Artificial intelligence text evaluation method and device for short video

Abstract

The invention relates to the technical field of artificial intelligence, in particular to an artificial intelligence text evaluation method and device for a short video, and the method comprises the steps: obtaining semantic information mapped by a to-be-converted text video, and carrying out the clustering of the semantic information; performing clustering processing on the obtained semantic information, performing weighted summation on the semantic information after clustering processing to sequentially calculate a quality index and a resource index, and sequentially generating a quality index priority and a resource index priority according to the quality index and the resource index; and establishing a weight relationship between the quality index priority and the resource index priority, deploying priority weights, then obtaining scoring priority configuration through a formula, and carrying out relative scoring rating on the to-be-converted texts participating in the test from large to small according to the scoring priority configuration. The script writing quality state can be improved, and improvement of high-quality short video content and short video output quality is facilitated.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to an artificial intelligence text evaluation method and device for short videos. Background Art

[0002] With the rapid development of technology, the demand for short video consumption is growing, and people's viewing of short videos has seen a surge in recent years.

[0003] The production of short videos generally involves converting text scripts into video lines to be read out by actors or assisted by narration. However, the quality of short video production cannot be guaranteed due to poor script writing, and the process of converting script-type text into short videos is often inefficient and has poor audience enthusiasm.

[0004] Therefore, how to improve the quality of script writing so as to facilitate subsequent conversion to obtain high-quality short video content is an urgent problem to be solved in this field. Summary of the invention

[0005] To achieve the above objectives, this application provides the following technical solutions:

[0006] According to a first aspect of the present invention, the present invention claims protection for an artificial intelligence text evaluation device for short videos, comprising: an acquisition unit, a processing unit, and an evaluation unit;

[0007] Acquisition unit: used to acquire semantic information of text-video mapping to be converted and cluster the semantic information;

[0008] Processing unit: used for clustering the acquired semantic information, and performing weighted summation on the semantic information after clustering to calculate the quality index and resource index in turn, and generating the quality index priority and resource index priority in turn according to the quality index and resource index;

[0009] Evaluation unit: used to establish a weight relationship between the quality index priority and the resource index priority. After the priority weight is deployed, the scoring priority configuration is obtained through the formula. The texts to be converted participating in the test are relatively scored and rated from large to small according to the scoring priority configuration.

[0010] Furthermore, the acquisition unit acquires grammatical accuracy information and semantic heat information of the text to be converted, the grammatical accuracy information includes part-of-speech accuracy, punctuation accuracy, paragraph accuracy, and the number of modifications in a preset period, and the semantic heat information includes chapter semantic type heat value, chapter semantic character heat value, and chapter semantic plot heat value.

[0011] Furthermore, the processing unit removes the dimensions of the part-of-speech accuracy, punctuation accuracy, paragraph accuracy, and the number of revisions in the preset period, and then calculates the accuracy by the formula The quality index Cl is calculated, where Hp is the part-of-speech accuracy, Ns is the punctuation accuracy, Te is the paragraph accuracy, and Cfu is the number of modifications in the preset period. α1, α2, α3, and α4 are reference factors for the part-of-speech accuracy Hp, the paragraph accuracy Te, the punctuation accuracy Ns, and the number of modifications in the preset period Cfu, respectively, and α1, α2, α3, and α4 are all greater than 0.

[0012] Furthermore, the processing unit removes the dimension of the chapter semantic type heat value, the chapter semantic character heat value and the chapter semantic plot heat value, and then calculates the heat value by the formula Calculate the resource index, where Wa is the heat value of the article semantic type, Ec is the heat value of the article semantic character, De is the heat value of the article semantic plot, β1, β2, β3 are the reference factors of the article semantic type heat value Wa, the article semantic character heat value Ec, and the article semantic plot heat value De, respectively, and β1, β2, β3 are all greater than 0;

[0013] The resource warning value Ct is set, and the resource index Co is compared with the resource warning value Ct. If the resource index Co is less than the resource warning value Ct, the device will issue an early warning.

