Stroma text evaluation method, apparatus and device, and readable storage medium

Abstract

The invention relates to the technical field of data processing, in particular to a story text evaluation method and device, equipment and a readable storage medium. According to the method, the evaluation result of each plot plot in the plot text is quantified into the plot grade score through the role description information and the plot summary information, and the plot distribution trend capable of reflecting the overall trend of the plot grade score is further obtained; and the quality of the plot text is measured by using the similarity between the plot distribution trend and the preset plot distribution index, so that the objectivity and intuitiveness of plot text evaluation are improved.

Description

Technical Field

[0001] The present disclosure relates to the field of data processing technology, and in particular to a method, device, equipment and readable storage medium for evaluating a plot text. Background Art

[0002] In the film and television industry, before filming an existing script, it is usually necessary to evaluate the script to predict the ratings of the script after it is adapted into a film or television program.

[0003] Normally, in the script evaluation work, an evaluator needs to read the script and obtain information such as the strength and rhythm of the plot in the script through the evaluator's subjective perception of the script, and then predict the audience's interest in the script or film and television work.

[0004] However, manual script evaluation does not have standardized evaluation indicators, may be affected by the evaluator's personal preferences, and lacks intuitive data presentation and standardized evaluation criteria. Summary of the invention

[0005] In order to solve the above technical problems, the present disclosure provides a plot text evaluation method, device, equipment and readable storage medium to improve the objectivity and intuitiveness of plot text evaluation.

[0006] In a first aspect, an embodiment of the present disclosure provides a method for evaluating a plot text, comprising:

[0007] dividing the plot text into a plurality of plot subtexts corresponding one-to-one to the plurality of plot plots;

[0008] Obtaining character description information corresponding to the plot text and plot summary information corresponding to each of the plot sub-texts;

[0009] Calculating a plot level score for each of the plot subtexts based on the character description information and the plot summary information;

[0010] Determining a plot distribution trend of the plot text according to the temporal positions of the multiple plot sub-texts in the plot text and the plot grade score of each of the plot sub-texts;

[0011] The similarity between the plot distribution trend and a preset plot distribution index is calculated to obtain an evaluation result of the plot text.

[0012] In some embodiments, the step of obtaining the character description information corresponding to the plot text and the plot summary information corresponding to each of the plot sub-texts includes:

[0013] Inputting the plot text and the role analysis guide into a large language model to obtain the role description information;

[0014] The plot subtext and the plot analysis guide are input into a large language model to obtain the plot summary information.

[0015] In some embodiments, the step of calculating the plot level score of each plot subtext according to the character description information and the plot summary information includes:

[0016] Determine a plurality of evaluation index values ​​of the plot subtext according to the character description information and the plot summary information, wherein the plurality of evaluation index values ​​correspond one-to-one to a plurality of preset evaluation indexes;

[0017] Determine the weight corresponding to each of the preset evaluation indicators according to the type of the plot text;

[0018] The plot level score of the plot subtext is calculated based on the multiple evaluation index values ​​and the weight corresponding to each of the preset evaluation indexes.

[0019] In some embodiments, the character description information includes the name of the main character, the main character's influence on the plot development is greater than a preset influence threshold, and the determining of the multiple evaluation index values ​​of the plot subtext according to the character description information and the plot summary information includes:

[0020] Extracting the characters appearing in the plot subtext according to the plot summary information;

[0021] If it is determined based on the character description information that the characters appearing in the plot subtext include main characters, then multiple evaluation index values ​​of the plot subtext are determined based on the character description information and the plot summary information; or,

[0022] If it is determined based on the character description information that the characters appearing in the plot sub-text do not include a main character, then the plot level score of the plot sub-text is determined to be a preset minimum score.

