreme comments can be reduced through fine-grained sentiment difference calculation, and the extraction accuracy of sentiment features can be improved by using a sentiment dictionary, thereby providing more abundant 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 as follows:

[0063]

[0064] wherein, 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 comments included in the global sentiment difference feature GED i represents the i'th microblog text S i to be detected.

[0065] It should be noted that by averaging the sentiment difference of other comments, the global sentiment deviation between the microblog text and the comments can be comprehensively and robustly captured, a global perspective sentiment feature is provided for false information detection, and the accuracy and robustness of detection are further enhanced.

[0066] S5: Splicing the content difference feature, the local sentiment difference feature and the global sentiment difference feature to obtain a content-sentiment dual difference feature.

[0067] The content-sentiment dual difference feature is a comprehensive feature vector formed by splicing the content difference feature and the sentiment difference feature, and expresses the overall relationship between the text and the comments in terms of content and sentiment.

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

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

[0070]

[0071] wherein, D i represents the i'th microblog text S i ​SD, the content sentiment double difference feature between the micro-blog comment to be detected and the content i represents the i-th micro-blog text to be detected S i LED, the content difference feature between the content-related comment i represents the i-th micro-blog text to be detected S i GED, the local sentiment difference feature between the content-related comment i represents the i-th micro-blog text to be detected S i the global sentiment difference feature between the other comments and the content represents the splicing operation.

[0072] It should be noted that by splicing multiple features into a comprehensive feature vector, the input of the model contains more information dimensions, which can improve the expression ability of the deep learning model (such as MLP), so that the model can better learn the complex relationship between the text and the comment, thereby more accurately distinguishing between false information and non-false information in classification.

[0073] S6: inputting the content sentiment double difference feature into a multi-layer perceptron model to output the false information true or false category of the micro-blog to be detected, wherein the false information true or false category includes false information and non-false information.

[0074] The multi-layer perceptron model (MLP, Multi-Layer Perceptron) is a classic neural network model that can learn the complex relationship between the input and the output through the linear and nonlinear combination of multiple neurons, and the false information true or false category is the result category output by the model, which is used to distinguish between false information and non-false information.

[0075] The main reasons for the rampant spread of micro-blog false information include: (1) users freely post and spread posts on the micro-blog platform, resulting in highly uncertain credibility of the information source; (2) the content of false information posts is often incendiary and concealed, and users are difficult to identify its authenticity, and are easy to believe and further spread false information; (3) the monitoring means of the micro-blog platform is relatively weak, and the punishment is low, so that the cost of spreading false information is low, thereby breeding false information.

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

[0077] In one possible implementation, S6 specifically includes:

[0078] The authenticity class of the microblog to be detected is output by the following formula:

[0079]

[0080] wherein, represents the authenticity prediction class result of the i-th microblog text to be detected, softmax represents a softmax activation function, and MLP represents a multi-layer perception model.

[0081] In the present application, the IP address, port range, domain name, URL and / or TCP packet that need to be shielded are determined according to the abnormal class in an automated manner, so that the network anomaly can be quickly and timely responded. Without manual intervention, the response time can be reduced, and the real-time performance and efficiency of the system can be improved. The target that needs to be blocked can be accurately determined according to the abnormal class, so that the spread of malicious attacks can be prevented. The source and related information of the abnormal data are shielded in time, so that the spread of attack behavior can be curbed, and the network security is protected. The corresponding bypass blocking is performed according to the abnormal class of the network transmission data, so that the network anomaly can be automatically responded, the spread of malicious attacks can be prevented, the system and data security can be protected, and the reliability of the network can be improved. At the same time, the present application also has flexibility and accuracy, and can effectively cope with various network security threats.

[0082] The technical scheme provided by the embodiment of the present application has at least the following beneficial effects:

[0083] In the present application, the data to be detected of the microblog including the microblog text to be detected and the microblog comment to be detected are acquired, so as to provide basic data for subsequent analysis. Then, the nature of the comment is further distinguished, and the content difference feature between the microblog text and the content-related comment, the local sentiment difference feature between the microblog text and the content-related comment, and the global sentiment difference feature between the microblog text and other comments are extracted. Then, the content difference feature, the local sentiment difference feature and the global sentiment difference feature are spliced to obtain the content-sentiment double difference feature. Finally, the comprehensive feature is classified by the multi-layer perception model, so as to realize the automatic identification of the false information. The present application divides the comment into the content-related comment and other comments for each source post text and the corresponding comment text, extracts the content difference feature between the microblog text and the content-related comment, and analyzes the sentiment difference feature between the microblog text and the two types of comments, so as to obtain the double difference feature. The effect of the false information detection and the quality of the information are improved, and the accuracy and robustness of the microblog false information detection are improved.

[0084] Reference is made to the accompanying drawings Figure 3 The accompanying drawings show a structure schematic diagram of a microblog false information detection system based on double difference features provided by the present application.

[0085] The application further provides a microblog false information detection system 20 based on double difference features, applied to the microblog false information detection method based on double difference features.

[0086] The processor 201.

