Text quality detection method and device, storage medium and electronic equipment

Abstract

Embodiments of the invention provide a text quality detection method and apparatus, a storage medium and an electronic device. The method comprises the steps of obtaining a to-be-detected text; segmenting the to-be-detected text into a plurality of paragraphs; inputting each paragraph into a trained text detection model for text detection to obtain a first detection result of each paragraph; according to the first paragraph, performing text duplicate checking on each second paragraph to obtain a plurality of second paragraphs after duplicate checking; calculating the similarity between the first paragraph and each second paragraph after duplicate checking; obtaining a second detection result of each paragraph according to the similarity between the first paragraph and each second paragraph after duplicate checking; determining a text quality detection result of each paragraph according to the first detection result and the second detection result; according to the text quality detection result of each paragraph, the text quality detection result of the to-be-detected text is obtained, and the text quality detection efficiency and accuracy can be improved.

Description

[0001] This application claims priority to Chinese patent application No. 202410541848.9 filed on April 30, 2024, entitled “Text quality detection method, device, storage medium and electronic device”, the entire contents of which are incorporated herein by reference. Technical Field

[0002] The embodiments of the present application relate to the field of text processing technology, and in particular to a text quality detection method, device, storage medium, and electronic device. Background Art

[0003] With the development of science and technology, many innovators are increasingly concerned about protecting scientific research results. However, for various reasons, the texts recording scientific research results may contain fabricated, forged, plagiarized, simply replaced, or piled up content. Therefore, it is necessary to conduct text quality inspections to detect these issues and ensure the text's standardization.

[0004] In the existing technology, when performing text quality detection on text, since the application of automated detection technology in text quality detection is still immature, the text quality detection is carried out through manual inspection. This method relies on the work experience and subjective judgment of relevant personnel, and has low detection efficiency and accuracy.

[0005] Therefore, providing an efficient, accurate and automated text quality detection method is a technical problem that needs to be solved urgently by those skilled in the art. Summary of the Invention

[0006] In order to overcome the problems existing in the related art, the present application provides a text quality detection method, device, storage medium and electronic device, which can improve the efficiency and accuracy of text quality detection.

[0007] According to a first aspect of an embodiment of the present application, a text quality detection method is provided, comprising the following steps:

[0008] Get the text to be detected;

[0009] Divide the text to be detected into several paragraphs;

[0010] Input each paragraph into the trained text detection model for text detection to obtain the first detection result of each paragraph;

[0011] One of the paragraphs segmented from the text to be detected is used as the first paragraph, and the other paragraphs from the multiple paragraphs segmented from the text to be detected, except the first paragraph, are used as the second paragraphs; based on the first paragraph, each second paragraph is checked for duplicate text to obtain multiple second paragraphs after duplicate text checking; and the similarity between the first paragraph and each second paragraph after duplicate text checking is calculated;

[0012] Obtaining a second detection result for each paragraph based on the similarity between the first paragraph and each second paragraph after the duplicate check;

[0013] Determining a text quality detection result for each paragraph based on the first detection result and the second detection result;

[0014] According to the text quality detection result of each paragraph, the text quality detection result of the text to be detected is obtained.

[0015] According to a second aspect of an embodiment of the present application, a text quality detection device is provided, comprising:

[0016] A module for acquiring text to be detected, used for acquiring text to be detected;

[0017] The text segmentation module is used to segment the text into several paragraphs.

[0018] A first detection result obtaining module is used to input each paragraph into a trained text detection model for text detection to obtain a first detection result for each paragraph;

[0019] A similarity calculation module is configured to take one of the paragraphs segmented from the text to be detected as a first paragraph, and take the other paragraphs from the plurality of paragraphs segmented from the text to be detected, excluding the first paragraph, as second paragraphs; perform a text duplication check on each second paragraph based on the first paragraph to obtain a plurality of second paragraphs after duplication checking; and calculate the similarity between the first paragraph and each second paragraph after duplication checking;

[0020] A second detection result obtaining module is used to obtain a second detection result for each paragraph based on the similarity between the first paragraph and each second paragraph after the duplicate check;

[0021] A text quality detection result determination module, configured to determine a text quality detection result of each paragraph based on the first detection result and the second detection result;

[0022] The text quality detection result obtaining module is used to obtain the text quality detection result of the text to be detected based on the text quality detection result of each paragraph.

[0023] According to a third aspect of an embodiment of the present application, an electronic device is provided, including a display, a processor and a memory; the memory stores a computer program, and the computer program is suitable for being loaded by the processor and executing the text quality detection method as described above.

[0024] According to a fourth aspect of an embodiment of the present application, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the text quality detection method as described above is implemented.

[0025] The embodiment of the present application obtains the text to be detected; divides the text to be detected into several paragraphs; inputs each paragraph into a trained text detection model for text detection to obtain a first detection result for each paragraph; uses one of the paragraphs divided from the text to be detected as the first paragraph, and uses the other paragraphs except the first paragraph from the several paragraphs divided from the text to be detected as the second paragraph; performs a text duplication check on each second paragraph based on the first paragraph to obtain several second paragraphs after duplication checking; calculates the similarity between the first paragraph and each second paragraph after duplication checking; obtains a second detection result for each paragraph based on the similarity between the first paragraph and each second paragraph after duplication checking; determines the text quality detection result of each paragraph based on the first detection result and the second detection result; obtains the text quality detection result of the text to be detected based on the text quality detection result of each paragraph. The present application determines the text quality detection result of each paragraph by inputting the paragraphs into a trained text detection model for text detection, checking the paragraphs for duplicates, and calculating the similarity between the paragraphs, thereby determining the text quality detection result of the text to be detected, thereby improving the efficiency and accuracy of text quality detection.

[0026] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application.

