A method, device, electronic device and storage medium for monitoring text quality

Abstract

The present application provides a method, device, electronic device and storage medium for monitoring text quality, wherein the monitoring method comprises: after obtaining a target text, determining the repetition of the target text according to the first participle in the target text; inputting the first participle in the target text into a topic model, obtaining the first probability that the target text output by the topic model belongs to each text topic; determining the second probability that the target text belongs to the text field according to the probability that each first participle in the target text belongs to the text field; obtaining multiple first characters in the target text, removing the first characters that meet specific rules from the multiple first characters, and obtaining multiple second characters; determining the quality monitoring result of the target text based on the repetition of the target text, the first probability of belonging to each text topic, the second probability of belonging to the text field and the number of second characters. The present application improves the timeliness and accuracy of text quality monitoring.

Description

Technical Field

[0001] The present application relates to the technical field of text quality detection, and in particular to a text quality monitoring method, device, electronic device and storage medium. Background Art

[0002] At present, with the rapid development of Internet technology, people in their daily lives tend to obtain real-time information through the Internet. In order to ensure the real-time nature of the information, machine writing is usually used to generate text information. Specifically, machine writing algorithms are used to generate corresponding text information based on given materials.

[0003] The quality of text information generated by machine writing algorithms varies greatly, and it is usually necessary to monitor the quality of the text information. In practice, quality evaluation results that match the feedback information are usually generated based on the feedback information provided by readers of the text information. For example, if a text information corresponds to multiple feedback messages about "unclear logic", the quality evaluation result of the text information is determined to be unqualified.

[0004] However, this text information quality monitoring method relies on readers' feedback on the text information, and the quality of the text information can only be determined after the text information is published. The timeliness of text information quality monitoring is poor, and the authenticity and objectivity of the feedback information provided by each reader cannot be guaranteed. The accuracy of the text information quality monitoring results is low. Summary of the invention

[0005] In view of this, the purpose of the embodiments of the present application is to provide a text quality monitoring method, device, electronic device and storage medium, which can complete the monitoring of the target text quality before the target text is published based on four dimensions: repetition, topic relevance, text field relevance and text effective length, thereby improving the timeliness and accuracy of text quality monitoring.

[0006] In a first aspect, an embodiment of the present application provides a method for monitoring text quality, the monitoring method comprising:

[0007] After acquiring the target text, determining the repetition degree of the target text according to the first segmentation word in the target text;

[0008] Inputting the first word segment in the target text into a pre-trained topic model, and obtaining a first probability output by the topic model that the target text belongs to each text topic;

[0009] Determining a second probability that the target text belongs to the text domain according to the probability that each first word segment in the target text belongs to the text domain;

[0010] Acquire multiple first characters in the target text, and remove first characters that meet a specific rule from the multiple first characters to obtain multiple second characters;

[0011] The quality monitoring result of the target text is determined based on the repetition of the target text, the first probability that the target text belongs to each text topic, the second probability that the target text belongs to the text field, and the number of second characters in the target text.

[0012] In a possible implementation, determining the repetition degree of the target text according to the first word segment in the target text includes:

[0013] Obtaining a first word segmentation in a target text, performing a deduplication process on the first word segmentation, and obtaining a second word segmentation;

[0014] The repetition degree of the target text is determined according to the number of the first segmented words and the number of the second segmented words.

[0015] In a possible implementation, the probability that each first word segment in the target text belongs to the text domain is determined in the following manner:

[0016] For each first participle in the target text, the first participle is input into a pre-trained text domain model, and the probability of the first participle belonging to the text domain output by the text domain model is obtained.

[0017] In a possible implementation manner, determining the second probability that the target text belongs to the text domain according to the probability that each first word segment in the target text belongs to the text domain includes:

[0018] The product of the probabilities that each first participle in the target text belongs to the text domain is calculated, and the result of raising the reciprocal of the product to the Nth power is determined as the second probability that the target text belongs to the text domain; wherein N is equal to the number of first participles included in the target text.

[0019] In a possible implementation manner, the first character among the multiple first characters that satisfies a specific rule is determined in the following manner:

[0020] If there are multiple consecutive identical first characters among the multiple first characters of the target text, and the multiple consecutive identical first characters do not belong to a preset standard character combination, then the other first characters among the multiple consecutive identical first characters except the first first character are determined as the first characters that meet the specific rule.