[0087] The memory 202, wherein the memory 202 stores computer readable instructions, and the computer readable instructions are executed by the processor 201 to realize the microblog false information detection method based on double difference features.

[0088] The microblog false information detection system 20 based on double difference features can execute the microblog false information detection method based on double difference features, and achieve the same or similar technical effects. To avoid repetition, the application will not be described again.

[0089] The technical scheme provided by the application embodiment has at least the following beneficial effects:

[0090] In the application, the microblog data to be detected including the microblog text to be detected and the microblog comment to be detected is acquired to provide basic data for subsequent analysis. Then, the nature of the comment is further distinguished, and the content difference feature between the microblog text and the content-related comment, the local sentiment difference feature between the microblog text and the content-related comment, and the global sentiment difference feature between the microblog text and other comments are extracted. Then, the content difference feature, the local sentiment difference feature and the global sentiment difference feature are spliced to obtain the content-sentiment double difference feature. Finally, the comprehensive feature is classified by the multilayer perception machine model to realize the automatic identification of false information. The method is aimed at each source post text and the corresponding comment text. The comments are divided into content-related comments and other comments. The content difference feature between the microblog text and the content-related comment is extracted, and the sentiment difference feature between the microblog text and the two types of comments is analyzed to obtain the double difference feature, thereby improving the effect of false information detection and the quality of information guarantee, and improving the accuracy and robustness of microblog false information detection.

[0091] It should be appreciated that a processor in the embodiments of the present application can be a central processing unit (CPU), and can also be other general purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs) or other programmable logic devices, discrete gates or transistor logic, discrete hardware components, etc. The general purpose processor can be a microprocessor or the processor can be any conventional processor.

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

[0093] The above-described embodiments can be implemented in part or in whole through software, hardware (e.g., circuitry), firmware, or any combination thereof. When implemented in software, the above-described embodiments can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When loaded and executed by a computer, the computer instructions or computer programs can generate the flow or function according to the embodiments of the present application in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable apparatus. The 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, such as from one website site, computer, server, or data center to another website site, computer, server, or data center through a wired (e.g., infrared, wireless, microwave, etc.) manner. The computer-readable storage medium can be any available medium accessible by a computer or a data storage device such as a server, data center, etc. containing a set of one or more available media. The available media can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. The semiconductor medium can be a solid-state disk.

[0094] It should be understood that the term "and / or" used herein is merely an association relationship between the associated objects, which means that there can be three relationships, for example, A and / or B can mean that A exists alone, A and B exist together, and B exists alone, where A and B can be singular or plural. In addition, the character " / " herein generally represents an "or" relationship between the associated objects before and after it, but it can also represent an "and / or" relationship, which can be understood in the context before and after it.

[0095] In the present application, "at least one" means one or more, and "multiple" means two or more. "At least one of the following" or the like means any combination of the items, including any combination of single or multiple items. For example, at least one of a, b, or c can mean a, b, c, a-b, a-c, b-c, or a-b-c, where a, b, and c can be single or multiple.

[0096] It should be understood that in various embodiments of the present application, the size of the sequence number of the above-described processes does not mean the order of execution, and the execution order of the processes should be determined by their functions and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

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

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

[0099] In several embodiments provided by the present application, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely schematic, for example, the division of units is only a logical function division, and actual implementation can have another division manner, for example, multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

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

[0101] In addition, each functional unit in each embodiment of the present application can be integrated into a processing unit, or each unit can exist physically independently, or two or more units can be integrated into one unit.

[0102] If the functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application essentially or the parts that contribute to the prior art or parts of the technical solutions can be embodied in the form of a software product, which is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various storage media that can store program codes.

[0103] The embodiment of the present application provides a computer readable storage medium, which stores a computer program, and the program is executed by a processor to realize the microblog false information detection method based on double difference features of the method embodiment.

[0104] The computer readable storage medium provided by the present application can realize the steps and effects of the microblog false information detection method based on double difference features of the above-mentioned method embodiment. To avoid repetition, the present application will not be described again.

[0105] The technical solutions provided by the embodiments of the present application have at least the following beneficial effects:

[0106] In the present application, by acquiring the to-be-detected microblog data including the to-be-detected microblog text and the to-be-detected microblog comment, the basic data for subsequent analysis is provided, then the nature of the comment is further distinguished, and the content difference feature between the microblog text and the content-related comment, the local sentiment difference feature between the microblog text and the content-related comment, and the global sentiment difference feature between the microblog text and other comments are extracted, then the content difference feature, the local sentiment difference feature and the global sentiment difference feature are spliced to obtain the content-sentiment double difference feature, finally, the comprehensive feature is classified through the multilayer perception machine model to realize the automatic identification of false information. The method is aimed at each source post text and its corresponding comment text, by dividing the comment into content-related comment and other comment, extracting the content difference feature between the microblog text and the content-related comment, and analyzing the sentiment difference feature between the microblog text and the two types of comments, the double difference feature is obtained, the effect of false information detection and the quality of information are improved, and the accuracy and robustness of microblog false information detection are improved.

