Text sharing compliance detection method and device, electronic equipment and program product

Abstract

The invention relates to the technical field of artificial intelligence, and provides a text sharing compliance detection method and device, electronic equipment and a computer program product. The method comprises the steps of obtaining a to-be-detected knowledge sharing text; inputting a first preset cue word into the large language model to indicate the large language model to execute quality evaluation on the multi-dimensional factors of the knowledge sharing text according to a built-in text quality evaluation rule, and outputting a result indicating whether the quality of the knowledge sharing text reaches the standard or not; detecting whether the knowledge sharing text has a plagiarism risk or not; and if the quality of the knowledge sharing text reaches the standard and the knowledge sharing text has no plagiarism risk, determining that the compliance detection of the knowledge sharing text passes. By adopting the method, compliance detection can be performed on the published knowledge sharing text, so that the fairness of reward issuing in an enterprise knowledge sharing scene is improved.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and in particular to a method, apparatus, electronic device, and computer program product for detecting text sharing compliance. Background Technology

[0002] In enterprise knowledge-sharing scenarios, employees can share reading notes, technical experiences, and other knowledge-sharing texts to earn rewards such as monetary rewards or points from the company. However, some employees, in order to obtain rewards, may publish low-quality or plagiarized knowledge-sharing texts that are clearly non-compliant, which seriously affects the fairness of reward distribution. Therefore, how to conduct compliance checks on published knowledge-sharing texts becomes a technical issue that those skilled in the art need to consider. Summary of the Invention

[0003] In view of this, embodiments of this application provide a method, apparatus, electronic device, and computer program product for detecting the compliance of published knowledge-sharing text, thereby improving the fairness of reward distribution.

[0004] The first aspect of this application provides a method for detecting text sharing compliance, including: Obtain the knowledge-sharing text to be tested; Input the first preset prompt word into the large language model to instruct the large language model to perform a quality assessment on the multi-dimensional factors of the knowledge sharing text according to the built-in text quality assessment rules, and output the result of whether the quality of the knowledge sharing text meets the standard. Detect whether knowledge-sharing texts pose a risk of plagiarism; If the quality of the knowledge-sharing text meets the standards and there is no risk of plagiarism, then the compliance test for the knowledge-sharing text is deemed passed.

[0005] In the technical solution of this application embodiment, the knowledge-sharing text to be detected is first obtained, and then a first preset prompt word is input into the large language model. After receiving the first preset prompt word, the large language model performs a quality assessment of the multi-dimensional factors of the knowledge-sharing text according to its built-in text quality assessment rules, and outputs the result of whether the quality of the knowledge-sharing text meets the standards. In addition, it also detects whether the knowledge-sharing text has a risk of plagiarism. If the quality of the knowledge-sharing text meets the standards and there is no risk of plagiarism, then the compliance test of the knowledge-sharing text is determined to be passed. By setting it up in this way, it is possible to effectively identify non-compliant situations such as low quality or plagiarism in knowledge-sharing texts, that is, to achieve compliance testing of knowledge-sharing texts, thereby improving the fairness of reward distribution in enterprise knowledge-sharing scenarios.

[0006] In one implementation of this application, detecting whether knowledge-sharing text has a plagiarism risk includes: By searching the historical shared text database, we can detect whether there is a risk of historical plagiarism in knowledge-sharing texts; Use online searches to detect whether knowledge-sharing texts are at risk of plagiarism. In cases where there is a risk of historical plagiarism and / or online plagiarism in the knowledge-sharing text, it is determined that the knowledge-sharing text is at risk of plagiarism.

[0007] Considering that the main sources of plagiarism in text content include historical shared texts and the internet, plagiarism risk detection can be divided into two parts: historical plagiarism risk detection and online plagiarism risk detection. By combining historical plagiarism risk detection and online plagiarism risk detection, it is possible to more comprehensively and accurately identify whether currently published knowledge-sharing texts pose a plagiarism risk.

[0008] In one implementation of this application, the detection of historical plagiarism risk in knowledge-sharing texts is achieved by searching a historical shared text library, including: Convert the knowledge-sharing text into the corresponding target text vector; The similarity of the target text vector with each historical text vector is compared. Based on the results of the similarity comparison, it is determined whether the knowledge-sharing text has a risk of historical plagiarism. Each historical text vector corresponds one-to-one with each historical shared text contained in the historical shared text library.

[0009] By using text vector similarity comparison, it is possible to more accurately identify whether there are historical shared texts in the historical shared text library that are the same as or similar to the currently published knowledge-sharing text, thereby more accurately determining whether the knowledge-sharing text has a historical plagiarism risk and effectively completing the historical plagiarism risk detection.

[0010] In one implementation of this application, the target text vector is compared with each historical text vector for similarity, and the risk of historical plagiarism in the knowledge-sharing text is determined based on the similarity comparison results, including: Calculate the inner product between the target text vector and each historical text vector; From all the historical text vectors, select the K historical text vectors with the largest inner product, where K ≥ 1; If, among K historical text vectors, there exists a similar historical text vector whose cosine similarity to the target text vector is greater than a first threshold, the knowledge-sharing text is determined to have a historical plagiarism risk; or, if, among K historical text vectors, there exists a similar historical text vector whose edit distance similarity to the target text vector is greater than a second threshold, the knowledge-sharing text is determined to have a historical plagiarism risk.

[0011] When comparing the similarity of the target text vector with each historical text vector, the inner product index technique can be used to conveniently and accurately achieve vector similarity retrieval and find potential similar historical text vectors.

[0012] In one implementation of this application, after determining whether the knowledge-sharing text has a historical plagiarism risk based on the similarity comparison result, the method further includes: If the knowledge-sharing text has a history of plagiarism risk, then similar historical text vectors that have a similar relationship with the target text vector are identified from each historical text vector. Identify similar historical shared texts corresponding to similar historical text vectors; The types of historical plagiarism risks are determined based on the authors and publication dates of similar historical shared texts, as well as the authors and publication dates of knowledge-sharing texts.

[0013] After determining that a knowledge-sharing text has a history of plagiarism risk, the type of this risk can be further determined based on information such as the text's author and publication date, thereby providing users with richer risk-related information.

