News content extraction and display method and system

Abstract

The invention relates to the related field of news push management, and discloses a news content extraction and display method and system, which are suitable for screening and managing user side quality of Internet promotion content, and can extract and display news content by performing natural language recognition on content and title of preloaded push news. The method comprises the following steps: judging a plurality of evaluation elements corresponding to news key point contents contained in the news, further tracing element contents of the news based on the plurality of evaluation elements, and judging the pieceability and integrity of the news contents according to a plurality of traceable result information, so as to achieve the purposes of evaluating the pushed news and screening the news quality for a user. Compared with a simple preference pushing mode, the method can ensure that the news pushed to the user can have higher quality and authenticity.

Description

Technical Field

[0001] The present invention relates to the field related to news push management, and in particular to a news content extraction and display method and system. Background Art

[0002] The Internet is an important way for people to obtain and receive information. Thanks to its ability to carry large amounts of data and transmit it quickly, the Internet has become the current information transmission channel, especially for time-sensitive news information. Various media continuously generate and push a large amount of news content to users. As consumers of news content, they receive thousands of news contents of varying quality every day.

[0003] In the current network environment, there is too much meaningless junk news content, which is full of clickbait titles, exaggerated content, and even completely fabricated fake news by patchwork to gain traffic. As a result, the information ultimately transmitted is extremely limited or even completely meaningless. It not only completely wastes the viewer's time, but also causes unnecessary public opinion turmoil. Therefore, in order to ensure that people can obtain higher quality news content and maintain the network environment, it is very necessary to refine and screen the news content. Summary of the invention

[0004] The purpose of the present invention is to provide a news content extraction and display method and system to solve the problems raised in the above background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions: A news content extraction and display method, comprising: Acquire push news and preload news content, process the news content based on natural language understanding, and acquire multiple evaluation elements, wherein the evaluation elements include a time element, a location element, an event subject element, and an event result element; Performing source retrieval according to a random combination of multiple evaluation elements, obtaining multiple first source information in a time series, extracting content elements from the multiple first source information, and evaluating the source quality of the current news content based on the multiple content elements, wherein the source quality is used to characterize the non-patchwork authenticity of the current news content; The relevance quality of the content ratio of the elements in the current news content and the news title is evaluated by natural language processing, wherein the content ratio of the elements is used to represent the proportion of effective information in the news content; An evaluation level for the news content is generated based on the traceability quality result and the associated quality result, and a floating mark is placed on the pushed news according to the evaluation level, wherein the floating mark is used to represent a position-variable display layer following the news preview title.

[0006] As a further solution of the present invention: it also includes a title relevance evaluation step, specifically including: Based on natural language processing, the content of news headlines is identified to obtain multiple core expressions; Performing correlation judgment on multiple core expressions, distinguishing between subject expressions and definition expressions, and judging the correlation between multiple definition expressions and the subject expressions based on the expression content, wherein the subject expressions are used to represent the event content, and the definition expressions are used to modify and define the subject expressions; Determine the logic and causal relationship between the definition expression and the subject expression, and perform description verification of the logic and causal relationship in the news content to generate a verification result; If the verification result indicates that the definition expression word or the subject expression word does not match the news content, the news is marked as news with a deceptive title; If the verification result indicates that the definition expression is exaggerated compared to the news content or has no direct causal relationship with the main expression in the news content, the news is marked as news with exaggerated content.

[0007] As a further solution of the present invention: the step of evaluating the quality of the current news content based on the multiple content elements comprises: Classify multiple content elements of multiple first source information and the current evaluation element into the same content category, and compare the information expression meanings of the evaluation element and the multiple content elements in the same content category to obtain a tracing result; If the tracing result is characterized by the inconsistency between the evaluation element and multiple content elements in the same content category, the news content is traced and evaluated based on the proportion of the inconsistent evaluation elements. The tracing evaluation process is characterized by a proportional scoring process, and each of the pushed news has a highest basic promotion score. If the tracing result indicates that there is a content category that the currently pushed news does not contain among the multiple content elements, multiple content elements of the first source information corresponding to the content category are compared, and if they are consistent, the first source information is marked as a news content supplement source and pushed.