[0107] The above merely illustrates the specific embodiments of the present application, and the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical scope disclosed by the present application, which should be covered by the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0108] The following points need to be explained:

[0109] (1) The attached drawings of the embodiments of the present application only involve the structures involved in the embodiments of the present application, and other structures can refer to the general design.

[0110] (2) In the drawings used for describing the embodiments of the present application, the thickness of a layer or region is exaggerated or reduced for clarity, that is, the drawings are not drawn according to the actual proportion. It can be understood that when an element such as a layer, a film, a region or a substrate is referred to as being located "on" or "under" another element, the element can be "directly" located on or under another element or there can be an intermediate element.

[0111] (3) In the case of no conflict, the embodiments of the present application and the features in the embodiments can be combined with each other to obtain new embodiments.

[0112] The above merely illustrates the specific embodiments of the present application, and the protection scope of the present application is not limited thereto, and the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A microblog false information detection method based on double difference features, characterized in that, The method comprises the following steps: S1: obtaining to-be-detected microblog data, wherein the to-be-detected microblog data comprises to-be-detected microblog text and to-be-detected microblog comments; S2: dividing the to-be-detected microblog comments into content-related comments and other comments; The S2 is specifically: S201: extracting a theme keyword of the to-be-detected microblog text, traversing the to-be-detected microblog comments, judging whether the theme keyword exists in the to-be-detected microblog comments, if the theme keyword exists, dividing the to-be-detected microblog comments into content-related comments, otherwise, dividing the to-be-detected microblog comments into other comments; S3: extracting a content difference feature between the to-be-detected microblog text and the content-related comments; The calculation formula of the content difference feature is specifically: ； in, SD i Indicates the first i The text of the Weibo post to be detected S i Content-related differences between comments Sem ( ) indicates that content features are extracted based on the BERT pre-trained model. This indicates the first item in the set of content-related comments. j Comments related to the content, , R This represents the set of comments related to the content of the Weibo data to be detected. The number of comments related to the content included. The summation symbol is used to represent the summation symbol. Indicates the absolute value sign; S4: respectively extracting a local sentiment difference feature between the to-be-detected microblog text and the content-related comments and a global sentiment difference feature between the to-be-detected microblog text and the other comments; The calculation formula of the local sentiment difference feature is specifically: ； wherein, LED i denotes the i th content-related comment S i local sentiment difference feature between the content-related comments, emo denotes extracting sentiment features using a sentiment lexicon, denotes the j th content-related comment in the set of content-related comments, denotes a summation symbol, denotes an absolute value symbol; The calculation formula of the global sentiment difference feature is specifically: ； wherein, represents the number of other comments in the other comment set of the micro-blog data to be detected, , O represents the number of other comments included in the other comment set of the micro-blog data to be detected, GED i represents the number of other comments in the other comment set of the micro-blog data to be detected, i represents the global sentiment difference feature between the first S i other comment, ​​​ S5: splicing the content difference feature, the local sentiment difference feature and the global sentiment difference feature to obtain a content-sentiment dual difference feature; S6: inputting the content-sentiment dual difference feature into a multi-layer perception machine model to output a real or false class of the to-be-detected microblog, wherein the real or false class of the to-be-detected microblog comprises real information and non-real information.

2. The microblog false information detection method based on double difference features according to claim 1, characterized in that, The S201 specifically comprises: S2011: extracting a plurality of theme keywords of different themes from the to-be-detected microblog text through a latent Dirichlet allocation theme model; S2012: performing keyword deduplication on the obtained theme keywords; S2013: traversing whether any theme keyword after deduplication exists in the to-be-detected microblog comments, if the theme keyword exists, dividing the to-be-detected microblog comments into content-related comments, otherwise, dividing the to-be-detected microblog comments into other comments.

3. The microblog false information detection method based on double difference features according to claim 1, characterized in that, The calculation formula of the content-sentiment dual difference feature is specifically: ； wherein, D i represents the i th piece of micro-blog text to be detected S i content sentiment double difference feature between the micro-blog text to be detected and the content related comments, SD i represents the i th piece of micro-blog text to be detected S i content difference feature between the micro-blog text to be detected and the content related comments, LED i represents the i th piece of micro-blog text to be detected S i local sentiment difference feature between the micro-blog text to be detected and the content related comments, GED i represents the i th piece of micro-blog text to be detected S i global sentiment difference feature between the micro-blog text to be detected and other comments, represents a splicing operation.​ 4. The microblog false information detection method based on double difference features according to claim 1, characterized in that, The S6 is specifically: outputting the real or false class of the to-be-detected microblog through the following formula: ； wherein, represents the first i bar false information true or false prediction category result of the micro-blog text to be detected, softmax represents softmax activation function, MLP represents a multi-layer perception model. 5.A micro-blog false information detection system based on double difference features, characterized in that, The method comprises: a processor; a memory, wherein the memory stores computer readable instructions, and the computer readable instructions are executed by the processor to implement the microblog false information detection method based on dual difference features according to any one of claims 1 to 4.

6. A computer-readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to implement the microblog false information detection method based on dual difference features according to any one of claims 1 to 4.

Images