[0027] For better understanding and implementation, the present invention is described in detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 This is a flow chart of a text quality detection method according to an embodiment of the present application;

[0029] Figure 2 This is a flow chart of step S40 in the text quality detection method according to an embodiment of the present application;

[0030] Figure 3 This is a flow chart of step S40 in the text quality detection method according to another embodiment of the present application;

[0031] Figure 4 This is a flow chart showing the steps after step S42 in the text quality detection method according to one embodiment of the present application;

[0032] Figure 5 This is a schematic block diagram of the structure of a text quality detection device according to one embodiment of the present application;

[0033] Figure 6 This is a schematic block diagram of the structure of an electronic device shown in one embodiment of the present application. DETAILED DESCRIPTION

[0034] In order to make the objectives, technical solutions and advantages of the present application clearer, the embodiments of the present application will be described in further detail below with reference to the accompanying drawings.

[0035] It should be clear that the embodiments described are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the embodiments of the present application.

[0036] When the following description refers to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. In the description of this application, it should be understood that the terms "first", "second", "third", etc. are only used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence, nor can they be understood as indicating or implying relative importance. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood according to the specific circumstances. The singular forms of "a", "said" and "the" used in this application and the appended claims are also intended to include the plural forms, unless the context clearly indicates other meanings. The words "if" / "if" used herein can be interpreted as "at the time of" or "when" or "in response to determination".

[0037] In addition, in this application, unless otherwise specified, "plurality" refers to two or more. "And / or" describes the relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can mean: A exists alone, A and B exist simultaneously, or B exists alone. The character " / " generally indicates that the associated objects are in an "or" relationship.

[0038] The text quality detection method of the embodiment of the present application can be applied to text quality detection scenarios, specifically, to detect whether the text contains fabricated and forged content, and to detect whether the text contains plagiarized, simply replaced, or piled up content.

[0039] The text quality detection method of the embodiments of the present application can be performed by a text quality detection device. The text quality detection device can be implemented through software and / or hardware. The text quality detection device can be composed of two or more physical entities, or a single physical entity. The hardware referred to by the text quality detection device is essentially a computer device. For example, the text quality detection device can be a computer, a mobile phone, a tablet, or an intelligent device such as an interactive tablet.

[0040] To this end, the embodiment of the present application determines the text quality detection result of each paragraph by inputting the paragraph into a trained text detection model for text detection, checking the paragraphs for duplication, calculating the similarity between paragraphs, and then determining the text quality detection result of the text to be detected, thereby improving the efficiency and accuracy of text quality detection.

[0041] Based on this, the present application proposes a text quality detection method, device, storage medium and electronic device.

[0042] See also Figure 1 , the text quality detection method provided in the embodiment of the present application includes:

[0043] S10: Obtain the text to be detected.

[0044] The text to be tested includes, but is not limited to, papers to be submitted and self-media text to be published. The text to be tested can be encrypted or unencrypted.

[0045] In the embodiment of the present application, the text to be detected is an encrypted text to be submitted. The encrypted text can be in various languages, which is not limited by the present application.

[0046] When receiving the text to be tested, the text quality detection device can receive all the encrypted texts of the user at one time; the text quality detection device can also receive several encrypted segment files uploaded by the user based on the encrypted text, decrypt the several encrypted segment files, obtain several decrypted segment files, perform document parsing and document merging on the several decrypted segment files, and obtain the final text to be tested.

[0047] S20: Segment the text to be detected into several paragraphs.

[0048] In an optional embodiment, the text to be detected is divided into natural paragraphs, and the divided text is processed to remove stop words and blocked words to obtain several paragraphs.

[0049] In another optional embodiment, in order to improve the accuracy of paragraph division, the text to be detected is segmented according to the punctuation marks of each text to be detected; the segmented text is processed to remove stop words and blocked words to obtain several paragraphs.

[0050] In an embodiment of the present application, paragraph separators in the text to be detected can be identified, and the text to be detected can be divided into several paragraphs according to the paragraph separators.

[0051] S30: Input each paragraph into the trained text detection model for text detection to obtain a first detection result for each paragraph.

[0052] The trained text detection models include, but are not limited to, machine learning models and deep learning models. The trained text detection models are used to determine whether text is generated by artificial intelligence.

[0053] Among them, the first detection result is used to indicate the possibility that the paragraph content is generated by artificial intelligence.

[0054] In the embodiment of the present application, considering that the text to be detected may be generated by artificial intelligence, that is, the content of the text to be detected is fabricated or forged, it is necessary to detect the possibility that the text to be detected has fabricated or forged content. Each paragraph segmented from the submitted text to be detected is input into a trained text detection model, and the trained text detection model outputs the probability that each paragraph is generated by artificial intelligence.

[0055] S40: One of the paragraphs segmented from the text to be detected is used as the first paragraph, and the other paragraphs except the first paragraph from the multiple paragraphs segmented from the text to be detected are used as the second paragraphs; based on the first paragraph, each second paragraph is checked for duplicate text to obtain multiple second paragraphs after the duplicate text is checked; and the similarity between the first paragraph and each second paragraph after the duplicate text is checked is calculated.

[0056] The first paragraph can be the first paragraph in the text to be detected, or any paragraph in the text to be detected. The second paragraph is different from the first paragraph, and can be any paragraph in the text to be detected except the first paragraph, or can be the next paragraph of the first paragraph in the text to be detected.

[0057] Among them, text duplication checking includes text structure duplication checking and text semantic duplication checking. Paragraph text structure duplication checking refers to determining whether the paragraph contains preset words and / or sentences. Specifically, by identifying whether two paragraphs both contain preset words and / or sentences, if both paragraphs contain preset words and / or sentences, it means that there are structurally repeated contents in the two paragraphs, and the preset words and / or sentences are determined as the structurally repeated contents in the two paragraphs.