[0023] In some embodiments, the plurality of preset evaluation indicators include one or more of a character fate line change indicator, a character relationship line change indicator, a dramatic conflict indicator, a suspense setting indicator, and an emotional concentration indicator; the determining of the plurality of evaluation indicator values ​​of the plot subtext according to the character description information and the plot summary information includes:

[0024] The character fate line change index value, the character relationship line change index value, and the emotion concentration index value of the plot sub-text are determined according to the character description information and the plot summary information, and the dramatic conflict index value and suspense setting index value of the plot sub-text are determined according to the plot summary information.

[0025] In some embodiments, determining a plurality of evaluation index values ​​of the plot subtext according to the character description information and the plot summary information includes:

[0026] The character description information and the plot summary information are input into a large language model, and the large language model is made to output a plurality of evaluation index values ​​of the plot subtext through a plurality of preset prompt word instructions, wherein the preset prompt word instructions correspond one to one to the plurality of evaluation index values.

[0027] In some embodiments, after calculating the plot level score of each plot subtext according to the character description information and the plot summary information, the method further includes:

[0028] The plot level score of each of the plot sub-texts is verified.

[0029] In some embodiments, the checking of the plot level score of each plot sub-text includes:

[0030] Inputting the character description information and the plot summary information into the large language model again to obtain a first verification reference score of the plot subtext;

[0031] If the difference between the first verification reference score and the plot level score is greater than a first preset threshold, adjusting the parameters of the large language model so as to reduce the difference between the first verification reference score and the plot level score, thereby obtaining an updated large language model;

[0032] Inputting the character description information and the plot summary information into the updated large language model to obtain a second verification reference score for the plot subtext;

[0033] An average of the plot level score, the first verification reference score, and the second verification reference score is calculated to obtain a verified plot level score.

[0034] In some embodiments, the checking of the plot level score of each plot sub-text includes:

[0035] Get the plot level scores of multiple plot subtexts;

[0036] If the variance of the plot level scores of the plurality of plot sub-texts is greater than a second preset threshold, updating the temperature parameter of the large language model to obtain an updated large language model;

[0037] For any plot sub-text, input the character description information and the plot summary information into the updated large language model to obtain a third verification reference value;

[0038] An average value of the plot level score and the third verification reference value is used as the verified plot level score.

[0039] In a second aspect, an embodiment of the present disclosure provides a plot text evaluation device, comprising:

[0040] a division module for dividing the plot text into a plurality of plot subtexts corresponding one-to-one to a plurality of plot plots;

[0041] An acquisition module, used to acquire character description information corresponding to the plot text and plot summary information corresponding to each plot sub-text;

[0042] A first calculation module, configured to calculate a plot level score of each of the plot subtexts according to the character description information and the plot summary information;

[0043] A determination module, configured to determine a plot distribution trend of the plot text according to the temporal positions of the plurality of plot sub-texts in the plot text and the plot grade score of each of the plot sub-texts;

[0044] The second calculation module is used to calculate the similarity between the plot distribution trend and a preset plot distribution index to obtain an evaluation result of the plot text.

[0045] In a third aspect, an embodiment of the present disclosure provides an electronic device, including:

[0046] Memory;

[0047] Processor; and

[0048] Computer programs;

[0049] The computer program is stored in the memory and is configured to be executed by the processor to implement the method as described in the first aspect.

[0050] In a fourth aspect, an embodiment of the present disclosure provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor to implement the method described in the first aspect.

[0051] In a fifth aspect, the embodiments of the present disclosure further provide a computer program product, which includes a computer program or instructions, and when the computer program or instructions are executed by a processor, the method for evaluating a plot text as described above is implemented.