[0008] As a further solution of the present invention, the auxiliary association step of low-quality news is also included, which specifically includes: If the traceability quality result or the correlation quality result of the current recommended news is less than a preset value, the recommended news is classified as low-quality news; Acquire multiple evaluation elements that do not match the first source information in the traceability search, perform traceability search on the unmatched elements, acquire the first source information on multiple time series, and establish an associated link based on the first source information, wherein the associated link is used to guide the jump to the corresponding first source information link; A guidance interface is generated at the corresponding entry of the evaluation element in the recommended news, and the associated link is connected to the guidance interface, and a visual guidance layer is generated accordingly.

[0009] As a further solution of the present invention, the step of tracing the source of the image content is also included, which specifically includes: Acquire image information in news content as the visual element of the pushed news, perform source tracing search on the visual element based on image features, and acquire first source information in a time series; The image-related content of the first source information is matched. If the information content is inconsistent with multiple evaluation factors of the current news content, an unrelated picture mark is added to the current recommended news, and the unrelated picture mark is output through a floating mark.

[0010] The embodiment of the present invention aims to provide a news content extraction and display system, comprising: A news preprocessing module, used to obtain push news and preload news content, process the news content based on natural language understanding, and obtain multiple evaluation elements, wherein the evaluation elements include time elements, location elements, event subject elements, and event result elements; A source retrieval module, used to perform source retrieval according to a random combination of multiple evaluation elements, obtain multiple first source information in a time series, extract content elements from the multiple first source information, and perform source quality evaluation on the current news content based on the multiple content elements, wherein the source quality is used to characterize the non-patchwork authenticity of the current news content; A relevance evaluation module is used to evaluate the relevance quality of the element relevance content ratio of the current news content and the news title through natural language processing, wherein the element relevance content ratio is used to represent the length ratio of effective information in the news content; The graded promotion module is used to generate an evaluation level for the news content based on the traceability quality results and the associated quality results, and to place a floating mark on the pushed news according to the evaluation level, wherein the floating mark is used to represent a positionally variable display layer following the news preview title.

[0011] As a further solution of the present invention: it also includes a title association module, specifically including: A core expression acquisition unit, used to perform content recognition on news titles based on natural language processing to acquire multiple core expression words; A core association evaluation unit, used to judge the association of multiple core expressions, distinguish between subject expressions and definition expressions, and judge the association between multiple definition expressions and the subject expressions based on the expression content, wherein the subject expressions are used to represent the event content, and the definition expressions are used to modify and define the subject expressions; A logic association verification unit, used to determine the logic and causal relationship between the definition expression and the subject expression, and to perform description verification of the logic and causal relationship in the news content to generate a verification result; a title non-matching determination unit, configured to mark the news as news with a deceptive title if the verification result indicates that the definition expression word or the subject expression word does not match the news content; The title exaggeration determination unit is used to mark the news as content-exaggerated news if the verification result is characterized by the definition expression word being exaggerated compared to the news content or having no direct causal relationship with the main expression word in the news content.

[0012] As a further solution of the present invention: the traceability retrieval module includes: An element comparison unit, configured to classify the content elements of the plurality of first source information and the current evaluation element into the same category, and compare the information expression meanings of the evaluation element and the plurality of content elements in the same content category to obtain a tracing result; An element judgment unit, for, if the tracing result is characterized as an inconsistency between the evaluation element and multiple content elements in the same content category, then tracing the evaluation of the news content based on the proportion of the inconsistent evaluation elements, the tracing evaluation process is characterized as a proportional scoring process, and each of the pushed news has a highest basic promotion score; The element extension unit is used to compare the multiple content elements of the first source information corresponding to the content category if the tracing result is characterized by the fact that the currently pushed news does not contain a content category among the multiple content elements. If they are consistent, the first source information is marked as a news content supplement source and pushed.

[0013] As a further solution of the present invention: it also includes a low-quality auxiliary association module, specifically including: A low-quality demarcation unit, configured to classify the recommended news as low-quality news if the traceability quality result or the association quality result of the currently recommended news is less than a preset value; An association guiding unit, used to obtain multiple evaluation elements that do not match the first source information in the traceability search, and perform traceability search on the unmatched elements to obtain the first source information on multiple time series, and establish an association link based on the first source information, wherein the association link is used to guide the jump to the corresponding first source information link; The guidance push unit is used to generate a guidance interface at the corresponding entry of the evaluation element in the recommended news, connect the associated link to the guidance interface, and generate a visual guidance layer accordingly.