[0058] Paragraph text semantic duplication checking involves identifying text within a paragraph that has similar meaning. Specifically, the text in the paragraph is converted into text vectors, and the similarity between the text vectors is calculated. A high similarity indicates that the text in the two paragraphs is semantically similar. Text in two paragraphs with similar semantics is considered semantically duplicated content.

[0059] In the embodiment of the present application, it is considered that in the text to be detected, due to the inherent provisions of the document, there may be identical content expressions in the previous and next paragraphs. For these identical contents, it cannot be judged as content plagiarism, simple replacement or piling. Therefore, it is necessary to screen these contents.

[0060] Specifically, before calculating the similarity between the first paragraph and the second paragraph, first determine whether the second paragraph contains structurally duplicated content and semantically duplicated content with the first paragraph. If so, remove the structurally duplicated content and semantically duplicated content in the second paragraph to obtain the second paragraph after the duplicate check, and then calculate the similarity between the first paragraph and the second paragraph after the duplicate check.

[0061] After calculating the similarity between the first paragraph and the second paragraph after checking for plagiarism, select the next second paragraph, check for plagiarism on the next second paragraph, and calculate the similarity between the first paragraph and the next second paragraph after checking for plagiarism, until the similarity between the first paragraph and each second paragraph after checking for plagiarism is obtained.

[0062] S50: Obtain a second detection result for each paragraph according to the similarity between the first paragraph and each second paragraph after the duplicate checking.

[0063] The second detection result is used to indicate the possibility that the paragraph contains plagiarism, simple replacement, and piling.

[0064] In the embodiment of the present application, considering that the paragraph content of the text to be detected contains plagiarism, simple replacement and piling, the similarity between each paragraph will be relatively high. Therefore, for the paragraphs segmented from the text to be detected, the similarity between the first paragraph and each second paragraph after the duplicate check is used to determine whether each paragraph contains plagiarism, simple replacement and piling.

[0065] S60: Determine a text quality detection result of each paragraph according to the first detection result and the second detection result.

[0066] In an embodiment of the present application, the text quality of each paragraph is evaluated based on the possibility of whether each paragraph is generated by artificial intelligence and the possibility of whether each paragraph contains plagiarism, simple replacement, and piling. Specifically, if the possibility of the paragraph content being generated by artificial intelligence is high, the text quality of the paragraph is determined to be unqualified. If the possibility of the paragraph containing plagiarism, simple replacement, and piling is high, the text quality of the paragraph is determined to be unqualified.

[0067] S70: Obtaining a text quality detection result of the text to be detected according to the text quality detection result of each paragraph.

[0068] In an embodiment of the present application, if the text quality of at least one paragraph among the several paragraphs segmented from the text to be detected is unqualified, the text quality of the text to be detected is determined to be unqualified. Specifically, if there is a high possibility that one or some paragraphs are generated by artificial intelligence, then there is a high possibility that the text to be detected is generated by artificial intelligence. If there is a high possibility that one or some paragraphs contain fabricated, forged, plagiarized, or simply replaced content, then there is a high possibility that the text to be detected contains fabricated, forged, plagiarized, or simply replaced content.

[0069] Applying the embodiment of the present application, by obtaining the text to be detected; dividing the text to be detected into several paragraphs; inputting each paragraph into a trained text detection model for text detection, obtaining a first detection result for each paragraph; taking one of the paragraphs segmented from the text to be detected as the first paragraph, and taking the other paragraphs other than the first paragraph from the several paragraphs segmented from the text to be detected as the second paragraph; performing a text duplication check on each second paragraph based on the first paragraph, obtaining several second paragraphs after duplication checking; calculating the similarity between the first paragraph and each second paragraph after duplication checking; obtaining a second detection result for each paragraph based on the similarity between the first paragraph and each second paragraph after duplication checking; determining the text quality detection result of each paragraph based on the first detection result and the second detection result; obtaining the text quality detection result of the text to be detected based on the text quality detection result of each paragraph. The present application can determine the text quality detection result of each paragraph by inputting the paragraph into a trained text detection model for text detection and checking the paragraphs for duplication and calculating the similarity between the paragraphs, thereby determining the text quality detection result of the text to be detected, thereby improving the efficiency and accuracy of text quality detection.

[0070] In an optional embodiment, before step S30, steps S301 to S303 are included, which are specifically as follows:

[0071] S301: Acquire training sample data; wherein the training sample data includes a number of sample paragraphs and a classification label for each sample paragraph.

[0072] In an embodiment of the present application, the training sample data includes several manually written paragraphs and several AI-generated paragraphs. The classification labels of the sample paragraphs include manually written labels and AI-generated labels.

[0073] S302: Extracting feature information of the training data; wherein the feature information includes one of language topic consistency, information density, knowledge coverage, rhetoric, and sentiment analysis.

[0074] In an embodiment of the present application, since there are great differences between manually written texts and texts generated by artificial intelligence in terms of language topic consistency, information density, knowledge coverage, rhetorical techniques and sentiment analysis, the text detection module is trained by extracting the language topic consistency, information density, knowledge coverage, rhetorical techniques and sentiment analysis feature information of the paragraphs.

[0075] S303: Using the feature information as input and the classification label as output, the text detection model to be trained is trained to obtain a trained text detection model.

[0076] In an embodiment of the present application, supervised learning is performed on a text detection model to be trained to obtain a trained text detection model. Specifically, an input vector is generated based on feature information of a paragraph, and the input vector is input into the text detection model to be trained. The text detection model to be trained can be one of a support vector machine, a decision tree, and a neural network. The text detection model to be trained is trained to obtain a trained text detection model.