[0014] In one implementation of this application, the type of historical plagiarism risk is determined based on the author and publication time of similar historical shared texts, as well as the author and publication time of knowledge-sharing texts, including: If the authors of similar historical shared texts are different from the authors of knowledge-sharing texts, then the type of historical plagiarism risk is determined to be duplicate. If the author of a similar historical shared text is the same as the author of a knowledge-sharing text, and the publication time of the similar historical shared text is within the same time period as the publication time of the knowledge-sharing text, then the type of historical plagiarism risk is determined to be multiple publications. If the authors of similar historical shared texts are the same as those of knowledge-sharing texts, and the publication times of similar historical shared texts and knowledge-sharing texts are not within the same time period, then the type of historical plagiarism risk is determined to be duplicate.

[0015] This setup allows for the identification of historical plagiarism risks based on the author and publication time of the shared text, thereby providing users with richer and more comprehensive information on historical plagiarism risks and enhancing the comprehensiveness and practicality of historical plagiarism risk detection.

[0016] In one implementation of this application, detecting the risk of online plagiarism in knowledge-sharing text through web search includes: Extract M sentences from the knowledge-sharing text, where M≥1; Using a web search interface, crawl the web to retrieve the N text search results with the highest similarity to the M sentences, where N≥1; Input a second preset prompt word into the large language model to instruct it to output a result indicating whether the knowledge-sharing text is at risk of online plagiarism, based on the similarity between M sentences and N text search results.

[0017] By introducing a large language model, we can compare the similarity of M sentences extracted from knowledge-sharing text with N text search results from the web, and thus determine more accurately whether the knowledge-sharing text is at risk of online plagiarism.

[0018] In one implementation of this application, M sentences are extracted from knowledge-sharing text, including: Find the target content in the author's reflections section of the knowledge-sharing text; Extract M sentences from the target content that have more than the third threshold number of words.

[0019] By extracting a certain number of long sentences from the author's reflections in a knowledge-sharing text and comparing their similarity with text search results obtained through online searches, it is possible to detect the risk of online plagiarism in the knowledge-sharing text with relatively high accuracy.

[0020] In another implementation of this application's embodiments, M sentences are extracted from the knowledge-sharing text, including: Input a third preset prompt word into the large language model to instruct it to extract M sentences from the knowledge-sharing text that have more than the third threshold number of words.

[0021] By using large language models, it is possible to conveniently and accurately extract the sentences needed to perform online plagiarism risk detection from knowledge-sharing texts.

[0022] In one implementation of this application, the quantities of M and N are set in the following manner: Determine the number of times the author of the knowledge-sharing text has plagiarized in the past, and the number of target historical shared texts published by the author of the knowledge-sharing text; The quantities of M and N are set based on the number of historical plagiarisms and the number of target historical shared texts; where the quantities of M and N are positively correlated with the number of historical plagiarisms and negatively correlated with the number of target historical shared texts.

[0023] By reasonably setting the values ​​of M and N based on the author's historical plagiarism count and the number of published and shared texts, the detection intensity of online plagiarism risks can be flexibly adjusted. On the one hand, this can reduce the consumption of system computing power and time, and on the other hand, there is no need to worry about missing online plagiarism.

[0024] In one implementation of this application, the method further includes: If the quality of the knowledge-sharing text does not meet the standards, or if the knowledge-sharing text poses a risk of plagiarism, then the compliance test for the knowledge-sharing text is deemed to have failed.

[0025] If the quality of the currently published knowledge-sharing text does not meet the standards, or if the knowledge-sharing text has a risk of plagiarism, it means that the knowledge-sharing text is a low-quality or plagiarized non-compliant text, and therefore it can be determined that its compliance test has failed.

[0026] In one implementation of this application, the method further includes: When the compliance check of the knowledge-sharing text passes, add a rewardable tag to the knowledge-sharing text; When a knowledge-sharing text fails the compliance check, a "non-rewardable" tag is added to it.

[0027] This setup effectively filters out low-quality or plagiarized non-compliant shared texts, reduces manual review costs, and greatly enhances the fairness and reasonableness of reward distribution in enterprise knowledge-sharing scenarios.

[0028] A second aspect of this application provides a text sharing compliance detection device, including: The shared text acquisition module is used to acquire the knowledge-sharing text to be detected; The text quality assessment module is used to input the first preset prompt word into the large language model, so as to instruct the large language model to perform quality assessment on the multi-dimensional factors of the knowledge sharing text according to the built-in text quality assessment rules, and output the result of whether the quality of the knowledge sharing text meets the standard. The plagiarism risk detection module is used to detect whether knowledge-sharing texts are at risk of plagiarism. The compliance assessment module is used to determine if the quality of the knowledge-sharing text meets the standards and there is no risk of plagiarism, thus confirming that the compliance test of the knowledge-sharing text has passed.

[0029] A third aspect of this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the text sharing compliance detection method provided in the first aspect of this application.

[0030] A fourth aspect of this application provides a computer program product that, when run on an electronic device, causes the electronic device to execute the text sharing compliance detection method provided in the first aspect of this application.

[0031] A fifth aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the text sharing compliance detection method provided in the first aspect of this application.

[0032] It is understood that the beneficial effects of the second to fifth aspects mentioned above can be found in the relevant descriptions in the first aspect mentioned above, and will not be repeated here. Attached Figure Description

[0033] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0034] Figure 1 This is a flowchart of a text sharing compliance detection method provided in an embodiment of this application; Figure 2 This is a schematic diagram of the operation process of the text sharing compliance detection method provided in this application embodiment in a practical application scenario; Figure 3 This is a schematic diagram of the structure of a text sharing compliance detection device provided in an embodiment of this application; Figure 4 This is a schematic diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0035] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application can also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail. Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only for distinguishing descriptions and should not be construed as indicating or implying relative importance.

[0036] To enhance employees' skills and enrich their leisure time, companies sometimes organize knowledge-sharing activities to encourage employees to share reading notes, technical experiences, and other knowledge-sharing texts. To incentivize active sharing, companies typically reward employees with appropriate material rewards or points. However, some employees, in pursuit of rewards, may publish low-quality or plagiarized knowledge-sharing texts, seriously affecting the fairness of reward distribution.

[0037] To address the aforementioned issues, this application proposes a method, apparatus, electronic device, and computer program product for detecting the compliance of published knowledge-sharing text, thereby improving the fairness of reward distribution. For more specific technical details regarding the embodiments of this application, please refer to the various method embodiments described below.