[0014] As a further solution of the present invention: it also includes an image tracing module, specifically including: A visual tracing unit, used to obtain image information in news content as a visual element of the pushed news, perform a tracing search on the visual element based on image features, and obtain first source information in a time series; The visual judgment unit is used to match the image-related content of the first source information. If the information content is inconsistent with multiple evaluation factors of the current news content, the current recommended news is marked with an unrelated picture, and the unrelated picture mark is output through a floating mark.

[0015] Compared with the prior art, the beneficial effects of the present invention are: it is suitable for user-side quality screening management of Internet promotion content, by performing natural language recognition on the content and title of pre-loaded push news, judging the evaluation factors of multiple corresponding news key points contained therein, and then tracing the source of the element content of the news based on the multiple evaluation factors, so as to judge the patchwork and integrity of the news content according to the multiple result information of tracing, so as to achieve the purpose of evaluating the pushed news and screening the news quality for users. Compared with the simple preference push mode, it can ensure that the pushed news to users can have higher quality and authenticity. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 The flowchart is a method for extracting and displaying news content.

[0017] Figure 2 The present invention is a flowchart of the title relevance evaluation step in a news content extraction and display method.

[0018] Figure 3 This is a block diagram of a news content extraction and display system. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0020] The specific implementation of the present invention is described in detail below in conjunction with specific embodiments.

[0021] like Figure 1 The present invention provides a method for extracting and displaying news content in accordance with an embodiment of the present invention, comprising the following steps: S10, acquiring push news and preloading news content, processing the news content based on natural language understanding, and acquiring multiple evaluation elements, wherein the evaluation elements include a time element, a location element, an event subject element, and an event result element; S20, performing source retrieval according to a random combination of multiple evaluation elements, obtaining multiple first source information in a time series, extracting content elements from the multiple first source information, and performing source tracing quality evaluation on the current news content based on the multiple content elements, wherein the source tracing quality is used to characterize the non-patchwork authenticity of the current news content; S30, performing association quality assessment on the element-related content ratio of the current news content and the news title through natural language processing, wherein the element-related content ratio is used to represent the length ratio of effective information in the news content; S40, generating an evaluation level for the news content based on the traceability quality result and the associated quality result, and placing a floating mark on the pushed news according to the evaluation level, wherein the floating mark is used to represent a position variable display layer following the news preview title.

[0022] In this embodiment, a news content extraction and display method is provided, which is suitable for user-side quality screening management of Internet promotion content. By performing natural language recognition on the content and title of pre-loaded push news, multiple evaluation factors corresponding to the news key points contained therein are judged, and then the news element content is traced based on the multiple evaluation factors, so as to judge the patchwork and integrity of the news content according to the multiple result information of tracing, so as to achieve the purpose of evaluating the push news and screening the news quality for users. Compared with the simple preference push mode, it can ensure that the push news to users can have higher quality and authenticity; the Internet is an important way for people to obtain and receive information, and benefit Due to the characteristics of the Internet that can carry a large amount of data and transmit it quickly, the Internet has become the current information transmission channel, especially for timely news information. Various media continuously generate and push a large amount of news content to users. As consumers of news content, they receive thousands of news contents of varying quality every day. Therefore, in order to ensure that consumers can click to watch the pushed content more frequently, it is also very important to screen and target the pushed content. The promotion screening methods in the prior art mainly include two directions. One is to screen and push based on popularity, which selects and pushes according to the popularity information of different contents on the Internet, and gives priority to pushing the popularity information. The other is to screen and push based on popularity. According to the preference filtering push, that is, according to the consumer's historical browsing preference matching related content to push; however, in the current filtering promotion mode, there are a lot of low-quality or even false news content in the pushed content, which affects the consumer's evaluation of real events and wastes a lot of consumer's time, including but not limited to various false types such as content patchwork, excessive exaggeration and content inconsistency; therefore, this embodiment introduces natural language understanding (NLP) technology. When the platform pushes news to consumers, the content of multiple pushed news is preloaded in the background, and the news content and title are identified through natural language understanding, and summarized into multiple evaluation factors according to the type of content elements (for example, For example, what happened at what time and place, and what were the consequences in the end. The multiple "whats" here can correspond to different evaluation factors, and the "whats" can be further divided into multiple factors such as the start, inducement, and process). These factors are randomly combined and searched (for example, evaluation factor A + evaluation factor C) to trace the source and find earlier matching news content. If these matching news contents are consistent, it means that the current news has factual basis. If the results of tracing back multiple factors correspond to factors of multiple different events, it means that the current content may be inaccurate and there is no point in browsing it. Such pushed content does not need to be pushed to consumers.Moreover, some media currently prefer to use exaggerated and outrageous titles that are even irrelevant to the content to attract consumers to click on their content. Such pushed news is also not worth pushing to consumers. Therefore, the conformity of the title can be judged by matching the key words and descriptive words of the title with the main content of the pushed news. The floating mark here is a newly established regional display layer, and the coordinates of the display area are followed through image recognition to achieve following the corresponding news display area when the page is turned. ;