[0077] By using the feature information and classification labels of the training sample data, a trained text detection model can be obtained automatically and quickly.

[0078] In an alternative embodiment, see Figure 2 In step S40, the step of checking for duplicate text in each second paragraph according to the first paragraph to obtain a plurality of second paragraphs after checking for duplicate text includes steps S401 to S402, which are specifically as follows:

[0079] S401: Perform a text structure check and a text semantic check on each second paragraph to determine the first content in each second paragraph that is repeated in the text structure with the first paragraph and the second content that is repeated in the text semantics with the first paragraph.

[0080] In an embodiment of the present application, in the text to be detected, some descriptions may have fixed expressions, resulting in similar structures of some descriptions. In this case, the fixed expressions can be pre-set, such as the preset wording "specifically including".

[0081] In the text to be detected, there may be some similar paragraphs in the text itself due to fixed content or format requirements. Specifically, in order to achieve the same purpose and obtain the same effect, the specific implementation process is different. That is, some paragraphs in the text to be detected have a causal relationship, the cause and effect are the same, but the specific process of obtaining the effect is different. Therefore, before calculating the similarity of the paragraphs, it is necessary to remove the cause and effect parts in the paragraphs, and only the similarity calculation needs to be performed on the specific process. By identifying the preset designated words in the paragraph, for example, "based on", "the effect is", "beneficial to", etc., the cause part and the result part in the paragraph are determined, and the cause part and the result part in the paragraph are removed to obtain the paragraph after the duplicate check.

[0082] S402: Remove the first content and the second content from each second paragraph to obtain a number of second paragraphs after duplication checking.

[0083] In an embodiment of the present application, after determining the first content in each second paragraph that is structurally similar to the text in the first paragraph and the second content that is semantically similar to the text in the first paragraph, for each second paragraph, the first content and the second content are removed from the second paragraph to obtain a second paragraph after the duplicate check. For example, based on paragraph A, paragraph B is checked for textual duplicates, and the content in paragraph B that is structurally and semantically similar to paragraph A is removed to obtain paragraph B after the duplicate check.

[0084] By checking the text structure and semantics of the second paragraph for plagiarism, the second paragraph after plagiarism checking can be automatically and quickly obtained.

[0085] In an alternative embodiment, see Figure 3 The step of calculating the similarity between the first paragraph and each second paragraph after checking for duplicates in step S40 includes steps S41 to S45, which are specifically as follows:

[0086] S41: Take the first character string of the first paragraph as the current character string, compare the current character string of the first paragraph with each character string of the second paragraph after the duplicate check, if there is a character string in the second paragraph after the duplicate check that is consistent with the current character string of the first paragraph, merge the current character string of the first paragraph with the next character string of the current character string of the first paragraph to obtain a merged character string.

[0087] The character string includes but is not limited to a single character and a single word. The first paragraph and the second paragraph after the duplicate check both include several character strings.

[0088] In the embodiment of the present application, we first start from the first string of the first paragraph, and then search the second paragraph after the duplicate check to see if it contains the first string of the first paragraph. If the second paragraph after the duplicate check contains the first string of the first paragraph, we merge the first string of the first paragraph with the second string of the first paragraph to obtain a merged string. For example, if the first paragraph is "ASDFHGJ" and the second paragraph after the duplicate check is "AQDFGKJ", then the second paragraph after the duplicate check contains the first string "A". We merge the first string "A" of the first paragraph with the second string "S" of the first paragraph to obtain a merged string "AS".

[0089] If the second paragraph after the duplicate check does not contain the first string of the first paragraph, the similarity between the first character of the first paragraph and the second paragraph after the duplicate check is 0. Starting from the second string of the first paragraph, the second paragraph after the duplicate check is searched to see if it contains the second string of the first paragraph.

[0090] S42: Compare the merged character string with each character string in the second paragraph after the duplicate check.

[0091] In an embodiment of the present application, the merged character string is compared one by one with each character string of the second paragraph after the duplicate check to determine whether the merged character string is included in the second paragraph after the duplicate check.

[0092] S43: If there is no string consistent with the merged string in the second paragraph after the duplicate check, obtain the string length of the current string of the first paragraph and the string length of the first paragraph; determine the similarity of the current string of the first paragraph based on the string length of the current string of the first paragraph and the string length of the first paragraph; update the next string of the current string of the first paragraph to the current string, and compare the current string of the first paragraph with each string of the second paragraph until the similarity of each string of the first paragraph with respect to one of the second paragraphs after the duplicate check is determined.

[0093] In an embodiment of the present application, if the second paragraph after the duplicate check does not contain the merged string, the ratio of the string length of the current string (i.e., the string before the merge) to the string length of the first paragraph is determined as the similarity of the current string of the first paragraph. Wherein, the string length of the first paragraph is the sum of the string lengths of all the strings in the first paragraph. For example, if the second paragraph after the duplicate check does not contain the merged string "AS", the string length 1 of the current string (i.e., the first string "A") is divided by the string length 7 of the first paragraph to obtain the similarity of the current string 1 / 7.

[0094] Update the next string of the current string of the first paragraph to the current string, and search the second paragraph after the duplicate check to see if it contains the current string of the first paragraph. Repeat steps S41 to S42 until the similarity of each string of the first paragraph to one of the second paragraphs after the duplicate check is determined. For example, search the second paragraph after the duplicate check to see if it contains the current string "S" of the first paragraph.

[0095] S44: Sum the similarities of the character strings in the first paragraph to obtain the similarity between the first paragraph and one of the second paragraphs after duplication checking.