[0038] It should be understood that the implementing entity of the various method embodiments proposed in this application can be various types of electronic devices, such as mobile phones, tablets, desktop computers, wearable devices, augmented reality (AR) / virtual reality (VR) devices, laptops, ultra-mobile personal computers (UMPCs), netbooks, personal digital assistants (PDAs), large-screen TVs, etc. The embodiments of this application do not impose any restrictions on the specific type of electronic device.

[0039] Please see Figure 1 This application illustrates a text sharing compliance detection method provided in an embodiment, comprising: 101. Obtain the knowledge-sharing text to be tested; First, the knowledge-sharing text to be performed for compliance testing is acquired. This knowledge-sharing text is published by a user and can be of various types, such as book reviews, technical experience, communication skills, or general knowledge dissemination. Users can publish this knowledge-sharing text through various means such as social applications, email, or SMS; this application embodiment does not impose any restrictions on this. For example, a user can publish knowledge-sharing text in a group chat interface of a social application, so that other group chat members can obtain the knowledge-sharing text through the group chat interface, thereby completing the text sharing process. The electronic device can obtain each published knowledge-sharing text by monitoring group chat information.

[0040] 102. Input the first preset prompt word into the large language model to instruct the large language model to perform a quality assessment on the multi-dimensional factors of the knowledge sharing text according to the built-in text quality assessment rules, and output the result of whether the quality of the knowledge sharing text meets the standard. When an electronic device detects knowledge-sharing text posted by a user through a group chat interface or other means, it automatically retrieves a pre-designed first preset prompt and inputs it into a Large Language Model (LLM). This application embodiment does not impose any restrictions on the specific type of LLM model used. The first preset prompt informs the LLM model to perform a quality assessment on the knowledge-sharing text. Upon receiving the first preset prompt, the LLM model parses and extracts semantics from the knowledge-sharing text, and performs a quality assessment on multiple dimensions of the text according to built-in text quality assessment rules, obtaining the corresponding quality assessment results to determine whether the quality of the knowledge-sharing text meets the standards. Generally, the LLM model can complete the quality assessment based on multiple dimensions of factors such as the authenticity, independent thinking, sincerity, writing style, whether it quotes from the original text, and whether it is AI-generated.

[0041] Taking the application scenario of sharing reading experiences as an example, the first preset prompt can inform the large language model to adopt the following quality assessment rules: (1) Quality meets the standard: The sharing includes the author's own real work scenarios, specific examples, specific experiences or insights, rather than general talk. The sharing reflects the author's independent and in-depth thinking on the theories in the book, with personalized understanding or specific application; (2) Quality does not meet the standard: The author only excerpts, restates or explains the original text of the book, without personalized in-depth thinking or specific application, and talks in a general way. The author lacks real work experience, specific examples or specific insights. Regarding the output of the results, the large language model can be informed to output the judgment results in the following format: Quality meets the standard / Quality does not meet the standard. In addition, the large language model can be asked to perform causal analysis, give the judgment basis, and briefly explain the reasons for outputting the corresponding judgment results so that relevant users can be convinced.

[0042] Furthermore, large language models can also rate or score the quality of knowledge-sharing texts and determine whether the quality meets the standards based on the rating or score. For example, the quality assessment result of knowledge-sharing text can include ratings such as excellent, good, average, passable, and fail. When the rating result is excellent or good, the quality of the knowledge-sharing text is considered to meet the standards; otherwise, the quality is considered to fail. Another example is that the quality assessment result of knowledge-sharing text can be a numerical score, ranging from 0 to 100. When the score reaches 80 or above, the quality is considered to meet the standards; otherwise, the quality is considered to fail. Yet another example is that the quality assessment result of knowledge-sharing text can be a star rating, ranging from 0 to 6 stars. When the star rating reaches 4 stars or above, the quality is considered to meet the standards; otherwise, the quality is considered to fail, and so on.

[0043] After using a large language model to assess the quality of knowledge-sharing texts, if the text fails to meet the quality standards, it can be directly determined that the compliance check has failed, and a low-quality tag is added, ending the process. This achieves the identification and filtering of low-quality text sharing. If the text meets the quality standards, a high-quality tag is added, and the plagiarism risk detection process described later continues.

[0044] 103. Detect whether knowledge-sharing texts pose a risk of plagiarism; In addition to checking whether the text quality meets the standards, compliance testing also needs to detect whether the text poses a risk of plagiarism. Considering that the main sources of plagiarism for text objects include historically shared texts and the Internet, plagiarism risk detection can be divided into two parts: historical plagiarism risk detection and online plagiarism risk detection, which will be described in detail below.

[0045] In one implementation of this application, detecting whether knowledge-sharing text has a plagiarism risk includes: (1) Detect the risk of historical plagiarism in knowledge-sharing texts by searching the historical shared text library; (2) Detect the risk of online plagiarism in knowledge-sharing texts through online searches; (3) In cases where there is a risk of historical plagiarism and / or online plagiarism in the knowledge-sharing text, it is determined that the knowledge-sharing text is at risk of plagiarism.

[0046] For historical plagiarism risk detection, all previously published shared texts by users can be collected in advance and stored in a historical shared text library. Then, a search of this library can be performed to find if there are any historical shared texts that are completely identical or similar to the current knowledge-sharing text, thus determining whether the current knowledge-sharing text carries a historical plagiarism risk. For online plagiarism risk detection, a web search interface can be called to search the internet for online texts that are completely identical or similar to the current knowledge-sharing text, thus determining whether the current knowledge-sharing text carries an online plagiarism risk. If the current knowledge-sharing text has at least one of the risks of historical plagiarism or online plagiarism, then it can be determined that the current knowledge-sharing text carries a plagiarism risk. Conversely, if the current knowledge-sharing text has neither the risk of historical plagiarism nor the risk of online plagiarism, then it can be determined that the current knowledge-sharing text does not carry a plagiarism risk. It can be seen that by combining historical plagiarism risk detection and online plagiarism risk detection, the possibility of plagiarism risk in the currently published knowledge-sharing text can be identified more comprehensively and accurately. The specific technical implementation details of historical plagiarism risk detection and online plagiarism risk detection are described below.

[0047] In one implementation of this application, the detection of historical plagiarism risk in knowledge-sharing texts is achieved by searching a historical shared text library, including: (1) Convert the knowledge-sharing text into the corresponding target text vector; (2) Compare the similarity between the target text vector and each historical text vector, and determine whether the knowledge sharing text has a risk of historical plagiarism based on the results of the similarity comparison; wherein, each historical text vector corresponds one-to-one with each historical sharing text contained in the historical sharing text library.