[0023] like Figure 2 As shown, as another preferred embodiment of the present invention, it also includes a title relevance evaluation step, specifically including: S51, performing content recognition on news headlines based on natural language processing to obtain multiple core expression words; S52, performing correlation judgment on the multiple core expressions, distinguishing the subject expression words from the definition expression words, and judging the correlation relationship between the multiple definition expression words and the subject expression words based on the expression content, wherein the subject expression words are used to represent the event content, and the definition expression words are used to modify and define the subject expression words; S53, determining the logic and causal relationship between the definition expression and the subject expression, and performing description verification of the logic and causal relationship in the news content to generate a verification result; S54, if the verification result indicates that the definition expression word or the subject expression word does not match the news content, marking the news as news with a deceptive title; S55: If the verification result indicates that the definition expression is exaggerated compared to the news content or has no direct causal relationship with the main expression in the news content, the news is marked as news with exaggerated content.

[0024] In this embodiment, the evaluation steps of the title relevance are supplemented. First, the title is identified through natural language processing (NLP) technology to determine multiple core expressions (i.e., element words) in the title, including content expressions and modifying expressions, and logical association judgments are made on these core expressions (for example, the degree to which a certain modifying word acts on the expression of a core keyword of a certain content expression). The content of the pushed news is matched according to these core expressions and the associations between the core expressions. If it does not match the content, association, or degree of modification of the core expressions in the news content, it indicates that the current title is falsely misleading or overly exaggerated, and the news does not make sense to be pushed to consumers for browsing.

[0025] As another preferred embodiment of the present invention, the step of evaluating the source quality of the current news content based on the multiple content elements includes: Classify multiple content elements of multiple first source information and the current evaluation element into the same content category, and compare the information expression meanings of the evaluation element and the multiple content elements in the same content category to obtain a tracing result; If the tracing result is characterized by the inconsistency between the evaluation element and multiple content elements in the same content category, the news content is traced and evaluated based on the proportion of the inconsistent evaluation elements. The tracing evaluation process is characterized by a proportional scoring process, and each of the pushed news has a highest basic promotion score. If the tracing result indicates that there is a content category that the currently pushed news does not contain among the multiple content elements, multiple content elements of the first source information corresponding to the content category are compared, and if they are consistent, the first source information is marked as a news content supplement source and pushed.

[0026] In this embodiment, in different results retrieved by different combinations of factors, there may be multiple news that are partially overlapping but the factors are not completely consistent and can be complementary. It should be noted that the inconsistency of factors here means that the content obtained by tracing back is basically consistent with the current content, but there are different content factors. This content factor expresses the key points of different events, rather than being used to express the key points of the same event but with different content. Therefore, the part of the traced content can be regarded as the relevant follow-up of the news and other content not explained in the current news. By tracing back the source of the content and supplementing it, it can help consumers to have a more complete understanding of the overall event content and avoid news writers from guiding viewers' emotions and positions by maliciously deleting content.

[0027] As another preferred embodiment of the present invention, the auxiliary association step of low-quality news is also included, which specifically includes: If the traceability quality result or the correlation quality result of the current recommended news is less than a preset value, the recommended news is classified as low-quality news; Acquire multiple evaluation elements that do not match the first source information in the traceability search, perform traceability search on the unmatched elements, acquire the first source information on multiple time series, and establish an associated link based on the first source information, wherein the associated link is used to guide the jump to the corresponding first source information link; A guidance interface is generated at the corresponding entry of the evaluation element in the recommended news, and the associated link is connected to the guidance interface, and a visual guidance layer is generated accordingly.