[0096] In an embodiment of the present application, after obtaining the similarities of the various character strings in the first paragraph, the sum of the similarities of the various character strings is used as the similarity between the first paragraph and one of the second paragraphs after the duplicate check. For example, if the first paragraph is "ASDFHGJ" and the second paragraph after the duplicate check is "AQDFGKJ", then the similarity between the character string "A" in the first paragraph and the second paragraph after the duplicate check is 1 / 7, the similarity between the character string "S" in the first paragraph and the second paragraph after the duplicate check is 0, the similarity between the character string "DF" in the first paragraph and the second paragraph after the duplicate check is 2 / 7, the similarity between the character string "H" in the first paragraph and the second paragraph after the duplicate check is 0, the similarity between the character string "G" in the first paragraph and the second paragraph after the duplicate check is 0, the similarity between the character string "J" in the first paragraph and the second paragraph after the duplicate check is 1 / 7, and the similarity between the first paragraph and the second paragraph after the duplicate check is 4 / 7.

[0097] S45: Select the second paragraph after the next duplicate check and calculate the similarity with the first paragraph, take the first character string of the first paragraph as the current character string, and compare the current character string of the first paragraph with each character string of the second paragraph after the next duplicate check, until the similarity between the first paragraph and each second paragraph after the duplicate check is obtained.

[0098] In an embodiment of the present application, after calculating the similarity between the first paragraph and the second paragraph after a plagiarism check, steps S41 to S44 are used to calculate the similarity between the first paragraph and the second paragraph after the next plagiarism check, until the similarity between the first paragraph and the second paragraphs after each plagiarism check is obtained.

[0099] By calculating the similarity between each character string in the first paragraph and the second paragraph after checking for duplicates, the similarity between the first paragraph and each second paragraph after checking for duplicates can be automatically and quickly obtained.

[0100] In an alternative embodiment, see Figure 4 After step S42, step S46 is included, which is specifically as follows:

[0101] S46: If there is a character string in the second paragraph after the duplicate check that is consistent with the merged character string, update the merged character string to the current character string of the first paragraph, merge the current character string of the first paragraph with the next character string of the current character string of the first paragraph to obtain a merged character string, and compare the merged character string with each character string of the second paragraph after the duplicate check until the similarity between each character string of the first paragraph and one of the second paragraphs after the duplicate check is determined.

[0102] In an embodiment of the present application, if the second paragraph after the duplicate check contains the merged string, the merged string is further merged with the next string, and the merged string is compared with each string of the second paragraph after the duplicate check. If the second paragraph after the duplicate check no longer contains the merged string, step S43 is executed until the similarity between each string of the first paragraph and one of the second paragraphs after the duplicate check is determined. For example, the first paragraph is "ABCDEF", the second paragraph after the duplicate check is "ABPHJK", and the second paragraph after the duplicate check includes the string "A" of the first paragraph, then the string "A" of the first paragraph is merged with the string "B" of the first paragraph to obtain the merged string "AB", the second paragraph after the duplicate check includes the merged string "AB", the string "C" of the first paragraph is merged with the merged string "AB" to obtain the merged string "ABC", the second paragraph after the duplicate check does not include the merged string "ABC", and step S43 is executed.

[0103] By comparing the merged character string with the second paragraph after checking for duplicates, the similarity of each character string of the first paragraph with respect to one of the second paragraphs after checking for duplicates can be determined.

[0104] In an optional embodiment, step S50 includes steps S51 to S53, which are specifically as follows:

[0105] S51: Obtain the highest similarity based on the similarity between the first paragraph and each second paragraph after duplication checking.

[0106] In the embodiment of the present application, after obtaining the similarity between the first paragraph and each second paragraph after checking for duplicate content, the similarities are sorted to obtain the highest similarity. For example, if the similarity between paragraph A and paragraph B after checking for duplicate content is 50%, and the similarity between paragraph A and paragraph C after checking for duplicate content is 65%, then the highest similarity is 65%.

[0107] S52: Obtain a second detection result for each first paragraph according to the difference between the highest similarity and a preset similarity threshold.

[0108] The preset similarity threshold can be set according to human needs. Specifically, the preset similarity threshold is 1.

[0109] In the embodiment of the present application, the highest similarity can be subtracted from the preset similarity threshold to obtain the probability that each first paragraph is manually written. The probability of the first paragraph being manually written is compared with the preset probability threshold to determine the second detection result of the first paragraph.

[0110] S53: Obtain the second detection result of each paragraph according to the second detection result of each first paragraph.

[0111] In the embodiment of the present application, since the first paragraph is one of the paragraphs segmented from the text to be detected, the second detection result of each paragraph is obtained.

[0112] Through the highest similarity and the preset similarity threshold, the second detection result of each paragraph can be automatically and quickly obtained.

[0113] In an optional embodiment, step S53 includes steps S531 to S533, which are specifically as follows:

[0114] S531: If the difference is greater than or equal to a first preset threshold, determine that the second detection result of the first paragraph is a first quality score.

[0115] The first preset threshold and the second preset threshold can be specifically set according to human needs, and the first preset threshold is greater than the second preset threshold. For example, the first preset threshold is 80%, and the second preset threshold is 50%.

[0116] In the embodiment of the present application, if the difference is greater than or equal to 80%, it means that the first paragraph contains fabricated, forged or altered content, experimental data or effects, or the probability of plagiarism or simple replacement is very low, and the first quality score corresponding to the first paragraph is high.

[0117] S532: If the difference is greater than or equal to the second preset threshold and less than the first preset threshold, determine that the second detection result of the first paragraph is a second quality score.

[0118] In the embodiment of the present application, if the difference is greater than or equal to 50% and less than 80%, it means that the probability that the first paragraph contains fabricated, forged or altered content, experimental data or effects, or plagiarized or simply replaced is medium, and the first quality score corresponding to the first paragraph is medium.