[0048] When detecting historical plagiarism risks by searching a historical shared text library, the currently published knowledge-sharing text is first converted into vector form, denoted as the target text vector. This conversion can be achieved using methods such as Word2Vec or Embedding. Additionally, each historical shared text in the library is also converted into a corresponding historical text vector. This step can be pre-completed, and the converted historical text vectors and their corresponding historical shared texts are stored together. Then, the target text vector is compared with each historical text vector for similarity. The results of the similarity comparison determine whether the knowledge-sharing text has a historical plagiarism risk. Specifically, similarity comparisons can be performed by calculating cosine similarity, Euclidean distance, or edit distance similarity between text vectors. If at least one historical text vector has a similarity higher than a set threshold with the target text vector, the knowledge-sharing text is considered to have a historical plagiarism risk. If no historical text vector has a similarity higher than the set threshold with the target text vector, the knowledge-sharing text is considered not to have a historical plagiarism risk. By comparing the similarity of the text vectors described above, it is possible to identify, relatively accurately, whether there are historical shared texts in the historical shared text library that are the same as or similar to the currently published knowledge-sharing text. This allows for a more accurate determination of whether the knowledge-sharing text has a historical plagiarism risk, effectively completing the historical plagiarism risk detection.

[0049] In one implementation of this application, the target text vector is compared with each historical text vector for similarity, and the risk of historical plagiarism in the knowledge-sharing text is determined based on the similarity comparison results, including: (1) Calculate the inner product between the target text vector and each historical text vector respectively; (2) Select the K historical text vectors with the largest inner product from each historical text vector, where K ≥ 1; (3) If there is a similar historical text vector among the K historical text vectors whose cosine similarity to the target text vector is greater than the first threshold, it is determined that the knowledge sharing text has a risk of historical plagiarism; or, if there is a similar historical text vector among the K historical text vectors whose edit distance similarity to the target text vector is greater than the second threshold, it is determined that the knowledge sharing text has a risk of historical plagiarism.

[0050] When comparing the similarity of a target text vector with each historical text vector, an inner product index technique can be used. The inner product index is a type of index used for vector similarity searching, particularly suitable for retrieving high-dimensional vectors. It measures the similarity between the query vector and the index vector by calculating the inner product. Based on approximate vector retrieval engines such as FAISS, Milvus, and Annoy, each text vector is normalized and written into the inner product index. The inner product between the target text vector and each historical text vector is then calculated. After this processing, each historical text vector obtains a corresponding inner product; the larger the inner product, the greater the similarity with the target text vector. Then, the K historical text vectors with the largest inner products are selected. The value of K can be set as needed, for example, to 10. Next, for each of the K historical text vectors, calculate its cosine similarity and edit distance similarity with the target text vector. If the cosine similarity between a historical text vector and the target text vector is greater than a first threshold, or the edit distance similarity is greater than a second threshold, then the historical text vector is recorded as a similar historical text vector. The first and second thresholds can be set as needed, for example, to 80% or 90%. If at least one similar historical text vector exists among the K historical text vectors, it can be determined that the currently published knowledge-sharing text has a historical plagiarism risk; if no similar historical text vector exists among the K historical text vectors, it can be determined that the currently published knowledge-sharing text does not have a historical plagiarism risk. By introducing the inner product index technique, vector similarity retrieval can be conveniently and accurately implemented to find potential similar historical text vectors.

[0051] After determining whether the knowledge-sharing text has a history of plagiarism risk, if the text does not have a history of plagiarism risk, the online plagiarism risk detection process can continue. If the knowledge-sharing text has a history of plagiarism risk, it indicates that the text has a plagiarism risk, and its compliance detection can be directly determined as failing. A plagiarism risk label will be added to the knowledge-sharing text, and the process will end, thus achieving the identification and filtering of text sharing plagiarism.

[0052] In one implementation of this application, after determining whether the knowledge-sharing text has a historical plagiarism risk based on the similarity comparison result, the method further includes: (1) If the knowledge-sharing text has a risk of historical plagiarism, then identify the similar historical text vectors that have a similar relationship with the target text vector from each historical text vector; (2) Determine the similar historical shared texts corresponding to the similar historical text vectors; (3) Determine the type of historical plagiarism risk based on the authors and publication time of similar historical shared texts and the authors and publication time of knowledge-sharing texts.

[0053] When storing each published shared text, the corresponding author and publication time can be stored together. This allows for further determination of the type of historical plagiarism risk after identifying a potential historical plagiarism risk in the currently published knowledge-sharing text. Specifically, if a knowledge-sharing text has a historical plagiarism risk, it means that there are similar historical text vectors among the various historical text vectors that have a similarity relationship to the target text vector, as described above. In this case, first, the similar historical shared texts corresponding to these similar historical text vectors are identified. Then, based on the author and publication time of these similar historical shared texts, as well as the author and publication time of the currently published knowledge-sharing text, the type of historical plagiarism risk is determined. For example, based on the date rules for author names and publication times, it can be determined whether the currently published knowledge-sharing text and similar historical shared texts belong to the same author or the same statistical time period, thereby determining whether the type of historical plagiarism risk is "duplicate" or "multiple submissions".

[0054] In one implementation of this application, the type of historical plagiarism risk is determined based on the author and publication time of similar historical shared texts, as well as the author and publication time of knowledge-sharing texts, including: (1) If the author of the similar historical shared text is different from the author of the knowledge shared text, then the type of historical plagiarism risk is determined to be duplicate; (2) If the author of the similar historical shared text is the same as the author of the knowledge shared text, and the publication time of the similar historical shared text is within the same time period as the publication time of the knowledge shared text, then the type of historical plagiarism risk is determined to be multiple publications. (3) If the author of the similar historical shared text is the same as the author of the knowledge shared text, and the publication time of the similar historical shared text is not in the same time period as the publication time of the knowledge shared text, then the type of historical plagiarism risk is determined to be duplicate.