[0014] Further, the final processing method of the semantic information by the processing unit;

[0015] The quality index Cl is calculated according to the grammatical accuracy information of the text to be converted, and the quality index priority u is established through the quality index. The resource index Co is calculated according to the semantic heat information of the text to be converted, and the resource index priority v is established through the resource index.

[0016] Furthermore, the evaluation unit establishes a strategy for scoring priority configuration;

[0017] The evaluation unit is evaluated by the formula The scoring priority configuration is obtained, where λ1 and λ2 are the reference factors of the quality index priority u and the resource index priority v respectively, and both λ1 and λ2 are greater than 0. According to the scoring priority configuration w, each text to be converted participating in the scoring and rating test is prioritized from large to small to generate a scoring priority table.

[0018] Furthermore, according to the scoring priority configuration, the strategy of scoring the relative priority of the texts to be converted in the test is as follows:

[0019] The higher the scoring priority configuration, the stronger the relative score of the text to be converted. The d texts to be converted participating in the test are defined with a sequence number Y. Where k is a positive integer greater than or equal to 1, and The larger the scoring priority configuration, the larger the sequence number.

[0020] Further, the acquisition strategy of part-of-speech accuracy Hp, punctuation accuracy Ns, paragraph accuracy Te, and preset period modification times Cfu;

[0021] The part-of-speech accuracy is calculated based on the vocabulary mapping table analysis of the word segmentation results of the text to be converted. The formula is: , where No is the correct number of part-of-speech pairs of the text to be converted, and Ps is the associated part-of-speech pair information in the vocabulary mapping table;

[0022] Punctuation accuracy is calculated according to the punctuation terminology standard, and its value is a positive integer greater than 0;

[0023] The paragraph accuracy is calculated based on the paragraph sentence counter and paragraph semantic analyzer analysis within the text to be converted, and its formula is: , starting from the time T0 of the initial segment, and ending at the final time T m Stop timing at t, the total time is t;

[0024] The preset cycle modification times are detected by the listener on the text modification mark. If the text modification mark does not detect the text modification, the preset cycle modification times are 1. If the text is detected, the sample is highlighted and recorded in categories. The listening device of the listener stimulates the sample's identification text, and marks different modified lengths with different colors. The modified length is defined as a serial number A. i , where i is {1, 2, 3...n}, and n is a positive integer greater than 0. The listener measures the density of the highlighted text to calculate the number of modified pages, and defines a sequence number b for each modified page number. j , where j is {1, 2, 3...m}, and m is a positive integer greater than 0, then the preset period modification times , where C is The reference factor of , C is a positive integer greater than 1, and .

[0025] According to the second aspect of the present invention, the present invention claims protection for an artificial intelligence text evaluation method for short videos, which is implemented based on the scoring and rating device, and the method has the following steps:

[0026] S1: Obtain the semantic information of the text-video mapping to be converted and cluster the data;

[0027] S2: clustering the semantic information, calculating the quality index and resource index, and generating an associated priority table;

[0028] S3: Based on different reference factors, the quality index priority and resource index priority can be prioritized and compared according to the associated standards, so as to achieve artificial intelligence text conversion for short videos.

[0029] The present application relates to the field of artificial intelligence technology, and in particular to an artificial intelligence text evaluation method and device for short videos, which obtains semantic information of the text video mapping to be converted and clusters the semantic information; clusters the obtained semantic information, and performs weighted summation on the semantic information after clustering to calculate the quality index and resource index in turn, and generates the quality index priority and resource index priority in turn according to the quality index and resource index; establishes a weight relationship between the quality index priority and the resource index priority, and obtains the scoring priority configuration through a formula after deploying the priority weight, and performs relative scoring and rating of the text to be converted participating in the test from large to small according to the scoring priority configuration. The present invention can improve the writing quality of the script, which is conducive to the improvement of high-quality short video content and the output quality of short videos. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 A structural unit diagram of an artificial intelligence text evaluation device for short videos as claimed in an embodiment of the present application;