[0119] S533: If the difference is less than the second preset threshold, determine that the second detection result of the first paragraph is a third quality score;

[0120] The first preset threshold is greater than the second preset threshold.

[0121] In the embodiment of the present application, if the difference is less than 50%, it means that the first paragraph contains fabricated, forged or altered content, experimental data or effects, or there is a very high probability of plagiarism or simple replacement, and the first quality score corresponding to the first paragraph is low.

[0122] By comparing the difference with the first preset threshold and the second preset threshold, the second detection result of the first paragraph can be automatically and quickly determined.

[0123] In an optional embodiment, step S60 includes steps S61 to S62, which are specifically as follows:

[0124] S61: If the first detection result of the paragraph indicates that the probability that the paragraph is generated by artificial intelligence is greater than a first preset probability threshold, determine that the text quality detection result of the paragraph is unqualified; or

[0125] The first preset probability threshold can be specifically set according to human needs.

[0126] In this embodiment of the present application, if the probability that a paragraph was generated by AI is greater than a first preset probability threshold, it indicates that the paragraph is likely generated by AI, and the paragraph's text quality is determined to be unqualified. If the probability that a paragraph was generated by AI is less than the first preset probability threshold, it indicates that the paragraph is unlikely to be generated by AI, and the paragraph's text quality is determined to be qualified.

[0127] S62: If the second detection result of the paragraph indicates that the probability of plagiarism, simple replacement and piling in the paragraph is greater than a second preset probability threshold, determine that the text quality detection result of the paragraph is unqualified.

[0128] The second preset probability threshold can be specifically set according to human needs.

[0129] In an embodiment of the present application, if the probability that a paragraph is fabricated or forged is greater than a second preset probability threshold, it indicates that there is a high possibility of fabricated, forged or altered content, experimental data or effects, or plagiarism or simple replacement, and the text quality of the paragraph is judged to be unqualified.

[0130] If the probability that a paragraph is fabricated or forged is less than a second preset probability threshold, it indicates that there is fabrication, forgery or alteration of content, experimental data or effects, or the possibility of plagiarism or simple replacement is small, and the text quality of the paragraph is judged to be qualified.

[0131] Whether the text quality of the paragraph is qualified can be determined through the first detection result and the second detection result.

[0132] To verify the effectiveness and reliability of the text quality detection method, the present embodiment prepared test data consisting of real text and fabricated text. The text quality detection method provided in the present embodiment was used to perform text quality detection on the test data, thereby detecting whether the test data was real text or fabricated text. Real text refers to text to be tested that is not generated by artificial intelligence, does not contain plagiarized content, is simply replaced, or is padded. Fabricated text refers to text to be tested that is generated by artificial intelligence, contains plagiarized content, is simply replaced, or is padded.

[0133] (1) Test data

[0134] The test data includes real texts and fabricated texts. Real texts refer to texts that have been rigorously reviewed, approved, and undisputed, selected from the database, covering multiple technical fields such as mechanical engineering, electronic information, and biomedicine, to ensure the diversity and representativeness of real texts. Fabricated texts refer to the generation of a series of fabricated texts through a combination of manual writing and the use of specific text generation tools. The manual writing part is done by professionals who deliberately fabricate some non-existent technical solutions and innovations based on the format and style of real texts; the text generation tool uses existing language models to generate fabricated texts that are somewhat confusing. At the same time, reference is made to some historical text cases that were judged to be fabricated and forged, and they are appropriately adapted and expanded to increase the complexity and challenge of the data.

[0135] The test data is preprocessed, including text cleaning, text segmentation, and annotation. Text cleaning refers to the removal of irrelevant information in the text, such as page numbers, formatting marks, etc., to ensure the neatness and consistency of the data. Text segmentation refers to the division of long texts into paragraph or sentence levels to facilitate subsequent feature extraction and analysis. For some complex content, such as text containing multiple solutions, reasonable segmentation is performed according to the logical structure to highlight the core content of each part. Annotation processing refers to labeling real text as "real" and fabricated text as "forged", providing clear label information for subsequent model training and performance evaluation.

[0136] (2) Testing

[0137] During the testing process, a total of 100 test data were prepared, including 50 real texts and 50 fabricated texts. Each fabricated text contained 1-3 fabricated paragraphs, totaling 113 items. The data covered multiple technical fields such as mechanical engineering, electronic information, and biomedicine. After text cleaning, segmentation, and annotation preprocessing, a test data set with a clear structure and clear labels was formed.

[0138] (3) Test results

[0139] The test results showed that among the real texts, the text quality detection method of this application correctly identified 50 pieces and misidentified 0 pieces as fabricated or forged, with an accuracy rate of 100%. Among the fabricated or forged texts, the text quality detection method of this application correctly identified all 50 pieces as containing fabricated or forged paragraphs, and the number of fabricated or forged paragraphs detected was 107, accounting for 94.69% of the total 113 pieces.

[0140] (4) Test conclusion

[0141] The above test results show that the text quality detection method provided by this application is effective and reliable in identifying real text and fabricated text, and can maintain stable performance in different technical fields and complex text environments.

[0142] See also Figure 5 The text quality detection device 8 provided in the embodiment of the present application includes:

[0143] A to-be-detected text acquisition module 81 is used to acquire the to-be-detected text;

[0144] The to-be-detected text segmentation module 82 is used to segment the to-be-detected text into several paragraphs;

[0145] A first detection result obtaining module 83 is configured to input each paragraph into a trained text detection model for text detection, and obtain a first detection result for each paragraph;

[0146] The similarity calculation module 84 is configured to take one of the paragraphs segmented from the text to be detected as a first paragraph, and take the other paragraphs from the plurality of paragraphs segmented from the text to be detected, excluding the first paragraph, as second paragraphs; perform a text duplication check on each second paragraph based on the first paragraph to obtain a plurality of second paragraphs after duplication checking; and calculate the similarity between the first paragraph and each second paragraph after duplication checking;

[0147] A second detection result obtaining module 85 is configured to obtain a second detection result for each paragraph based on the similarity between the first paragraph and each second paragraph after duplication checking;

[0148] a text quality detection result determination module 86 for determining a text quality detection result of each paragraph based on the first detection result and the second detection result;

[0149] The text quality detection result obtaining module 87 is used to obtain the text quality detection result of the text to be detected based on the text quality detection result of each paragraph.