[0055] If the author of the similar historical shared text is different from the author of the knowledge-sharing text, it indicates that the knowledge-sharing text published by the current user may plagiarize historical shared text published by other users. In this case, the type of historical plagiarism risk is determined as "duplicate". If the author of the similar historical shared text is the same as the author of the knowledge-sharing text, it indicates that the knowledge-sharing text published by the current user may plagiarize its own previously published historical shared text. In this case, the plagiarism risk type needs to be classified in conjunction with the text's publication time. If the publication time of the similar historical shared text and the knowledge-sharing text are within the same time period (meaning the same statistical time period, such as the same month), it indicates that the current knowledge-sharing text is a duplicate publication within the same time period, and therefore the type of historical plagiarism risk is determined as "multiple publication". If the publication time of the similar historical shared text and the knowledge-sharing text are not within the same time period, it indicates that the current knowledge-sharing text plagiarizes historical shared text across different time periods, and therefore the type of historical plagiarism risk is determined as "duplicate". By setting it up in this way, the type of historical plagiarism risk can be identified based on the author and publication time of the shared text, thereby providing users with richer and more complete historical plagiarism risk information and improving the comprehensiveness and practicality of historical plagiarism risk detection.

[0056] The above describes the relevant content regarding historical plagiarism risk detection. This section will now describe the relevant content regarding online plagiarism risk detection.

[0057] In one implementation of this application, detecting the risk of online plagiarism in knowledge-sharing text through web search includes: (1) Extract M sentences from the knowledge sharing text, where M≥1; (2) Using the web search interface, crawl the web to retrieve the N text search results with the highest similarity to the M sentences, where N≥1; (3) Input the second preset prompt word into the large language model to instruct the large language model to output the result of whether the knowledge sharing text is at risk of online plagiarism based on the similarity between M sentences and N text search results.

[0058] To detect online plagiarism risks, firstly, M sentences are extracted from the currently published knowledge-sharing text, where M is any configurable value greater than or equal to 1, for example, 2. Then, using various web search interfaces, such as BochaAI, Bing Search API, and Google CSE, N text search results with the highest similarity to the M sentences are crawled online. N is also any configurable value greater than or equal to 1, for example, 20. As an example, assuming two sentences S1 and S2 are extracted from the knowledge-sharing text, a web search is performed on S1 and S2, collecting the top 50 text search results with the highest similarity to S1 and S2. These top 50 search results can then be deduplicated and sorted in descending order of similarity. Finally, the top 20 search results are selected as the comparison objects for subsequent assessment of online plagiarism risks. After obtaining the aforementioned N text search results, a second preset prompt word with a fixed design is retrieved and input into the large language model. This instructs the large language model to output a result indicating whether the knowledge-sharing text carries a risk of online plagiarism, based on the similarity between the extracted M sentences and the N text search results. Specifically, upon receiving the second preset prompt word, the large language model calculates the similarity between the extracted M sentences and the N text search results according to its built-in text similarity calculation rules. Then, based on the magnitude of the similarity, it outputs either "risk of online plagiarism exists" or "no risk of online plagiarism exists."

[0059] In one implementation of this application, M sentences are extracted from knowledge-sharing text, including: (1) Find the target content of the author's reflections in the knowledge sharing text; (2) Extract M sentences from the target content whose word count exceeds the third threshold.

[0060] Assuming the current application scenario is sharing book reviews, knowledge-sharing text typically includes two parts: a restatement of the book's content and the author's reflections. The restatement itself is a repetition of the book's content and does not fall under the scope of plagiarism detection; therefore, online plagiarism detection targets the author's reflections. Furthermore, to ensure the accuracy of online plagiarism detection, the extracted sentences should not be too short and need to contain relatively rich and complete semantics. Based on this, firstly, from the currently published knowledge-sharing text, the author's reflections section is identified using keyword recognition or semantic recognition, and this is recorded as the target content. Then, M sentences exceeding a third threshold in length are extracted from the target content according to a predefined method; for example, two long sentences of 30-40 characters each can be randomly extracted. By extracting a certain number of long sentences from the author's reflections section of the knowledge-sharing text and comparing their similarity with text search results obtained through online searches, the risk of online plagiarism in the knowledge-sharing text can be detected relatively accurately.

[0061] In another implementation of this application's embodiments, M sentences are extracted from the knowledge-sharing text, including: Input a third preset prompt word into the large language model to instruct it to extract M sentences from the knowledge-sharing text that have more than the third threshold number of words.

[0062] Another approach is to utilize a large language model to automatically extract sentences. Specifically, a pre-defined third set of prompts can be obtained and input into the large language model. This instructs the model to perform semantic analysis on the currently published knowledge-sharing text, identify the author's reflections, and extract a representative number of long sentences. Ultimately, this extracts M sentences exceeding the third threshold from the knowledge-sharing text. Clearly, using a large language model allows for convenient and accurate extraction of sentences needed for online plagiarism risk detection from knowledge-sharing text.

[0063] In one implementation of this application, the quantities of M and N are set in the following manner: (1) Determine the number of times the author of the knowledge-sharing text has plagiarized in the past, and the number of target historical shared texts published by the author of the knowledge-sharing text; (2) Set the number of M and N based on the number of historical plagiarisms and the number of target historical shared texts; where the number of M and N is positively correlated with the number of historical plagiarisms and negatively correlated with the number of target historical shared texts.

[0064] In the technical solution of this application embodiment, the large language model needs to compare the similarity between M sentences extracted from knowledge-sharing texts and N text search results obtained through network searches. It is foreseeable that if the number of M and N is large, problems such as excessive system computing power and time consumption may occur. To solve this problem, this application embodiment can assess the probability of an author engaging in online plagiarism based on the author's historical plagiarism count and the number of published shared texts, and reasonably set the number of M and N accordingly. Specifically, as described above, when storing each published shared text, the corresponding publishing author and publishing time can be associated and stored together, so the number of historical shared texts published by each author is known. Furthermore, for any author, since each knowledge-sharing text they publish undergoes corresponding compliance checks to identify plagiarism risks, and if a plagiarism risk exists, the historical plagiarism count is accumulated once, so the historical plagiarism count for each author is also known. For currently published knowledge-sharing texts, first, obtain the author's historical plagiarism count C1 and the number of previously published shared texts C2. Then, based on the historical plagiarism count C1 and the number of previously published shared texts C2, reasonably assess the probability of the author engaging in online plagiarism, and accordingly set the values ​​of M and N. Generally speaking, if the historical plagiarism count C1 is higher and the number of previously published shared texts C2 is lower, the probability of the author engaging in online plagiarism is higher. In this case, the intensity of online plagiarism risk detection needs to be increased, so a larger value of M and N can be set. Conversely, if the historical plagiarism count C1 is lower and the number of previously published shared texts C2 is higher, the probability of the author engaging in online plagiarism is lower. In this case, the intensity of online plagiarism risk detection can be reduced to decrease the consumption of system computing time, so a smaller value of M and N can be set. That is, the values ​​of M and N are positively correlated with the historical plagiarism count and negatively correlated with the number of target previously published shared texts. Therefore, by reasonably setting the values ​​of M and N based on the author's historical plagiarism count and the number of published and shared texts, the detection intensity of online plagiarism risk can be flexibly adjusted. On the one hand, this can reduce the consumption of system computing power and time, and on the other hand, there is no need to worry about missing online plagiarism.