[0031] Figure 2 This is a workflow diagram of an artificial intelligence text evaluation device for short videos as claimed in an embodiment of the present application. DETAILED DESCRIPTION

[0032] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0033] The terms "first", "second" and "third" in this application are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features. Therefore, the features defined as "first", "second" and "third" can explicitly or implicitly include at least one of the features. In the description of this application, the meaning of "multiple" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined. All directional indications (such as up, down, left, right, front, back...) in the embodiments of this application are only used to explain the relative position relationship, movement, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication also changes accordingly. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions. For example, a process, method, device, product or equipment that includes a series of steps or units is not limited to the listed steps or units, but optionally also includes steps or units that are not listed, or optionally also includes other steps or units inherent to these processes, methods, products or equipment.

[0034] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0035] Example 1

[0036] like Figure 1 and 2 As shown, this example describes an artificial intelligence text evaluation device for short videos, and the rating device is composed of the following parts: including an acquisition unit, a processing unit and an evaluation unit;

[0037] After the acquisition unit acquires multiple parameters of the text-video mapping to be converted, the parameters are clustered and sent to the processing unit;

[0038] The processing unit establishes a quality index Cl based on the parameters of the text to be converted, and establishes a quality index priority u based on the quality index, establishes a resource index Co based on the parameters of the text to be converted, and establishes a resource index priority v based on the resource index;

[0039] The evaluation unit is evaluated by the formula The scoring priority configuration is obtained, and the various texts to be converted that participate in the scoring and rating test are prioritized from large to small according to the scoring priority configuration w, and a scoring priority table is generated.

[0040] The present invention obtains various data of the text to be converted through an acquisition unit, and establishes a quality index and a resource index for various parameters through formulas, wherein the larger the quality index is, the larger its priority u is, and the smaller the resource index is, the smaller its priority v is. The finally generated scoring priority configuration w is arranged in order from large to small, and the higher the scoring priority configuration is, the larger the sequence number of the text to be converted is, so that the text to be converted is relatively scored and rated.

[0041] The text data to be converted that the acquisition unit needs to acquire include part-of-speech accuracy, punctuation accuracy, paragraph accuracy and the number of changes in a preset period. Each parameter is calibrated as part-of-speech accuracy Hp, punctuation accuracy Ns, paragraph accuracy Te, and the number of changes in a preset period Cfu. The part-of-speech accuracy Hp, punctuation accuracy Ns, paragraph accuracy Te, and the number of changes in a preset period Cfu are clustered, and the quality index Cl is established after removing the unit. The formula is:

[0042]

[0043] In the formula, α1, α2, α3, and α4 are reference factors for part-of-speech accuracy Hp, paragraph accuracy Te, punctuation accuracy Ns, and preset periodic modification times Cfu, respectively, and α1, α2, α3, and α4 are all greater than 0. The specific values ​​of α1, α2, α3, and α4 are set by technical personnel in this field according to the specific model of the text to be converted being tested.

[0044] The present invention obtains the part-of-speech accuracy, punctuation accuracy, paragraph accuracy and the number of preset periodic modification times and then performs clustering processing, removes the units from each data and then performs comprehensive processing, which helps to improve the data processing efficiency.

[0045] In this embodiment, the part-of-speech accuracy is analyzed based on the vocabulary mapping table of the word segmentation result of the text to be converted. The correct number of part-of-speech pairs of the text to be converted is No. The water pressure Ps is obtained according to the vocabulary mapping table data, and the formula is combined The part-of-speech accuracy Hp is calculated;

[0046] The paragraph accuracy is analyzed by the paragraph sentence counter inside the text to be converted, starting from the initial paragraph time T0 to the final paragraph time T m The timing stops at t, the total time is t, and the paragraph accuracy is given by the formula: calculate;