[0052] The plot text evaluation method, device, equipment and readable storage medium provided by the embodiments of the present disclosure quantify the evaluation result of each plot in the plot text into a plot grade score through character description information and plot summary information, further obtain the plot distribution trend that can reflect the overall trend of the plot grade score, and use the similarity between the plot distribution trend and the preset plot distribution index to measure the quality of the plot text, thereby improving the objectivity and intuitiveness of the plot text evaluation. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0054] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0055] Figure 1 A flow chart of a method for evaluating a plot text provided in an embodiment of the present disclosure;

[0056] Figure 2 A plot level score line graph provided for an embodiment of the present disclosure;

[0057] Figure 3 A flow chart of a plot text evaluation method provided by another embodiment of the present disclosure;

[0058] Figure 4 A diagram of an evaluation method for plot text provided in an embodiment of the present disclosure;

[0059] Figure 5 A schematic diagram of a comparison between actual broadcast data and plot level scores provided in an embodiment of the present disclosure;

[0060] Figure 6 A schematic diagram of the structure of a plot text evaluation device provided by an embodiment of the present disclosure;

[0061] Figure 7 A schematic diagram of the structure of an electronic device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0062] In order to more clearly understand the above-mentioned objectives, features and advantages of the present disclosure, the scheme of the present disclosure will be further described below. It should be noted that the embodiments of the present disclosure and the features in the embodiments can be combined with each other without conflict.

[0063] In the following description, many specific details are set forth to facilitate a full understanding of the present disclosure, but the present disclosure may also be implemented in other ways different from those described herein; it is obvious that the embodiments in the specification are only part of the embodiments of the present disclosure, rather than all of the embodiments.

[0064] The present disclosure provides a method for evaluating a plot text, which is described below in conjunction with a specific embodiment.

[0065] Figure 1 Flowchart of the evaluation method of the plot text provided by the embodiment of the present disclosure. The method can be applied to terminal devices with data processing functions, such as smart phones, PDAs, tablet computers, wearable devices with display screens, desktop computers, laptop computers, all-in-one machines, smart home devices, etc. The terminal device can call a large language model (LLM), which can be deployed in a server. It is understandable that the evaluation method of the plot text provided by the embodiment of the present disclosure can also be applied in other scenarios.

[0066] Below Figure 1 The evaluation method of the plot text shown is introduced, and the specific steps of the method are as follows:

[0067] S101, dividing a plot text into a plurality of plot sub-texts corresponding one-to-one to a plurality of plot plots.

[0068] The plot text is used to express the storyline. It is the textual basis for artistic creations such as drama and film, and is also an important basis for directors and actors to perform.

[0069] The plot text includes multiple scenes, wherein one or more scenes constitute a plot plot, and the plot text is divided into multiple plot sub-texts according to the plot plot. That is, a single plot sub-text corresponds to a plot plot, and a plot plot corresponds to one or more scenes.

[0070] Optionally, the division of plot texts and plot sub-texts can be done manually, or the plot texts can be submitted one by one or in batches to the large language model for judgment and division through a sliding window.

[0071] Among them, a single plot usually takes place in the same scene and is explained by the same set of tasks. For example, scene 1 is character A on the way to the playground, scene 2 is character A arriving at the playground, and scene 3 is character B getting up; in the above scenes, scene 1 and scene 2 constitute a plot, and scene 3 is a plot.

[0072] S102: Obtain character description information corresponding to the plot text and plot summary information corresponding to each plot sub-text.

[0073] The role description information corresponding to the plot text is the relevant information of each role in the plot text, such as the name of each role in the script, and the role attribute of each role, whether it is a main role, etc.

[0074] The plot summary information is the summary information of the plot subtext.

[0075] In some embodiments, character description information corresponding to the plot text and plot summary information corresponding to each of the plot sub-texts may be acquired through a large language model.

[0076] Specifically, the plot text and the role analysis guide are input into the large language model to obtain the role description information; the plot sub-text and the plot analysis guide are input into the large language model to obtain the plot summary information.

[0077] Among them, the role analysis guide words and the plot analysis guide words refer to instructions that can guide the large language model to perform related functions, such as "please extract the role description information of the plot text" or "please extract the plot summary information of the plot text."

[0078] Specifically, the large language model has the ability to understand the plot text or plot sub-text. The role analysis guide can guide the large language model to understand the plot text, and extract and analyze the relevant information of each character therein, and finally obtain the character description information; the plot analysis guide can guide the large language model to understand the plot text and extract the plot summary information corresponding to the plot sub-text.