[0065] After determining whether the currently published knowledge-sharing text poses a risk of online plagiarism, if the text does not pose a risk, it can be confirmed that the text does not have a historical plagiarism risk, since it has already been determined that the text does not have a plagiarism risk. If the text does pose a risk of online plagiarism, it means that the text has a plagiarism risk, its compliance check has failed, and a plagiarism risk label has been added to the text.

[0066] 104. If the quality of the knowledge-sharing text meets the standards and there is no risk of plagiarism, then the compliance test of the knowledge-sharing text is deemed to have passed.

[0067] After completing the compliance testing processes described above, including quality assessment, historical plagiarism risk detection, and online plagiarism risk detection, if the quality of the currently published knowledge-sharing text meets the standards and there is no risk of plagiarism (neither historical nor online), then the compliance test for the knowledge-sharing text can be considered passed. Conversely, if the quality of the currently published knowledge-sharing text does not meet the standards, or if there is a risk of plagiarism (both historical and / or online), then the compliance test for the knowledge-sharing text can be considered failed.

[0068] It should be noted that the compliance testing process described above executes the steps of quality assessment, historical plagiarism risk detection, and online plagiarism risk detection sequentially. However, this embodiment does not restrict the execution order of these steps. For example, it is also feasible to execute them in the order of quality assessment, online plagiarism risk detection, and historical plagiarism risk detection, or in the order of historical plagiarism risk detection, online plagiarism risk detection, and quality assessment. The compliance test is considered passed only if all steps are successfully completed.

[0069] In one implementation of this application, the method further includes: (1) When the compliance check of the knowledge sharing text passes, add a reward tag to the knowledge sharing text; (2) When the compliance test of the knowledge sharing text fails, add a non-reward tag to the knowledge sharing text.

[0070] For enterprise knowledge-sharing scenarios, when a knowledge-sharing text passes the compliance check, it indicates that the text is high-quality, original content and meets the reward criteria; therefore, a reward-eligible tag can be added to the text. Conversely, when a knowledge-sharing text fails the compliance check, it indicates that the text is low-quality or plagiarized and does not meet the reward criteria; therefore, a non-reward tag is added to the text. This setting effectively filters out low-quality or plagiarized non-compliant shared text, reduces manual review costs, and greatly improves the fairness and reasonableness of reward distribution in enterprise knowledge-sharing scenarios.

[0071] On the other hand, for each knowledge-sharing text, information such as its author, publication time, main content, quality assessment results and reasons, plagiarism detection results and reasons, relevant information on suspected plagiarized historical shared texts or online texts, and reward / non-reward tags can be integrated into a report and stored in a database for users to access when needed.

[0072] In the technical solution of this application embodiment, the knowledge-sharing text to be detected is first obtained, and then a first preset prompt word is input into the large language model. After receiving the first preset prompt word, the large language model performs a quality assessment of the multi-dimensional factors of the knowledge-sharing text according to its built-in text quality assessment rules, and outputs the result of whether the quality of the knowledge-sharing text meets the standards. In addition, it also detects whether the knowledge-sharing text has a risk of plagiarism. If the quality of the knowledge-sharing text meets the standards and there is no risk of plagiarism, then the compliance test of the knowledge-sharing text is determined to be passed. By setting it up in this way, it is possible to effectively identify non-compliant situations such as low quality or plagiarism in knowledge-sharing texts, that is, to achieve compliance testing of knowledge-sharing texts, thereby improving the fairness of reward distribution in enterprise knowledge-sharing scenarios.

[0073] For a better understanding of the technical solutions in the embodiments of this application, please refer to [link / reference]. Figure 2 This illustrates the operation flow of the text sharing compliance detection method provided in this application embodiment under a practical application scenario. Figure 2 When a user posts a knowledge-sharing text, the AI ​​large language model, guided by preset prompts, can automatically assess the quality of the text and obtain the corresponding quality assessment results. If the quality assessment result is substandard, the text is considered non-compliant and low-quality, and a "non-rewardable" tag is added. If the quality assessment result is satisfactory, the text is considered high-quality, and the plagiarism risk detection process begins. This process includes historical plagiarism risk detection and online plagiarism risk detection. If the text is determined to have historical and / or online plagiarism risks, it is considered non-compliant and a "non-rewardable" tag is added. If the text is determined to have no historical or online plagiarism risks, it is considered compliant and original, and a "rewardable" tag is added.

[0074] In summary, the embodiments of this application, through a three-level detection chain of "semantic quality assessment - historical record comparison - internet retrieval comparison", can accurately determine the compliance of knowledge-sharing texts, achieve low-quality content filtering, historical plagiarism identification, and online plagiarism identification, significantly reducing the cost of manual review and improving the fairness and rationality of reward distribution.

[0075] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0076] The above mainly describes a method for detecting text sharing compliance. The following will describe a device for detecting text sharing compliance.

[0077] Please see Figure 3 This application illustrates a text sharing compliance detection device provided in an embodiment, comprising: The shared text acquisition module 301 is used to acquire the knowledge sharing text to be detected. The text quality assessment module 302 is used to input the first preset prompt word into the large language model, so as to instruct the large language model to perform quality assessment on the multi-dimensional factors of the knowledge sharing text according to the built-in text quality assessment rules, and output the result of whether the quality of the knowledge sharing text meets the standard. The plagiarism risk detection module 303 is used to detect whether there is a risk of plagiarism in knowledge-sharing texts. The compliance determination module 304 is used to determine that the compliance test of the knowledge sharing text has passed if the quality of the knowledge sharing text meets the standards and there is no risk of plagiarism.