[0047] The preset cycle modification times are detected by the listener on the text modification mark. If the text modification mark does not detect the text modification, the preset cycle modification times are 1. If the text is detected, the sample is highlighted and recorded in categories. The listening device of the listener stimulates the sample's identification text, and marks different modified lengths with different colors. The modified length is defined as a serial number A.i , where i is {1, 2, 3...n}, and n is a positive integer greater than 0. The listener measures the density of the highlighted text to calculate the number of modified pages, and defines a sequence number b for each modified page number. j , where j is {1, 2, 3...m}, and m is a positive integer greater than 0, then the preset period modification times , where C is The reference factor of , C is a positive integer greater than 1, and .

[0048] For the text to be converted, data operations can be performed according to the above parameters. The greater the part-of-speech accuracy, the stronger the cleaning ability of the text to be converted.

[0049] The higher the paragraph accuracy, the better the paragraph distinction function is;

[0050] The higher the accuracy of punctuation marks, the more reasonable the spatial structure of the text to be converted is and the better the semantic interpretation effect is;

[0051] The lower the number of preset cycle modifications, the stronger the coherence of the text to be converted and the better the plot consistency, that is, the better the score.

[0052] The text data to be converted obtained by the acquisition unit includes the chapter semantic type heat value, the chapter semantic character heat value and the chapter semantic plot heat value. The parameters are calibrated as the chapter semantic type heat value Wa, the chapter semantic character heat value Ec, and the chapter semantic plot heat value De. The chapter semantic type heat value Wa, the chapter semantic character heat value Ec, and the chapter semantic plot heat value De are clustered, and the resource index Co is established after removing the unit. The formula is:

[0053]

[0054] In the formula, β1, β2, and β3 are reference factors of the chapter semantic type heat value Wa, the chapter semantic character heat value Ec, and the chapter semantic plot heat value De respectively, and β1, β2, and β3 are all greater than 0. β1, β2, and β3 are set by technical personnel in this field according to the specific model of the text to be converted being tested.

[0055] Set the resource warning value Ct, and compare the resource index Co with the resource warning value Ct. If the resource index Co is less than the resource warning value Ct, the device will issue an early warning, indicating that the text resources to be converted are too large; if the resource index Co is greater than the resource warning value Ct, the device will not issue an early warning.

[0056] In this embodiment, the heat value of the semantic type of the text is obtained by the cumulative flow index of the vocabulary mapping table on the word segmentation result of the text to be converted;

[0057] The semantic character heat value of the article is derived from the search ranking on the search engine;

[0058] The semantic plot heat value of an article is obtained by analyzing the video viewing volume of the short video media network. It should be pointed out that for technical personnel in this field, the sampling test location is determined by the specific situation and meets the consistency of the scoring and rating test conditions.

[0059] Example 2

[0060] The present invention is to score and rate the text to be converted. Therefore, in the above embodiment 1, two different dimensions of artificial intelligence text conversion for short videos are introduced, namely, quality index and resource index. Through comprehensive evaluation of the quality index and resource index, the relative scoring order of the text to be converted is obtained. This example includes the following contents:

[0061] The evaluation unit generates a quality index priority u according to the quality index, and generates a resource index priority v according to the resource index. For the quality index priority u, the larger the quality index, the larger the quality index priority u; for the resource index priority v, the smaller the resource index, the smaller the resource index priority v.

[0062] Priority u and priority v are combined to establish a scoring priority configuration w, which is expressed as:

[0063]

[0064] Where λ1 and λ2 are the reference factors of priority u and priority v respectively. The priority table prioritizes each text to be converted in descending order according to the scoring priority configuration w, so as to determine the relative scoring order of each text to be converted.

[0065] The present invention establishes a quality index and a resource index by acquiring external expression parameters and internal resource parameters of the text to be converted, and generates a quality index priority u and a resource index priority v according to the quality index and the resource index. The quality index priority u and the resource index priority v are used to establish a scoring priority configuration w through a formula. The priority table is arranged in descending order according to the scoring priority configuration w, and can accurately represent the relative strength of the scores between the texts to be converted.