[0150] Applying the embodiment of the present application, by obtaining the text to be detected; dividing the text to be detected into several paragraphs; inputting each paragraph into a trained text detection model for text detection, obtaining a first detection result for each paragraph; taking one of the paragraphs segmented from the text to be detected as the first paragraph, and taking the other paragraphs other than the first paragraph from the several paragraphs segmented from the text to be detected as the second paragraph; performing a text duplication check on each second paragraph based on the first paragraph, obtaining several second paragraphs after duplication checking; calculating the similarity between the first paragraph and each second paragraph after duplication checking; obtaining a second detection result for each paragraph based on the similarity between the first paragraph and each second paragraph after duplication checking; determining the text quality detection result of each paragraph based on the first detection result and the second detection result; obtaining the text quality detection result of the text to be detected based on the text quality detection result of each paragraph. The present application can determine the text quality detection result of each paragraph by inputting the paragraph into a trained text detection model for text detection and checking the paragraphs for duplication and calculating the similarity between the paragraphs, thereby determining the text quality detection result of the text to be detected, thereby improving the efficiency and accuracy of text quality detection.

[0151] This application also provides a device embodiment that can be used to perform the content of the text quality detection method in the embodiment of this application. For details not disclosed in the device embodiment of this application, please refer to the content of the text quality detection method in the embodiment of this application.

[0152] See also Figure 6 The present application further provides an electronic device 300, which can be a computer, a text quality detection device, etc. In an exemplary embodiment of the present application, the electronic device 300 is a text quality detection device, which includes: at least one processor 301, at least one memory 302, at least one display, at least one network interface 303, a user interface 304, and at least one communication bus 305.

[0153] The user interface 304 is mainly used to provide an input interface for the user and obtain data input by the user. Optionally, the user interface can also include a standard wired interface or a wireless interface.

[0154] The network interface 303 may optionally include a standard wired interface or a wireless interface (such as a WI-FI interface).

[0155] The communication bus 305 is used to realize the connection and communication between these components.

[0156] Among them, the processor 301 may include one or more processing cores. The processor uses various interfaces and lines to connect the various parts of the entire electronic device, and performs various functions of the electronic device and processes data by running or executing instructions, programs, code sets or instruction sets stored in the memory, and calling data stored in the memory. Optionally, the processor can be implemented in the form of at least one hardware of digital signal processing (DSP), field programmable gate array (FPGA), and programmable logic array (PLA). The processor can integrate one or more combinations of a central processing unit (CPU), a graphics processing unit (GPU), and a modem. Among them, the CPU mainly processes the operating system, user interface, and application programs; the GPU is responsible for rendering and drawing the content to be displayed by the display layer; and the modem is used to handle wireless communications. It is understandable that the above-mentioned modem may not be integrated into the processor and may be implemented separately through a chip.

[0157] Among them, the memory 302 may include a random access memory (Random Access Memory, RAM) and may also include a read-only memory (Read-Only Memory). Optionally, the memory includes a non-transitory computer-readable storage medium. The memory can be used to store instructions, programs, codes, code sets or instruction sets. The memory may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area may store data involved in the above-mentioned various method embodiments, etc. The memory may also optionally be at least one storage device located away from the aforementioned processor. As Figure 3 As shown, the memory as a computer storage medium may include an operating system, a network communication module, a user interface module, and an operating application program.

[0158] The processor can be used to call the application of the text quality detection method of the text quality detection device stored in the memory, and specifically execute the method steps of the above-mentioned embodiment. The specific execution process can be referred to the specific description shown in the method embodiment, which will not be repeated here.

[0159] This application also provides a computer-readable storage medium storing a computer program, which contains instructions suitable for being loaded by a processor and executing the method steps of the above-described embodiments. The specific execution process can be referred to the specific description of the embodiments and is not described in detail here. The device containing the storage medium can be a personal computer, laptop computer, smartphone, tablet computer, text quality testing equipment, or other electronic device.

[0160] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to the partial description of the method embodiments. The device embodiments described above are merely illustrative, wherein the components described as separate parts may or may not be physically separated, and the parts shown as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the present application scheme. A person of ordinary skill in the art can understand and implement it without paying any creative work.

[0161] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0162] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 These computer program instructions can also be stored in a computer-readable memory that can guide a computer or other programmable data processing device to work in a specific way, so that the instructions stored in the computer-readable memory produce a product including an instruction device, which implements the function selected in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 function selected in a box or multiple boxes.

[0163] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 steps for the function selected in a box or multiple boxes.

[0164] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0165] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0166] Computer-readable media includes permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media (transitory media), such as modulated data signals and carrier waves.

[0167] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0168] The above are merely embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.

Claims

1. A text quality detection method, characterized in that: include: Get the text to be detected; Segmenting the text to be detected into several paragraphs; Inputting each of the paragraphs into a trained text detection model for text detection to obtain a first detection result for each of the paragraphs; One of the paragraphs segmented from the text to be detected is used as a first paragraph, and the other paragraphs from the multiple paragraphs segmented from the text to be detected, except the first paragraph, are used as second paragraphs; based on the first paragraph, each of the second paragraphs is checked for duplicate text to obtain multiple second paragraphs after the duplicate text is checked; and the similarity between the first paragraph and each of the second paragraphs after the duplicate text is checked is calculated; Obtaining a second detection result for each of the paragraphs based on the similarity between the first paragraph and each of the second paragraphs after the duplicate checking; Determining a text quality detection result of each paragraph according to the first detection result and the second detection result; According to the text quality detection result of each paragraph, the text quality detection result of the text to be detected is obtained.