[0078] In one implementation of this application, the plagiarism risk detection module includes: The historical plagiarism detection unit is used to detect whether knowledge-sharing texts have historical plagiarism risks by searching the historical shared text library; The online plagiarism detection unit is used to detect whether knowledge-sharing texts are at risk of online plagiarism through online searches; The plagiarism risk determination unit is used to determine whether a knowledge-sharing text is at risk of plagiarism when there is a risk of historical plagiarism and / or online plagiarism.

[0079] In one implementation of this application, the historical plagiarism detection unit includes: The text vector conversion subunit is used to convert knowledge-sharing text into corresponding target text vectors; The vector similarity comparison subunit is used to compare the similarity between the target text vector and each historical text vector, and to determine whether the knowledge-sharing text has a risk of historical plagiarism based on the similarity comparison results; wherein, each historical text vector corresponds one-to-one with each historical shared text contained in the historical shared text library.

[0080] In one implementation of this application, the vector similarity comparison subunit includes: The inner product calculation subunit is used to calculate the inner product between the target text vector and each historical text vector, respectively. The vector selection sub-unit is used to select the K historical text vectors with the largest inner product from each historical text vector, where K≥1; The historical plagiarism risk determination subunit is used to determine if a knowledge-sharing text has a historical plagiarism risk if there is a similar historical text vector among K historical text vectors whose cosine similarity to the target text vector is greater than a first threshold; or, if there is a similar historical text vector among K historical text vectors whose edit distance similarity to the target text vector is greater than a second threshold.

[0081] In one implementation of this application, the historical plagiarism detection unit further includes: The similarity vector determination subunit is used to determine similar historical text vectors that have a similar relationship with the target text vector from various historical text vectors if there is a risk of historical plagiarism in the knowledge-sharing text. The similar text determination subunit is used to determine similar historical shared texts corresponding to similar historical text vectors; The risk type determination subunit is used to determine the type of historical plagiarism risk based on the authors and publication times of similar historical shared texts, as well as the authors and publication times of knowledge-sharing texts.

[0082] In one implementation of this application, the risk type determination subunit includes: The first type of identification subunit is used to determine the type of historical plagiarism risk as "duplicate" if the author of the similar historical shared text is different from the author of the knowledge shared text. The second type of determination subunit is used to determine the type of historical plagiarism risk as multiple publications if the author of the similar historical shared text is the same as the author of the knowledge shared text, and the publication time of the similar historical shared text is within the same time period as the publication time of the knowledge shared text. The third type of sub-unit is used to determine the type of historical plagiarism risk as "duplicate" if the author of the similar historical shared text is the same as the author of the knowledge shared text, and the publication time of the similar historical shared text is not in the same time period as the publication time of the knowledge shared text.

[0083] In one implementation of this application, the online plagiarism detection unit includes: The sentence extraction subunit is used to extract M sentences from the knowledge-sharing text, where M≥1; The web search subunit is used to crawl the web and retrieve the N text search results with the highest similarity to M sentences, where N≥1, using the web search interface. The online plagiarism risk determination subunit is used to input a second preset prompt word into the large language model, so as to instruct the large language model to output the result of whether the knowledge sharing text has the risk of online plagiarism based on the similarity between M sentences and N text search results.

[0084] In one implementation of this application embodiment, the sentence extraction subunit includes: The "Reflections" sub-unit is used to find the target content of the author's reflections from knowledge-sharing texts; The first long sentence extraction sub-unit is used to extract M sentences from the target content whose word count exceeds the third threshold.

[0085] In another implementation of this application embodiment, the sentence extraction subunit includes: The second long sentence extraction subunit is used to input the third preset prompt word into the large language model, instructing the large language model to extract M sentences from the knowledge-sharing text that exceed the third threshold in word count.

[0086] In one implementation of this application, the network plagiarism detection unit further includes: The historical parameter determination subunit is used to determine the number of times an author of a knowledge-sharing text has plagiarized in the past, as well as the number of target historical shared texts that the author of the knowledge-sharing text has published. The parameter setting subunit is used to set the quantity of M and N based on the number of historical plagiarisms and the number of target historical shared texts; wherein the quantity of M and N is positively correlated with the number of historical plagiarisms and negatively correlated with the number of target historical shared texts.

[0087] In one implementation of this application, the text sharing compliance detection device further includes: The compliance determination module is used to determine if the compliance test of a knowledge-sharing text fails if the quality of the text does not meet the standards or if the text is at risk of plagiarism.

[0088] In one implementation of this application, the text sharing compliance detection device further includes: The first tag addition module is used to add rewardable tags to knowledge-sharing texts when the compliance check is passed. The second tag-adding module is used to add a non-reward tag to knowledge-sharing text when the compliance check fails.

[0089] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the text sharing compliance detection method described in any of the above embodiments.

[0090] This application also provides a computer program product that, when run on an electronic device, causes the electronic device to execute the text sharing compliance detection method as described in any of the above embodiments.

[0091] Figure 4 This is a schematic diagram of an electronic device provided in an embodiment of this application. Figure 4 As shown, the electronic device 4 in this embodiment includes: a processor 40, a memory 41, and a computer program 42 stored in the memory 41 and executable on the processor 40. When the processor 40 executes the computer program 42, it implements the steps in the embodiments of the various text sharing compliance detection methods described above, for example... Figure 1 Steps 101-104 are shown. Alternatively, when the processor 40 executes the computer program 42, it implements the functions of each module / unit in the above-described device embodiments, for example, implementing... Figure 3 The functions of modules 301-304 of the device shown.

[0092] The computer program 42 can be divided into one or more modules / units, which are stored in the memory 41 and executed by the processor 40 to complete this application. The one or more modules / units can be a series of computer program instruction segments capable of performing specific functions, which describe the execution process of the computer program 42 in the electronic device 4.

[0093] The processor 40 may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0094] The memory 41 can be an internal storage unit of the electronic device 4, such as a hard disk or memory. The memory 41 can also be an external storage device of the electronic device 4, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc. Furthermore, the memory 41 can include both internal and external storage units of the electronic device 4. The memory 41 is used to store the computer program and other programs and data required by the electronic device. The memory 41 can also be used to temporarily store data that has been output or will be output.

[0095] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0096] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0097] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0098] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented 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 this application.

[0099] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection between devices or units through some interfaces, and may be electrical, mechanical, or other forms.