[0021] In a possible implementation, the monitoring method further includes:

[0022] Determine the target text with qualified quality monitoring results as the text to be recommended;

[0023] Determining a priority score of the text to be recommended according to the repetition of the text to be recommended, a first probability that the text to be recommended belongs to each text theme, a second probability that the text to be recommended belongs to the text field, and the number of second characters in the text to be recommended;

[0024] Based on the priority scores of the texts to be recommended, a recommendation order of the texts to be recommended is determined, and based on the recommendation order, the texts to be recommended are recommended to the first user terminal.

[0025] In a possible implementation, the monitoring method further includes:

[0026] Determining the target text with unqualified quality monitoring results as the text to be modified;

[0027] Based on the repetition of the text to be modified, the first probability that the text to be modified belongs to each text topic, the second probability that the text to be modified belongs to the text field, and the number of second characters in the text to be modified, modification prompt information matching the text to be modified is generated, and the generated modification prompt information is sent to the second user terminal corresponding to the text to be modified.

[0028] In a second aspect, an embodiment of the present application provides a device for monitoring text quality, the monitoring device comprising:

[0029] A first determination module is used to determine the repetition degree of the target text according to the first word segmentation in the target text after acquiring the target text;

[0030] A first acquisition module, used for inputting the first word segmentation in the target text into a pre-trained topic model, and obtaining a first probability that the target text belongs to each text topic output by the topic model;

[0031] A second determination module, configured to determine a second probability that the target text belongs to the text domain according to the probability that each first word segment in the target text belongs to the text domain;

[0032] A removal module is used to obtain a plurality of first characters in the target text, remove the first characters that meet a specific rule among the plurality of first characters, and obtain a plurality of second characters;

[0033] The third determination module is used to determine the quality monitoring result of the target text based on the repetition of the target text, the first probability that the target text belongs to each text topic, the second probability that the target text belongs to the text field, and the number of second characters in the target text.

[0034] In a third aspect, an embodiment of the present application provides an electronic device, comprising: a processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor communicates with the memory via the bus, and the processor executes the machine-readable instructions to perform the steps of the text quality monitoring method described in any one of the first aspects.

[0035] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the text quality monitoring method described in any one of the first aspects are executed.

[0036] A text quality monitoring method, device, electronic device and storage medium provided by the embodiment of the present application, after obtaining the target text, according to the first participle in the target text, determine the repetition of the target text; input the first participle in the target text into the pre-trained topic model, obtain the first probability that the target text output by the topic model belongs to each text theme; according to the probability that each first participle in the target text belongs to the text field, determine the second probability that the target text belongs to the text field; obtain multiple first characters in the target text, remove the first characters that meet specific rules in the multiple first characters, and obtain multiple second characters; based on the repetition of the target text, the first probability that the target text belongs to each text theme, the second probability that the target text belongs to the text field and the number of second characters in the target text, determine the quality monitoring result of the target text. The embodiment of the present application can complete the monitoring of the quality of the target text before the target text is released based on the four dimensions of repetition, topic relevance, text field relevance and text effective length, thereby improving the timeliness and accuracy of text quality monitoring.

[0037] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, preferred embodiments are specifically cited below and described in detail with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0039] Figure 1 A flowchart of a text quality monitoring method provided by an embodiment of the present application is shown;

[0040] Figure 2 A flowchart of another text quality monitoring method provided in an embodiment of the present application is shown;

[0041] Figure 3 A flowchart of another text quality monitoring method provided in an embodiment of the present application is shown;

[0042] Figure 4 A schematic diagram of the structure of a text quality monitoring device provided in an embodiment of the present application is shown;

[0043] Figure 5 A schematic diagram of an electronic device provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0044] To make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. The components of the embodiments of the present application generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.

[0045] At present, with the rapid development of Internet technology, people are more inclined to obtain real-time information through the Internet in their daily lives. In order to ensure the real-time nature of the information, machine writing is usually used to generate text information. Specifically, the corresponding text information is generated based on the given materials using machine writing algorithms. The quality of text information generated by machine writing algorithms varies, and it is usually necessary to monitor the quality of the text information. In practice, quality evaluation results that match the feedback information are usually generated based on the feedback information provided by readers who read the text information. For example, if a certain text information corresponds to multiple feedback messages about "unclear logic", the quality evaluation result of the text information is determined to be unqualified.