2. The text quality detection method according to claim 1, characterized in that: The step of performing a text duplication check on each of the second paragraphs according to the first paragraph to obtain a plurality of second paragraphs after duplication checking includes: Performing a text structure check and a text semantic check on each of the second paragraphs to determine first content that is structurally duplicated with the first paragraph and second content that is semantically duplicated with the first paragraph in each of the second paragraphs; The first content and the second content are removed from each of the second paragraphs to obtain a number of second paragraphs after duplication checking.

3. The text quality detection method according to claim 1, wherein: The step of calculating the similarity between the first paragraph and each of the second paragraphs after the duplicate checking comprises: Taking the first string of the first paragraph as the current string, comparing the current string of the first paragraph with each string of one of the second paragraphs after the duplicate check, and if there is a string in the second paragraph after the duplicate check that is consistent with the current string of the first paragraph, merging the current string of the first paragraph with the string next to the current string of the first paragraph to obtain a merged string; Comparing the combined character string with each character string of the second paragraph after the duplicate checking; If there is no string consistent with the merged string in the second paragraph after the duplicate check, obtaining the string length of the current string of the first paragraph and the string length of the first paragraph; determining the similarity of the current string of the first paragraph based on the string length of the current string of the first paragraph and the string length of the first paragraph; updating the next string of the current string of the first paragraph as the current string, and comparing the current string of the first paragraph with each string of the second paragraph until the similarity of each string of the first paragraph with respect to one of the second paragraphs is determined; Summing up the similarities of the character strings in the first paragraph to obtain the similarity between the first paragraph and one of the second paragraphs after the duplicate check; Select the second paragraph after the next plagiarism check and calculate the similarity with the first paragraph, take the first character string of the first paragraph as the current character string, and compare the current character string of the first paragraph with each character string of the second paragraph after the next plagiarism check until the similarity between the first paragraph and each second paragraph after the plagiarism check is obtained.

4. The text quality detection method according to claim 3, wherein: After the step of comparing the combined character string with each character string of the second paragraph after the duplicate check, the method further includes: If there is a string in the second paragraph after the duplicate check that is consistent with the merged string, the merged string is updated to the current string of the first paragraph, the current string of the first paragraph is merged with the next string of the current string of the first paragraph to obtain a merged string, and the merged string is compared with each string of the second paragraph after the duplicate check until the similarity between each string of the first paragraph and one of the second paragraphs after the duplicate check is determined.

5. The text quality detection method according to claim 1, wherein: The step of obtaining a second detection result for each paragraph according to the similarity between the first paragraph and each second paragraph after the duplicate checking comprises: Obtaining a highest similarity based on the similarity between the first paragraph and each of the second paragraphs after the duplicate checking; Obtaining a second detection result for each of the first paragraphs according to a difference between the highest similarity and a preset similarity threshold; According to the second detection result of each of the first paragraphs, a second detection result of each of the paragraphs is obtained.

6. The text quality detection method according to claim 5, characterized in that: The step of obtaining the second detection result of each first paragraph according to the difference between the highest similarity and a preset similarity threshold comprises: If the difference is greater than or equal to a first preset threshold, determining that the second detection result of the first paragraph is a first quality score; If the difference is greater than or equal to a second preset threshold and less than the first preset threshold, determining that the second detection result of the first paragraph is a second quality score; If the difference is less than the second preset threshold, determining that the second detection result of the first paragraph is a third quality score; The first preset threshold is greater than the second preset threshold.

7. The text quality detection method according to any one of claims 1 to 6, characterized in that: The step of determining the text quality detection result of each paragraph according to the first detection result and the second detection result includes: If the first detection result of the paragraph indicates that the probability that the paragraph is generated by artificial intelligence is greater than a first preset probability threshold, determining that the text quality detection result of the paragraph is unqualified; or If the second detection result of the paragraph indicates that the probability of plagiarism, simple replacement and piling in the paragraph is greater than a second preset probability threshold, it is determined that the text quality detection result of the paragraph is unqualified.

8. A text quality detection device, characterized in that: include: A module for acquiring text to be detected, used for acquiring text to be detected; A text segmentation module for segmenting the text to be detected into several paragraphs; a first detection result obtaining module, configured to input each of the paragraphs into a trained text detection model for performing text detection, and obtain a first detection result for each of the paragraphs; A similarity calculation module is configured to take one of the paragraphs segmented from the text to be detected as a first paragraph, and take the other paragraphs, except the first paragraph, from the plurality of paragraphs segmented from the text to be detected as second paragraphs; perform a text duplication check on each of the second paragraphs based on the first paragraph to obtain a plurality of second paragraphs after duplication checking; and calculate a similarity between the first paragraph and each of the second paragraphs after duplication checking; A second detection result obtaining module, configured to obtain a second detection result for each paragraph based on the similarity between the first paragraph and each second paragraph after the duplicate checking; a text quality detection result determination module, configured to determine a text quality detection result of each of the paragraphs based on the first detection result and the second detection result; The text quality detection result obtaining module is used to obtain the text quality detection result of the text to be detected based on the text quality detection result of each paragraph.

9. An electronic device comprising a display, a processor, and a memory; characterized in that: The memory stores a computer program, and the computer program is suitable for being loaded by the processor and executing the text quality detection method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the text quality detection method according to any one of claims 1 to 7 is implemented.

Images