[0028] In this embodiment, an auxiliary association step of low-quality news is introduced. When it is determined through tracing that the current news is fabricated and pieced together as low-quality news, multiple different original news used for its piecemeal can be found based on the tracing results of multiple evaluation factors, and these original news are associated with the keywords of their pieced-together parts and displayed in the current push. If the current consumer has a great interest in some of the key contents, the associated first source information link can be used to quickly jump to the corresponding pieced-together news source, so that he can quickly obtain the content of interest.

[0029] As another preferred embodiment of the present invention, the step of tracing the source of the image content is also included, which specifically includes: Acquire image information in news content as the visual element of the pushed news, perform source tracing search on the visual element based on image features, and acquire first source information in a time series; The image-related content of the first source information is matched. If the information content is inconsistent with multiple evaluation factors of the current news content, an unrelated picture mark is added to the current recommended news, and the unrelated picture mark is output through a floating mark.

[0030] In this embodiment, in addition to the traditional text information expression, the news content also includes image reference content. Through the image content, one can intuitively and quickly understand the status and situation of the scene corresponding to the news content. Therefore, the image content is also a very important part of the news information. However, there are currently some editors who randomly quote images of irrelevant events, which causes news consumers to obtain incorrect information and even some serious misguidance of public opinion. Therefore, it is also very important to trace the source of the image. The conformity of the image is judged by tracing the source, and the rationality of the quoted image is judged based on the understanding and recognition of the image content and the news content. The news is evaluated and divided according to the rationality of the result.

[0031] like Figure 3 As shown, the present invention also provides a news content extraction and display system, which includes: The news preprocessing module 100 is used to obtain the pushed news and preload the news content, process the news content based on natural language understanding, and obtain multiple evaluation elements, wherein the evaluation elements include time elements, location elements, event subject elements, and event result elements; The source retrieval module 200 is used to perform source retrieval according to a random combination of multiple evaluation elements, obtain multiple first source information in a time series, extract content elements from the multiple first source information, and perform source quality evaluation on the current news content based on the multiple content elements, wherein the source quality is used to characterize the non-patchwork authenticity of the current news content; The association evaluation module 300 is used to evaluate the association quality of the element-related content ratio and the news title of the current news content through natural language processing, wherein the element-related content ratio is used to represent the length ratio of effective information in the news content; The graded promotion module 400 is used to generate an evaluation level for the news content based on the traceability quality results and the associated quality results, and to place a floating mark on the pushed news according to the evaluation level, wherein the floating mark is used to represent a positionally variable display layer following the news preview title.

[0032] As another preferred embodiment of the present invention, it also includes a title association module, specifically including: A core expression acquisition unit, used to perform content recognition on news titles based on natural language processing to acquire multiple core expression words; A core association evaluation unit, used to judge the association of multiple core expressions, distinguish between subject expressions and definition expressions, and judge the association between multiple definition expressions and the subject expressions based on the expression content, wherein the subject expressions are used to represent the event content, and the definition expressions are used to modify and define the subject expressions; A logic association verification unit, used to determine the logic and causal relationship between the definition expression and the subject expression, and to perform description verification of the logic and causal relationship in the news content to generate a verification result; a title non-matching determination unit, configured to mark the news as news with a deceptive title if the verification result indicates that the definition expression word or the subject expression word does not match the news content; The title exaggeration determination unit is used to mark the news as content-exaggerated news if the verification result is characterized by the definition expression word being exaggerated compared to the news content or having no direct causal relationship with the main expression word in the news content.

[0033] As another preferred embodiment of the present invention, the traceability retrieval module includes: An element comparison unit, configured to classify the content elements of the plurality of first source information and the current evaluation element into the same category, and compare the information expression meanings of the evaluation element and the plurality of content elements in the same content category to obtain a tracing result; An element judgment unit, for, if the tracing result is characterized as an inconsistency between the evaluation element and multiple content elements in the same content category, then tracing the evaluation of the news content based on the proportion of the inconsistent evaluation elements, the tracing evaluation process is characterized as a proportional scoring process, and each of the pushed news has a highest basic promotion score; The element extension unit is used to compare the multiple content elements of the first source information corresponding to the content category if the tracing result is characterized by the fact that the currently pushed news does not contain a content category among the multiple content elements. If they are consistent, the first source information is marked as a news content supplement source and pushed.