[0066] The higher the scoring priority configuration, the stronger the relative score of the text to be converted. The d texts to be converted participating in the test are defined with a sequence number Y. , where k is a positive integer greater than or equal to 1, and the larger the scoring priority configuration, the larger the defined sequence number.

[0067] In this embodiment, the scoring priority configuration w composed of priority u and priority v is expressed as: The reference factors λ1 and λ2 are 70% and 30% respectively, so the expression of the scoring priority configuration w is: .

[0068] Example 3

[0069] This embodiment describes a method for scoring and evaluating text to be converted, and the method comprises the following steps:

[0070] After the acquisition unit acquires multiple parameters of the text-video mapping to be converted, it clusters the parameters and sends them to the processing unit; the processing unit establishes a quality index Cl with the acquired part of the parameters, and establishes a quality index priority u through the quality index Cl, and then establishes a resource index Co with the acquired part of the parameters, and establishes a resource index priority v through the resource index Co; the evaluation unit uses the formula A scoring priority configuration is obtained, and scoring priorities are performed on each text to be converted in the test from large to small according to the scoring priority configuration w to generate a priority table.

[0071] The above formulas are all dimensionless and numerical calculations. The formula is a formula obtained by obtaining a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0072] In the several embodiments provided in the present application, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of units is only a strategic functional division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another device, 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 an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0073] In addition, each functional unit in each embodiment of the present application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The above integrated unit can be implemented in the form of hardware or in the form of software functional units. The above is only an implementation method of the present application, and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the specification and drawings of this application, or directly or indirectly used in other related technical fields, is also included in the patent protection scope of the present application.

[0074] The specific implementation methods of the invention are described in detail above, but they are only examples, and the present application is not limited to the specific implementation methods described above. For those skilled in the art, any equivalent modification or substitution of the invention is also within the scope of the present application, and therefore, the equalization, modification, and improvement made without departing from the spirit and principle of the present application should be included in the scope of the present application.

Claims

1. An artificial intelligence text evaluation device for short videos, characterized by: It includes an acquisition unit, a processing unit, and an evaluation unit; Acquisition unit: used to acquire semantic information of text-video mapping to be converted and cluster the semantic information; Processing unit: used for clustering the acquired semantic information, and performing weighted summation on the semantic information after clustering to calculate the quality index and resource index in turn, and generating the quality index priority and resource index priority in turn according to the quality index and resource index; Evaluation unit: used to establish a weight relationship between the quality index priority and the resource index priority. After the priority weight is deployed, the scoring priority configuration is obtained through the formula. The texts to be converted participating in the test are relatively scored and rated from large to small according to the scoring priority configuration.

2. The artificial intelligence text evaluation device for short videos according to claim 1, characterized in that: The acquisition unit acquires grammatical accuracy information and semantic heat information of the text to be converted, the grammatical accuracy information includes part of speech accuracy, punctuation accuracy, paragraph accuracy, and the number of modifications in a preset period, and the semantic heat information includes chapter semantic type heat value, chapter semantic character heat value, and chapter semantic plot heat value.

3. The artificial intelligence text evaluation device for short videos according to claim 1, characterized in that: The processing unit removes the dimensions of part of speech accuracy, punctuation accuracy, paragraph accuracy, and the number of revisions in the preset period, and then calculates the accuracy using the formula The quality index Cl is calculated, where Hp is the part-of-speech accuracy, Ns is the punctuation accuracy, Te is the paragraph accuracy, and Cfu is the number of modifications in the preset period. α1, α2, α3, and α4 are reference factors for the part-of-speech accuracy Hp, the paragraph accuracy Te, the punctuation accuracy Ns, and the number of modifications in the preset period Cfu, respectively, and α1, α2, α3, and α4 are all greater than 0.