[0079] In other embodiments, the character description information and plot summary information may also be maintained manually, which is not limited in the embodiments of the present disclosure.

[0080] S103: Calculate the plot level score of each plot subtext according to the character description information and the plot summary information.

[0081] The plot level score reflects whether the dramatic function of a plot sub-text is strong, and the intensity of the dramatic function. Generally speaking, the stronger the dramatic function of a plot sub-text with conflicts, plot reversals, and emotional outbursts, the higher the plot level score.

[0082] In this step, the plot level score of each plot sub-text is obtained respectively.

[0083] Optionally, for each plot sub-text, the plot level score can be comprehensively judged through multiple dimensions. For example, the intensity of the dramatic function of the dimension of changing the character's fate line can be determined according to the deviation or transformation of the character's ultimate goal in the plot sub-text; the intensity of the dramatic function of the dimension of changing the character's relationship line can be determined by the substantial change in the relationship between the characters; the intensity of the dramatic function of the dimension of dramatic conflict can be determined by the degree of success of the characters in completing the task goals; the intensity of the dramatic function of the dimension of suspense setting can be determined by the clarity of the expectations of known action settings in the plot sub-text; the intensity of the dramatic function of the dimension of emotional concentration can be determined by the degree of change in the emotional trends of the characters themselves and between the characters, etc.

[0084] One possible implementation method is to use multiple preset evaluation indicators to represent the intensity of the dramatic function of each dimension in each plot sub-text, obtain multiple evaluation indicator values, and further calculate the sum of the multiple evaluation indicator values ​​to obtain the plot level score of the plot sub-text; or, set a corresponding weight for each preset evaluation indicator, calculate the weighted average or weighted total of the multiple evaluation indicator values, and obtain the plot level score of the plot sub-text.

[0085] For example, the preset evaluation indicators include the character fate line change index, the character relationship line change index, the dramatic conflict index, the suspense setting index, and the emotional concentration index. If the values ​​of the above evaluation indicators corresponding to a certain plot sub-text are all 1 point, then the plot level score corresponding to the plot sub-text is 5 points.

[0086] S104: Determine a plot level distribution trend of the plot text according to the temporal positions of the multiple plot sub-texts in the plot text and the plot level score of each of the plot sub-texts.

[0087] After obtaining the plot level score of each plot sub-text, these plot level scores are visualized according to the order of the plot sub-texts in the plot text, so that users can intuitively understand the distribution trend of the plot level scores in the entire plot text.

[0088] Optionally, multiple plot sub-texts and their corresponding plot level scores are formed into a line graph for display processing, with the chronological position of the plot sub-text (or the chronological position of the scene) as the horizontal axis and the plot level score as the vertical axis.

[0089] Figure 2 The plot level score line chart provided by the embodiment of the present disclosure. Figure 2As shown in the figure, the horizontal axis represents the temporal position of each scene in the plot text from left to right, and the vertical axis represents the plot level score of each scene. A single plot sub-text includes one or more scenes, and the scenes belonging to the same plot sub-text have the same plot level score, which is the plot level score of the plot sub-text. It can be seen that the visualized plot level score can reflect the distribution position and proportion of the more intense and more bland plots in the entire plot text, and can also reflect the overall structure and rhythm of the plot.

[0090] Specifically, Figure 2 For example, the plot level scores of the 2nd to 8th scenes are the same, all 1, indicating that the 2nd to 8th scenes belong to the same plot subtext. Combined with one of the calculation rules in S103, the sum of the multiple evaluation index values ​​of the plot subtext is 1, which is reflected in the plot level score line graph as a line segment composed of multiple consecutive points with a vertical coordinate of 1.

[0091] S105, calculating the similarity between the plot distribution trend and a preset plot distribution index to obtain an evaluation result of the plot text.