[0034] As another preferred embodiment of the present invention, a low-quality auxiliary association module is also included, specifically including: A low-quality demarcation unit, configured to classify the recommended news as low-quality news if the traceability quality result or the association quality result of the currently recommended news is less than a preset value; An association guiding unit, used to obtain multiple evaluation elements that do not match the first source information in the traceability search, and perform traceability search on the unmatched elements to obtain the first source information on multiple time series, and establish an association link based on the first source information, wherein the association link is used to guide the jump to the corresponding first source information link; The guidance push unit is used to generate a guidance interface at the corresponding entry of the evaluation element in the recommended news, connect the associated link to the guidance interface, and generate a visual guidance layer accordingly.

[0035] As another preferred embodiment of the present invention, it also includes an image tracing module, which specifically includes: A visual tracing unit, used to obtain image information in news content as a visual element of the pushed news, perform a tracing search on the visual element based on image features, and obtain first source information in a time series; The visual judgment unit is used to match the image-related content of the first source information. If the information content is inconsistent with multiple evaluation factors of the current news content, the current recommended news is marked with an unrelated picture, and the unrelated picture mark is output through a floating mark.

[0036] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0037] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the disclosure in the specification and examples. This application is intended to cover any variations, uses or adaptations of the present disclosure, which follow the general principles of the present disclosure and include common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present disclosure are indicated by the claims.

[0038] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A news content extraction and display method, characterized in that: Include: Acquire push news and preload news content, process the news content based on natural language understanding, and acquire multiple evaluation elements, wherein the evaluation elements include a time element, a location element, an event subject element, and an event result element; Performing source retrieval according to a random combination of multiple evaluation elements, obtaining multiple first source information in a time series, extracting content elements from the multiple first source information, and evaluating the source quality of the current news content based on the multiple content elements, wherein the source quality is used to characterize the non-patchwork authenticity of the current news content; The relevance quality of the content ratio of the elements in the current news content and the news title is evaluated by natural language processing, wherein the content ratio of the elements is used to represent the proportion of effective information in the news content; An evaluation level for the news content is generated based on the traceability quality result and the associated quality result, and a floating mark is placed on the pushed news according to the evaluation level, wherein the floating mark is used to represent a position-variable display layer following the news preview title.

2. A news content extraction and display method according to claim 1, characterized in that: It also includes the title relevance evaluation step, including: Based on natural language processing, the content of news headlines is identified to obtain multiple core expressions; Performing correlation judgment on multiple core expressions, distinguishing between subject expressions and definition expressions, and judging the correlation between multiple definition expressions and the subject expressions based on the expression content, wherein the subject expressions are used to represent the event content, and the definition expressions are used to modify and define the subject expressions; Determine the logic and causal relationship between the definition expression and the subject expression, and perform description verification of the logic and causal relationship in the news content to generate a verification result; If the verification result indicates that the definition expression word or the subject expression word does not match the news content, the news is marked as news with a deceptive title; If the verification result indicates that the definition expression is exaggerated compared to the news content or has no direct causal relationship with the main expression in the news content, the news is marked as news with exaggerated content.

3. A news content extraction and display method according to claim 2, characterized in that: The step of evaluating the source quality of the current news content based on the multiple content elements comprises: Classify multiple content elements of multiple first source information and the current evaluation element into the same content category, and compare the information expression meanings of the evaluation element and the multiple content elements in the same content category to obtain a tracing result; If the tracing result is characterized by the inconsistency between the evaluation element and multiple content elements in the same content category, the news content is traced and evaluated based on the proportion of the inconsistent evaluation elements. The tracing evaluation process is characterized by a proportional scoring process, and each of the pushed news has a highest basic promotion score. If the tracing result indicates that there is a content category that the currently pushed news does not contain among the multiple content elements, multiple content elements of the first source information corresponding to the content category are compared, and if they are consistent, the first source information is marked as a news content supplement source and pushed.

4. A news content extraction and display method according to claim 3, characterized in that: It also includes auxiliary association steps for low-quality news, including: If the traceability quality result or the correlation quality result of the current recommended news is less than a preset value, the recommended news is classified as low-quality news; Acquire multiple evaluation elements that do not match the first source information in the traceability search, perform traceability search on the unmatched elements, acquire the first source information on multiple time series, and establish an associated link based on the first source information, wherein the associated link is used to guide the jump to the corresponding first source information link; A guidance interface is generated at the corresponding entry of the evaluation element in the recommended news, and the associated link is connected to the guidance interface, and a visual guidance layer is generated accordingly.