[0046] However, this text information quality monitoring method relies on readers' feedback on the text information, and the quality of the text information can only be determined after the text information is published. The timeliness of text information quality monitoring is poor, and the authenticity and objectivity of the feedback information provided by each reader cannot be guaranteed. The accuracy of the text information quality monitoring results is low.

[0047] Based on the above problems, the embodiment of the present application provides a method, device, electronic device and storage medium for monitoring text quality. After obtaining the target text, the repetition of the target text is determined according to the first participle in the target text; the first participle in the target text is input into the pre-trained topic model to obtain the first probability that the target text belongs to each text topic output by the topic model; according to the probability that each first participle in the target text belongs to the text field, the second probability that the target text belongs to the text field is determined; multiple first characters in the target text are obtained, and the first characters that meet specific rules in the multiple first characters are eliminated to obtain multiple second characters; based on the repetition of the target text, the first probability that the target text belongs to each text topic, the second probability that the target text belongs to the text field and the number of second characters in the target text, the quality monitoring result of the target text is determined. The embodiment of the present application can complete the monitoring of the quality of the target text before the target text is released based on the four dimensions of repetition, topic relevance, text field relevance and text effective length, thereby improving the timeliness and accuracy of text quality monitoring.

[0048] The defects existing in the above solutions are the results obtained by the inventor after practice and careful research. Therefore, the discovery process of the above problems and the solutions proposed by this application for the above problems below should be the contributions made by the inventor to this application during the application process.

[0049] The technical solutions in the present application will be clearly and completely described below in conjunction with the drawings in the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The components of the present application generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the present application for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.

[0050] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0051] To facilitate understanding of this embodiment, a text quality monitoring method disclosed in the embodiment of the present application is first introduced in detail. The executor of the text quality monitoring method provided in the embodiment of the present application is a computer server.

[0052] See also Figure 1 As shown, Figure 1 A flowchart of a method for monitoring text quality provided in an embodiment of the present application, the monitoring method comprising the following steps:

[0053] S101 . After acquiring a target text, determine the repetition degree of the target text according to the first word segmentation in the target text.

[0054] In an embodiment of the present application, the target text may be a text generated by machine writing or a manually generated text. The target text is segmented to obtain a plurality of first segmentations included in the target text, wherein each first segmentation is associated with a text field to which the target text belongs and a text subject to which the target text belongs. The repetition degree of the target text is determined according to the number of first segmentations included in the target text and the number of times each first segmentation appears. The more times the same first segmentation appears, the higher the repetition degree of the target text.

[0055] In practice, the repetition of the target text can be quantitatively analyzed in the following ways:

[0056] 1. Obtain the first word segmentation in the target text, perform deduplication processing on the first word segmentation, and obtain the second word segmentation.

[0057] 2. Determine the repetition degree of the target text according to the number of the first participles and the number of the second participles.

[0058] The multiple first participles in the target text may include the same first participle. For example, in the target text 1 "I like apples, apples are delicious", the first participle "apple" appears twice. In order to determine the repetition of the target text, the multiple first participles included in the target text are deduplicated to obtain different second participles. For example, the second participles of the above target text 1 include: I, like, apple, delicious.

[0059] Then, the repetition of the target text is determined according to the number of the first segmented words included in the target text and the number of the second segmented words included in the target text. Specifically, the repetition of the target text is determined as the ratio of the number of the first segmented words included in the target text to the number of the second segmented words included in the target text. For example, the repetition of the target text 1 is equal to 1.25 (5 / 4).

[0060] In practice, a first threshold is pre-set according to actual needs of the user. If the repetition degree of the target text is greater than the first threshold, the target text is determined to be too repetitive and the quality monitoring result of the target text is unqualified. Otherwise, the target text is determined to be qualified in the monitoring dimension of "repetition degree". Optionally, the first threshold is equal to 2.5.

[0061] S102: Input the first word segment in the target text into a pre-trained topic model, and obtain a first probability output by the topic model that the target text belongs to each text topic.

[0062] In an embodiment of the present application, multiple first word segmentations included in the target text are input into a pre-trained topic model. The output result of the pre-trained topic model is a vector, which includes multiple elements, each element corresponds to a text topic, and the value of each element is the first probability that the target text belongs to the text topic corresponding to the element. After the topic model is trained, the specific meaning of each text topic can be determined, such as food topics and clothing topics.