[0100] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of the embodiments of this application, depending on actual needs.

[0101] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0102] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals.

[0103] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for detecting compliance in text sharing, characterized in that, include: Obtain the knowledge-sharing text to be tested; Input a first preset prompt word into the large language model to instruct the large language model to perform a quality assessment on the multi-dimensional factors of the knowledge sharing text according to the built-in text quality assessment rules, and output the result of whether the quality of the knowledge sharing text meets the standard; Detect whether the knowledge-sharing text is at risk of plagiarism; If the quality of the knowledge-sharing text meets the standards and there is no risk of plagiarism, then the compliance test of the knowledge-sharing text is deemed to have passed.

2. The method as described in claim 1, characterized in that, The detection of whether the knowledge-sharing text is at risk of plagiarism includes: The risk of historical plagiarism in the knowledge-sharing text is detected by searching the historical shared text library. The knowledge-sharing text was checked for plagiarism risks through online searches. If the knowledge-sharing text has historical and / or online plagiarism risks, it is determined that the knowledge-sharing text has plagiarism risks.

3. The method as described in claim 2, characterized in that, The method of detecting whether the knowledge-sharing text has a historical plagiarism risk through searching a historical shared text library includes: Convert the knowledge-sharing text into the corresponding target text vector; The target text vector is compared with each historical text vector for similarity, and the results of the similarity comparison are used to determine whether the knowledge-sharing text has a risk of historical plagiarism; wherein each historical text vector corresponds one-to-one with each historical shared text contained in the historical shared text library.

4. The method as described in claim 3, characterized in that, The step of comparing the similarity of the target text vector with each historical text vector, and determining whether the knowledge-sharing text has a historical plagiarism risk based on the similarity comparison results, includes: Calculate the inner product between the target text vector and each of the historical text vectors respectively; From the various historical text vectors, select the K historical text vectors with the largest inner product, where K ≥ 1; If, among the K historical text vectors, there is a similar historical text vector whose cosine similarity to the target text vector is greater than a first threshold, the knowledge-sharing text is determined to have a historical plagiarism risk; or, if, among the K historical text vectors, there is a similar historical text vector whose edit distance similarity to the target text vector is greater than a second threshold, the knowledge-sharing text is determined to have a historical plagiarism risk.

5. The method as described in claim 3, characterized in that, After determining whether the knowledge-sharing text has a historical plagiarism risk based on the similarity comparison results, the method further includes: If the knowledge-sharing text has a history of plagiarism risk, then similar historical text vectors that have a similar relationship with the target text vector are determined from the various historical text vectors; Determine the similar historical shared texts corresponding to the similar historical text vectors; The type of historical plagiarism risk is determined based on the authors and publication times of the similar historical shared texts and the authors and publication times of the knowledge-sharing texts.

6. The method as described in claim 5, characterized in that, The method of determining the type of historical plagiarism risk based on the authors and publication times of the similar historical shared texts, and the authors and publication times of the knowledge-sharing texts, includes: If the author of the similar historical shared text is different from the author of the knowledge shared text, then the type of historical plagiarism risk is determined to be duplicate. If the author of the similar historical shared text is the same as the author of the knowledge shared text, and the publication time of the similar historical shared text is within the same time period as the publication time of the knowledge shared text, then the type of historical plagiarism risk is determined to be multiple publications. If the author of the similar historical shared text is the same as the author of the knowledge shared text, and the publication time of the similar historical shared text is not within the same time period as the publication time of the knowledge shared text, then the type of historical plagiarism risk is determined to be duplicate.

7. The method as described in claim 2, characterized in that, The method of detecting whether the knowledge-sharing text is at risk of online plagiarism through network search includes: Extract M sentences from the knowledge-sharing text, where M ≥ 1; Using a web search interface, crawl the web to retrieve the N text search results with the highest similarity to the M sentences, where N≥1; Input a second preset prompt word into the large language model to instruct the large language model to output a result indicating whether the knowledge-sharing text is at risk of online plagiarism based on the similarity between the M sentences and the N text search results.

8. The method as described in claim 7, characterized in that, The step of extracting M sentences from the knowledge-sharing text includes: From the knowledge-sharing text, find the target content in the author's reflections section; From the target content, extract the M sentences whose word count exceeds the third threshold.

9. The method as described in claim 7, characterized in that, The step of extracting M sentences from the knowledge-sharing text includes: Input a third preset prompt word into the large language model to instruct the large language model to extract the M sentences from the knowledge-sharing text that have more than a third threshold number of words.

10. The method as described in claim 7, characterized in that, The quantities of M and N are set in the following manner: Determine the number of times the author of the knowledge-sharing text has plagiarized in the past, and the number of target historical shared texts published by the author of the knowledge-sharing text; Based on the number of historical plagiarisms and the number of target historical shared texts, the quantities of M and N are set; wherein the quantities of M and N are positively correlated with the number of historical plagiarisms and negatively correlated with the number of target historical shared texts.

11. The method according to any one of claims 1 to 10, characterized in that, The method further includes: If the quality of the knowledge-sharing text does not meet the standards, or if the knowledge-sharing text has a risk of plagiarism, then the compliance test of the knowledge-sharing text is determined to have failed.

12. The method as described in claim 11, characterized in that, The method further includes: When the compliance check of the knowledge-sharing text passes, a rewardable tag is added to the knowledge-sharing text; When the compliance check of the knowledge-sharing text fails, a "non-rewardable" tag is added to the knowledge-sharing text.

13. A text-sharing compliance detection device, characterized in that, include: The shared text acquisition module is used to acquire the knowledge-sharing text to be detected; The text quality assessment module is used to input a first preset prompt word into the large language model to instruct the large language model to perform a quality assessment on the multi-dimensional factors of the knowledge sharing text according to the built-in text quality assessment rules, and output the result of whether the quality of the knowledge sharing text meets the standard. The plagiarism risk detection module is used to detect whether the knowledge-sharing text is at risk of plagiarism. The compliance determination module is used to determine that the compliance test of the knowledge-sharing text has passed if the quality of the knowledge-sharing text meets the standards and there is no risk of plagiarism.

14. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the text sharing compliance detection method as described in any one of claims 1 to 12.

15. A computer program product, characterized in that, When the computer program product is run on an electronic device, the electronic device performs the text sharing compliance detection method as described in any one of claims 1 to 12.

Images