4. The artificial intelligence text evaluation device for short videos according to claim 1, characterized in that: The processing unit removes the dimension of the article semantic type heat value, the article semantic character heat value and the article semantic plot heat value, and then calculates the heat value through the formula Calculate the resource index, where Wa is the heat value of the article semantic type, Ec is the heat value of the article semantic character, De is the heat value of the article semantic plot, β1, β2, β3 are the reference factors of the article semantic type heat value Wa, the article semantic character heat value Ec, and the article semantic plot heat value De, respectively, and β1, β2, β3 are all greater than 0; The resource warning value Ct is set, and the resource index Co is compared with the resource warning value Ct. If the resource index Co is less than the resource warning value Ct, the device will issue an early warning.

5. The artificial intelligence text evaluation device for short videos according to claim 1, characterized in that: The final processing method of semantic information by the processing unit; The quality index Cl is calculated according to the grammatical accuracy information of the text to be converted, and the quality index priority u is established through the quality index. The resource index Co is calculated according to the semantic heat information of the text to be converted, and the resource index priority v is established through the resource index.

6. An artificial intelligence text evaluation device for short videos as claimed in claim 5, characterized in that: The evaluation unit establishes a strategy for assigning scoring priorities; The evaluation unit is evaluated by the formula The scoring priority configuration is obtained, where λ1 and λ2 are the reference factors of the quality index priority u and the resource index priority v respectively, and both λ1 and λ2 are greater than 0. According to the scoring priority configuration w, each text to be converted participating in the scoring and rating test is prioritized from large to small to generate a scoring priority table.

7. An artificial intelligence text evaluation device for short videos as claimed in claim 6, characterized in that: According to the scoring priority configuration, the strategy for scoring the relative priority of the text to be converted in the test is as follows: The higher the scoring priority configuration, the stronger the relative score of the text to be converted. The d texts to be converted participating in the test are defined with a sequence number Y. Where k is a positive integer greater than or equal to 1, and The larger the scoring priority configuration, the larger the sequence number.

8. The artificial intelligence text evaluation device for short videos as claimed in claim 3, characterized in that: The acquisition strategy of part-of-speech accuracy Hp, punctuation accuracy Ns, paragraph accuracy Te, and preset period modification times Cfu; The part-of-speech accuracy is calculated based on the vocabulary mapping table analysis of the word segmentation results of the text to be converted. The formula is: , where No is the correct number of part-of-speech pairs of the text to be converted, and Ps is the associated part-of-speech pair information in the vocabulary mapping table; Punctuation accuracy is calculated according to the punctuation terminology standard, and its value is a positive integer greater than 0; The paragraph accuracy is calculated based on the paragraph sentence counter and paragraph semantic analyzer analysis within the text to be converted, and its formula is: , starting from the time T0 of the initial segment, and ending at the final time T m Stop timing at t, the total time is t; The preset cycle modification times are detected by the listener on the text modification mark. If the text modification mark does not detect the text modification, the preset cycle modification times are 1. If the text is detected, the sample is highlighted and classified and recorded. The listening device of the listener stimulates the sample's identification text, and marks different modified lengths with different colors. The modified length is defined as a serial number A. i , where i is {1, 2, 3...n}, and n is a positive integer greater than 0. The listener measures the density of the highlighted text to calculate the number of modified pages, and defines a sequence number b for each modified page number. j , where j is {1, 2, 3...m}, and m is a positive integer greater than 0, then the preset period modification times , where C is The reference factor of , C is a positive integer greater than 1, and .

9. An artificial intelligence text evaluation method for short videos, implemented based on the evaluation device according to any one of claims 1 to 8, characterized in that: The method comprises the following steps: S1: Obtain the semantic information of the text-video mapping to be converted and cluster the data; S2: clustering the semantic information, calculating the quality index and resource index, and generating the associated priority table; S3: Based on different reference factors, the quality index priority and resource index priority can be prioritized and compared according to the associated standards, so as to achieve artificial intelligence text conversion for short videos.

Images