[0063] For example, the output result of topic model 1 is a vector consisting of four elements. The first element corresponds to the text topic "food", the second element corresponds to the text topic "clothing", the third element corresponds to the text topic "car", and the fourth element corresponds to the text topic "house". Multiple first participles of target text 1 are input into topic model 1, and the output result of topic model 1 is (0.7, 0.1, 0.1, 0.1), that is, the first probabilities of target text 1 belonging to the text topics "food", "clothing", "car", and "house" are 0.7, 0.1, 0.1, and 0.1, respectively.

[0064] Optionally, multiple word segments included in the target text are written into a txt file separated by spaces, and the txt file is input into a pre-trained topic model to obtain a first probability that the target text belongs to each text topic.

[0065] Whether it is machine writing or human writing, text creation usually revolves around a specific theme. For a specific creative theme, when the text theme deviates too much from the creative theme, the quality of the generated text is considered to be unqualified. For example, the computer is commanded to write around the theme of cosmetics, but the generated text content is about dairy products. At this time, it is considered that the text is off-topic and the text quality is unqualified. Therefore, based on the first probability of the target text belonging to each text theme output by the topic model, it is determined whether the text theme corresponding to the largest first probability is consistent with the pre-set creative theme. If not, the quality monitoring result of the target text is determined to be unqualified. If consistent, the target text is determined to be qualified in the monitoring dimension of "topic relevance".

[0066] S103 . Determine a second probability that the target text belongs to the text domain according to the probability that each first word segment in the target text belongs to the text domain.

[0067] In an embodiment of the present application, for a specific text field, different target texts have different probabilities of belonging to the specific text field. For example, for the automobile field, the target text 2 "Tesla is delicious" is less likely to appear than the target text 3 "Tesla is an electric car". When determining the possibility that the target text belongs to a specific text field (hereinafter referred to as the text field), the probability of each first participle in the target text belonging to the text field is first determined, that is, the possibility of each first participle in the target text appearing in the text field, and then based on the probability corresponding to each first participle in the target text, the second probability of the target text belonging to the text field is determined, that is, the impossibility of the target text appearing in the text field. Here, the larger the second probability, the lower the possibility of the target text appearing in the text field.

[0068] Specifically, the probability that each first word in the target text belongs to the text field is determined in the following manner:

[0069] For each first participle in the target text, the first participle is input into a pre-trained text domain model, and the probability of the first participle belonging to the text domain output by the text domain model is obtained.

[0070] Different text domain models are trained for different text domains. For example, after model training, text domain model 1 corresponding to the automobile domain and text domain model 2 corresponding to the housing domain are obtained. The first segmentation word 1 "Tesla" is input into text domain model 1 and text domain model 2 respectively to obtain probability 1 that the first segmentation word 1 belongs to the automobile domain and probability 2 that the segmentation word 1 belongs to the housing domain. Here, probability 1 is greater than probability 2, that is, the probability of the first segmentation word 1 "Tesla" appearing in the automobile domain is greater than the probability of its appearing in the housing domain. As a possible implementation method, the above text domain model is an N-gram model.

[0071] After determining the text domain, a matching pre-trained text domain model is selected, and each first word segment in the target text is input into the above text domain model respectively to obtain the probability that each first word segment in the target text output by the above text domain model belongs to the text domain.

[0072] After determining the probability that each first word in the target text belongs to the text domain, the perplexity formula is usually used to determine the second probability that the target text belongs to the text domain:

[0073] The product of the probabilities that each first participle in the target text belongs to the text domain is calculated, and the result of raising the reciprocal of the product to the Nth power is determined as the second probability that the target text belongs to the text domain; wherein N is equal to the number of first participles included in the target text.

[0074] Specifically, the calculation formula for the second probability (perplexity) that the target text belongs to the text field is:

[0075]

[0076] The target text includes N first participles, and the probability of each first participle belonging to the text field is p(w i ), where i∈[1,N], PP(S) is the second probability that the target text belongs to the text domain.

[0077] In practice, the second threshold is pre-set according to the actual needs of the user. If the second probability that the target text belongs to the text field is greater than the second threshold, it is determined that the possibility of the target text appearing in the text field is low, and the quality monitoring result of the target text is unqualified. Otherwise, it is determined that the target text is qualified in the monitoring dimension of "text field relevance". Optionally, the second threshold is equal to 200.