5. A news content extraction and display method according to claim 1, characterized in that: It also includes the steps of tracing the source of the image content, including: Acquire image information in news content as the visual element of the pushed news, perform source tracing search on the visual element based on image features, and acquire first source information in a time series; The image-related content of the first source information is matched. If the information content is inconsistent with multiple evaluation factors of the current news content, an unrelated picture mark is added to the current recommended news, and the unrelated picture mark is output through a floating mark.

6. A news content extraction and display system, characterized in that: Include: A news preprocessing module, used to obtain push news and preload news content, process the news content based on natural language understanding, and obtain multiple evaluation elements, wherein the evaluation elements include time elements, location elements, event subject elements, and event result elements; A source retrieval module, used to perform source retrieval according to a random combination of multiple evaluation elements, obtain multiple first source information in a time series, extract content elements from the multiple first source information, and perform source quality evaluation on the current news content based on the multiple content elements, wherein the source quality is used to characterize the non-patchwork authenticity of the current news content; A relevance evaluation module is used to evaluate the relevance quality of the element relevance content ratio of the current news content and the news title through natural language processing, wherein the element relevance content ratio is used to represent the length ratio of effective information in the news content; The graded promotion module is used to generate an evaluation level for the news content based on the traceability quality results and the associated quality results, and to place a floating mark on the pushed news according to the evaluation level, wherein the floating mark is used to represent a positionally variable display layer following the news preview title.

7. A news content extraction and display system according to claim 6, characterized in that: It also includes title association modules, including: A core expression acquisition unit, used to perform content recognition on news titles based on natural language processing to acquire multiple core expression words; A core association evaluation unit, used to judge the association of multiple core expressions, distinguish between subject expressions and definition expressions, and judge the association between multiple definition expressions and the subject expressions based on the expression content, wherein the subject expressions are used to represent the event content, and the definition expressions are used to modify and define the subject expressions; A logic association verification unit, used to determine the logic and causal relationship between the definition expression and the subject expression, and to perform description verification of the logic and causal relationship in the news content to generate a verification result; a title non-matching determination unit, configured to mark the news as news with a deceptive title if the verification result indicates that the definition expression word or the subject expression word does not match the news content; The title exaggeration determination unit is used to mark the news as content-exaggerated news if the verification result is characterized by the definition expression word being exaggerated compared to the news content or having no direct causal relationship with the main expression word in the news content.

8. A news content extraction and display system according to claim 7, characterized in that: The traceability retrieval module includes: An element comparison unit, configured to classify the content elements of the plurality of first source information and the current evaluation element into the same category, and compare the information expression meanings of the evaluation element and the plurality of content elements in the same content category to obtain a tracing result; An element judgment unit, for, if the tracing result is characterized as an inconsistency between the evaluation element and multiple content elements in the same content category, then tracing the evaluation of the news content based on the proportion of the inconsistent evaluation elements, the tracing evaluation process is characterized as a proportional scoring process, and each of the pushed news has a highest basic promotion score; The element extension unit is used to compare the multiple content elements of the first source information corresponding to the content category if the tracing result is characterized by the fact that the currently pushed news does not contain a content category among the multiple content elements. If they are consistent, the first source information is marked as a news content supplement source and pushed.

9. A news content extraction and display system according to claim 8, characterized in that: It also includes low-quality auxiliary association modules, including: A low-quality demarcation unit, configured to classify the recommended news as low-quality news if the traceability quality result or the association quality result of the currently recommended news is less than a preset value; An association guiding unit, used to obtain multiple evaluation elements that do not match the first source information in the traceability search, and perform traceability search on the unmatched elements to obtain the first source information on multiple time series, and establish an association link based on the first source information, wherein the association link is used to guide the jump to the corresponding first source information link; The guidance push unit is used to generate a guidance interface at the corresponding entry of the evaluation element in the recommended news, connect the associated link to the guidance interface, and generate a visual guidance layer accordingly.

10. A news content extraction and display system according to claim 6, characterized in that: It also includes an image tracing module, which includes: A visual tracing unit, used to obtain image information in news content as a visual element of the pushed news, perform a tracing search on the visual element based on image features, and obtain first source information in a time series; The visual judgment unit is used to match the image-related content of the first source information. If the information content is inconsistent with multiple evaluation factors of the current news content, the current recommended news is marked with an unrelated picture, and the unrelated picture mark is output through a floating mark.

Images