[0078] S104: Acquire multiple first characters in the target text, and remove first characters that meet specific rules from the multiple first characters to obtain multiple second characters.

[0079] In the embodiments of the present application, a character refers to letters, numbers, words, and symbols used in a computer, including: "1", "2", "3", "A", "B", "C", ",", ".", etc., which are characters representing meanings. The target text includes multiple first characters. For example, there is the target text 4 "天气气真好,,大家都高高兴兴。", and the first characters in the target text 4 include: "天", "气", "气", "真", "好", ",", ",", "大", "家", "都", "高", "高", "兴", "兴", "。". To determine the effective text length of the target text, it is necessary to remove the first characters in the target text that meet specific rules, that is, remove the continuously repeated characters to obtain the second characters of the target text.

[0080] Specifically, the first characters that meet the specific rules among the multiple first characters are determined in the following manner:

[0081] If there are multiple consecutive identical first characters among the multiple first characters of the target text, and the multiple consecutive identical first characters do not belong to a preset standard character combination, then the other first characters except the first first character among the multiple consecutive identical first characters are determined as the first characters that meet the specific rules.

[0082] Among them, the preset standard character combination is a word segmentation containing reduplicated words. For example, 高高兴兴 (happy), 黄灿灿 (bright yellow), 黑乎乎 (black), etc. For the above target text 4, there are two consecutive first characters "气", and these two consecutive first characters do not belong to the preset standard character combination, so the second "气" is removed; there are two consecutive first characters ",", and these two consecutive first characters do not belong to the preset standard character combination, so the second "," is removed; there are two consecutive first characters "高", and two consecutive first characters "兴", and 高高兴兴 belongs to the preset standard character combination, so 高高兴兴 is not modified. The second characters of the target text 4 are obtained, including: "天", "气", "真", "好", ",", "大", "家", "都", "高", "高", "兴", "兴", "。".

[0083] In practice, a third threshold is preset according to the actual needs of the user. If the number of second characters included in the target text is greater than the third threshold, it is determined that the effective text length of the target text is unqualified, that is, the quality monitoring result of the target text is unqualified. Otherwise, it is determined that the target text is qualified in the monitoring dimension of "effective text length". Optionally, the third threshold is equal to half of the number of first characters included in the target text.

[0084] S105, determining a quality monitoring result of the target text based on the repetition of the target text, the first probability that the target text belongs to each text topic, the second probability that the target text belongs to the text field, and the number of second characters in the target text.

[0085] In the embodiment of the present application, the quality monitoring results of the target text include qualified and unqualified.

[0086] As a possible implementation, if the repetition of the target text is less than or equal to the above-mentioned first threshold, and the text theme corresponding to the largest first probability of the target text is consistent with the creation theme, and the second probability that the target text belongs to the text field is less than or equal to the second threshold, and the number of second characters in the target text is greater than or equal to the third threshold, then the quality monitoring result of the target text is determined to be qualified; otherwise, the quality monitoring result of the target text is determined to be unqualified.

[0087] As another possible implementation, based on the repetition of the target text, the first probability that the target text belongs to each text topic, the second probability that the target text belongs to the text field, and the number of second characters in the target text, the corresponding relationship between each monitoring result and the monitoring score under the four dimensions of repetition, topic relevance, text field relevance, and text effective length is determined. If the monitoring score is greater than a pre-set fourth threshold, the quality monitoring result of the target text is determined to be qualified, otherwise, the quality monitoring result of the target text is determined to be unqualified. Among them, the above-mentioned fourth threshold can be set according to the actual needs of the user.

[0088] The text quality monitoring method provided in the embodiment of the present application can monitor the quality of the target text before the target text is released based on four dimensions: repetition, topic relevance, text field relevance, and text effective length, thereby improving the timeliness and accuracy of text quality monitoring.

[0089] For further information, see Figure 2 As shown, in the text quality monitoring method provided in the embodiment of the present application, the monitoring method also includes:

[0090] S201: Determine the target text with qualified quality monitoring result as the text to be recommended.

[0091] In the embodiment of the present application, only target texts with qualified quality are recommended to readers. Therefore, after determining the quality monitoring result of the target text, it is judged whether the quality monitoring result of the target text is qualified. If so, the target text is determined as the text to be recommended.

[0092] S202, determining a priority score of the text to be recommended according to the repetition of the text to be recommended, a first probability that the text to be recommended belongs to each text theme, a second probability that the text to be recommended belongs to the text field, and the number of second characters in the text to be recommended.

[0093] In an embodiment of the present application, there are multiple texts to be recommended. Before recommending the texts to be recommended to readers, the multiple texts to be recommended need to be sorted. Specifically, based on the repetition of the texts to be recommended, the first probability that the texts to be recommended belong to each text theme, the second probability that the texts to be recommended belong to the text field, and the number of second characters in the texts to be recommended, according to the correspondence between each monitoring result and the priority score under the four dimensions of repetition, theme relevance, text field relevance, and text effective length, the priority score corresponding to the target text is determined.

[0094] S203: Determine a recommendation order of the texts to be recommended based on the priority scores of the texts to be recommended, and recommend the texts to be recommended to the first user terminal based on the recommendation order.

[0095] In the embodiment of the present application, multiple texts to be recommended are sorted according to the priority score of each text to be recommended, and the higher the priority score of the text to be recommended, the higher the recommendation order of the text to be recommended. The multiple texts to be recommended are recommended to the first user terminal in sequence according to the recommendation order, wherein the first user terminal is the terminal corresponding to the reader.

[0096] For further information, see Figure 3 As shown, in the text quality monitoring method provided in the embodiment of the present application, the monitoring method also includes:

[0097] S301, determining the target text with unqualified quality monitoring results as the text to be modified.

[0098] In an embodiment of the present application, modification prompt information can be provided for a target text with unqualified quality. After determining the quality monitoring result of the target text, it is determined whether the quality monitoring result of the target text is unqualified. If so, the target text is determined as a text to be modified.

[0099] S302. Generate modification prompt information matching the text to be modified based on the repetition of the text to be modified, the first probability that the text to be modified belongs to each text topic, the second probability that the text to be modified belongs to the text field, and the number of second characters in the text to be modified, and send the generated modification prompt information to the second user terminal corresponding to the text to be modified.

[0100] In an embodiment of the present application, for each text to be modified, if the repetition of the text to be modified is greater than the above-mentioned first threshold, it is determined that the repetition of the text to be modified is unqualified, and a first modification prompt information matching the repetition is generated; if the text theme corresponding to the largest first probability of the text to be modified is inconsistent with the creation theme, it is determined that the theme relevance of the text to be modified is unqualified, and a second modification prompt information matching the theme relevance is generated; if the second probability that the text to be modified belongs to the text field is greater than the second threshold, it is determined that the field relevance of the text to be modified is unqualified, and a third modification prompt information matching the field relevance is generated; if the number of second characters in the text to be modified is less than the third threshold, it is determined that the effective length of the text to be modified is unqualified, and a fourth modification prompt information matching the effective length of the text is generated, and at least one modification prompt information corresponding to the text to be modified is sent to the second user terminal, wherein the modification prompt information includes the first modification prompt information, the second modification prompt information, the third modification prompt information, and the fourth modification prompt information. Here, if the target text is a text generated by machine writing, the second user terminal is the terminal corresponding to the staff who monitors the text quality, or the terminal corresponding to the staff who develops the machine algorithm; if the target text is a manually generated text, the second user terminal is the terminal corresponding to the author of the target text.

[0101] Based on the same inventive concept, the embodiment of the present application also provides a text quality monitoring device corresponding to the text quality monitoring method. Since the principle of solving the problem by the device in the embodiment of the present application is similar to the above-mentioned text quality monitoring method in the embodiment of the present application, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be repeated.

[0102] Reference Figure 4 As shown, Figure 4 A schematic diagram of a text quality monitoring device provided in an embodiment of the present application, the monitoring device comprising:

[0103] A first determination module 401 is used to determine the repetition degree of the target text according to the first word segmentation in the target text after acquiring the target text;

[0104] A first acquisition module 402 is used to input the first word segmentation in the target text into a pre-trained topic model, and obtain a first probability that the target text belongs to each text topic output by the topic model;

[0105] A second determination module 403 is used to determine a second probability that the target text belongs to the text field according to the probability that each first word segment in the target text belongs to the text field;

[0106] A removal module 404 is used to obtain a plurality of first characters in the target text, and remove the first characters that meet a specific rule from the plurality of first characters to obtain a plurality of second characters;

[0107] The third determination module 405 is used to determine the quality monitoring result of the target text based on the repetition of the target text, the first probability that the target text belongs to each text topic, the second probability that the target text belongs to the text field, and the number of second characters in the target text.

[0108] In a possible implementation manner, when the first determination module 401 determines the repetition degree of the target text according to the first segmentation word in the target text, it includes:

[0109] Obtaining a first word segmentation in a target text, performing a deduplication process on the first word segmentation, and obtaining a second word segmentation;

[0110] The repetition degree of the target text is determined according to the number of the first segmented words and the number of the second segmented words.

[0111] In a possible implementation manner, the text quality monitoring device further includes:

[0112] The second acquisition module is used to input each first participle in the target text into a pre-trained text domain model to obtain the probability of the first participle belonging to the text domain output by the text domain model.

[0113] In a possible implementation manner, the second determination module 403, when determining the second probability that the target text belongs to the text domain according to the probability that each first word segment in the target text belongs to the text domain, includes:

[0114] The product of the probabilities that each first participle in the target text belongs to the text domain is calculated, and the result of raising the reciprocal of the product to the Nth power is determined as the second probability that the target text belongs to the text domain; wherein N is equal to the number of first participles included in the target text.

[0115] In a possible implementation manner, the text quality monitoring device further includes:

[0116] The fourth determination module is used to determine the other first characters among the multiple first characters of the target text except the first first character as the first characters that meet the specific rule if there are multiple consecutive identical first characters among the multiple first characters of the target text, and the multiple consecutive identical first characters do not belong to a preset standard character combination.

[0117] In a possible implementation manner, the text quality monitoring device further includes:

[0118] A fifth determination module, configured to determine the target text with a qualified quality monitoring result as a text to be recommended;

[0119] a sixth determination module, configured to determine a priority score of the text to be recommended according to the repetition of the text to be recommended, a first probability that the text to be recommended belongs to each text theme, a second probability that the text to be recommended belongs to the text field, and the number of second characters in the text to be recommended;

[0120] A seventh determination module, configured to determine a recommendation order of the texts to be recommended based on the priority scores of the texts to be recommended;

[0121] A recommendation module is used to recommend the text to be recommended to the first user terminal based on the recommendation order.

[0122] In a possible implementation manner, the text quality monitoring device further includes:

[0123] An eighth determination module, configured to determine the target text whose quality monitoring result is unqualified as the text to be modified;

[0124] A generating module, for generating modification prompt information matching the text to be modified according to the repetition of the text to be modified, the first probability that the text to be modified belongs to each text theme, the second probability that the text to be modified belongs to the text field, and the number of second characters in the text to be modified;

[0125] The sending module is used to send the generated modification prompt information to the second user terminal corresponding to the text to be modified.

[0126] The text quality monitoring device provided in the embodiment of the present application can monitor the quality of the target text before the target text is released based on four dimensions: repetition, topic relevance, text field relevance, and text effective length, thereby improving the timeliness and accuracy of text quality monitoring.

[0127] See also Figure 5 As shown, Figure 5 An electronic device 500 is provided for an embodiment of the present application, and the electronic device 500 includes: a processor 501, a memory 502 and a bus, wherein the memory 502 stores machine-readable instructions executable by the processor 501, and when the electronic device is running, the processor 501 communicates with the memory 502 via the bus, and the processor 501 executes the machine-readable instructions to perform the steps of the text quality monitoring method as described above.

[0128] Specifically, the above-mentioned memory 502 and processor 501 can be general-purpose memories and processors, which are not specifically limited here. When the processor 501 runs the computer program stored in the memory 502, the above-mentioned text quality monitoring method can be executed.

[0129] Corresponding to the above-mentioned text quality monitoring method, an embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned text quality monitoring method are executed.

[0130] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, the specific working process of the system and device described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here. In the several embodiments provided in the present application, it should be understood that the disclosed system, device and method can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the modules is only a logical function division. There may be other division methods in actual implementation. For example, multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0131] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, that is, they may be located in one place or distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0132] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0133] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium that can be executed by a processor. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0134] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present application, which are used to illustrate the technical solutions of the present application, rather than to limit them. The protection scope of the present application is not limited thereto. Although the present application is described in detail with reference to the above-mentioned embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-mentioned embodiments within the technical scope disclosed in the present application, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A method for monitoring text quality, characterized in that: The monitoring method comprises: After acquiring the target text, determining the repetition degree of the target text according to the first segmentation word in the target text; Inputting the first word segment in the target text into a pre-trained topic model, and obtaining a first probability output by the topic model that the target text belongs to each text topic; Determining a second probability that the target text belongs to the text field according to the probability that each first participle in the target text belongs to the text field includes: calculating the product of the probabilities that each first participle in the target text belongs to the text field, and determining the result of raising the reciprocal of the product to the Nth power as the second probability that the target text belongs to the text field; wherein N is equal to the number of first participles included in the target text; Acquire multiple first characters in the target text, and remove first characters that meet a specific rule from the multiple first characters to obtain multiple second characters; Determining a quality monitoring result of the target text based on the repetitiveness of the target text, a first probability that the target text belongs to each text topic, a second probability that the target text belongs to the text field, and the number of second characters in the target text; Determining the first character that satisfies a specific rule among the multiple first characters includes: if there are multiple consecutive identical first characters among the multiple first characters of the target text, and the multiple consecutive identical first characters do not belong to a preset standard character combination, then determining the other first characters among the multiple consecutive identical first characters except the first first character as the first characters that satisfy the specific rule.

2. The text quality monitoring method according to claim 1, characterized in that: The step of determining the repetition degree of the target text according to the first word segment in the target text includes: Obtaining a first word segmentation in a target text, performing a deduplication process on the first word segmentation, and obtaining a second word segmentation; The repetition degree of the target text is determined according to the number of the first segmented words and the number of the second segmented words.

3. The text quality monitoring method according to claim 1, characterized in that: The probability that each first word in the target text belongs to the text domain is determined by: For each first participle in the target text, the first participle is input into a pre-trained text domain model, and the probability of the first participle belonging to the text domain output by the text domain model is obtained.

4. The text quality monitoring method according to claim 1, characterized in that: The monitoring method further comprises: Determine the target text with qualified quality monitoring results as the text to be recommended; Determining a priority score of the text to be recommended according to the repetition of the text to be recommended, a first probability that the text to be recommended belongs to each text theme, a second probability that the text to be recommended belongs to the text field, and the number of second characters in the text to be recommended; Based on the priority scores of the texts to be recommended, a recommendation order of the texts to be recommended is determined, and based on the recommendation order, the texts to be recommended are recommended to the first user terminal.

5. The text quality monitoring method according to claim 1, characterized in that: The monitoring method further comprises: Determining the target text with unqualified quality monitoring results as the text to be modified; Based on the repetition of the text to be modified, the first probability that the text to be modified belongs to each text topic, the second probability that the text to be modified belongs to the text field, and the number of second characters in the text to be modified, modification prompt information matching the text to be modified is generated, and the generated modification prompt information is sent to the second user terminal corresponding to the text to be modified.

6. A device for monitoring text quality, characterized in that: The monitoring device comprises: A first determination module is used to determine the repetition degree of the target text according to the first word segmentation in the target text after acquiring the target text; A first acquisition module, used for inputting the first word segmentation in the target text into a pre-trained topic model, and obtaining a first probability that the target text belongs to each text topic output by the topic model; The second determination module is used to determine the second probability that the target text belongs to the text field according to the probability that each first participle in the target text belongs to the text field, including: calculating the product of the probability that each first participle in the target text belongs to the text field, and determining the result after the reciprocal of the product is raised to the Nth power as the second probability that the target text belongs to the text field; wherein N is equal to the number of first participles included in the target text; A removal module is used to obtain a plurality of first characters in the target text, remove the first characters that meet a specific rule among the plurality of first characters, and obtain a plurality of second characters; A third determination module is used to determine the quality monitoring result of the target text based on the repetitiveness of the target text, the first probability that the target text belongs to each text topic, the second probability that the target text belongs to the text field, and the number of second characters in the target text; Determining the first character that satisfies a specific rule among the multiple first characters includes: if there are multiple consecutive identical first characters among the multiple first characters of the target text, and the multiple consecutive identical first characters do not belong to a preset standard character combination, then determining the other first characters among the multiple consecutive identical first characters except the first first character as the first characters that satisfy the specific rule.

7. An electronic device, characterized in that: include: A processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor communicates with the memory via the bus, and the processor executes the machine-readable instructions to perform the steps of the text quality monitoring method as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which, when executed by a processor, executes the steps of the text quality monitoring method according to any one of claims 1 to 5.

Images