[0092] The preset plot distribution index is used to evaluate whether the plot distribution trend of the plot text meets the expected expectations. Optionally, the preset plot distribution index is calculated based on the actual broadcast data of the same type of popular dramas that have been broadcast, reflecting the distribution of viewing popularity as the plot develops. Alternatively, the preset plot distribution index can also be set in combination with expert experience.

[0093] The higher the similarity between the plot distribution trend and the preset plot distribution index, the higher the quality of the plot text is, and the plot distribution, the overall rhythm of the plot, etc. are considered to meet the expected expectations.

[0094] The disclosed embodiment divides a plot text into a plurality of plot sub-texts corresponding to a plurality of plot plots one by one; obtains character description information corresponding to the plot text and plot summary information corresponding to each of the plot sub-texts; calculates a plot grade score of each of the plot sub-texts according to the character description information and the plot summary information; determines a plot distribution trend of the plot text according to the temporal positions of the plurality of plot sub-texts in the plot text and the plot grade score of each of the plot sub-texts; calculates the similarity between the plot distribution trend and a preset plot distribution index to obtain an evaluation result of the plot text, quantifies the evaluation result of each plot plot in the plot text into a plot grade score according to the character description information and the plot summary information, further obtains a plot distribution trend that can reflect the overall trend of the plot grade score, and uses the similarity between the plot distribution trend and the preset plot distribution index to measure the quality of the plot text, thereby improving the objectivity and intuitiveness of the plot text evaluation.

[0095] Figure 3 This is a flow chart of a method for evaluating a plot text provided by another embodiment of the present disclosure. Figure 3 As shown, the method includes the following steps.

[0096] S301, dividing the plot text into a plurality of plot sub-texts corresponding to a plurality of plot plots one by one.

[0097] S302: Obtain character description information corresponding to the plot text and plot summary information corresponding to each plot sub-text.

[0098] Specifically, the implementation process and principle of S301-S302 are consistent with those of S101-S102, and will not be repeated here.

[0099] For each plot sub-text, S303 to S309 are executed in a loop until the plot level score corresponding to each plot sub-text is obtained.

[0100] S303: Extract the characters appearing in the plot subtext according to the plot summary information.

[0101] The plot summary information includes the characters in the plot subtext and the corresponding plot summary, and the characters appearing in the plot subtext can be extracted according to the plot summary information. Specifically, the characters appearing in the plot text or the plot subtext are divided into two types: main characters and non-main characters.

[0102] S304: Based on the character description information, determine whether the main character appears in the plot subtext. If yes, execute S305; if no, execute S308.

[0103] For a plot subtext, the premise of plot grade score evaluation is that the dramatic function of the plot subtext acts on the main character. Therefore, before calculating the plot grade score, it is necessary to determine whether the main character appears in the plot subtext.

[0104] The character description information includes the name of each character in the plot text and whether each character is a main character. When extracting the character description information, for each character, whether the character is a main character is determined according to the degree of influence of the character on the plot development; when the degree of influence of the character on the plot development is greater than a preset influence degree threshold, the character is determined to be a main character.

[0105] Among them, the degree of influence of the role on the development of the plot can be evaluated comprehensively from multiple dimensions, including the following dimensions:

[0106] Character activity: Count the number of times a character appears in the plot text and the duration of each appearance. The more times a character appears and the longer the average duration of each appearance, the higher the exposure and activity of the character, and the greater the impact on the development of the plot.

[0107] Plot involvement: Assess the character's involvement in the plot, such as decision-making, actions, conflicts, etc. The more plots a character participates in, the greater the degree of influence the character has on the development of the plot.

[0108] Character relationship network: Extract the relationship network between characters based on the plot text, and analyze the position and influence of the characters in the relationship network. The closer the position of the character in the relationship network is to the central hub, and the more characters are connected, the greater the influence of the character on the development of the plot.

[0109] For each character, based on the above-mentioned dimensional data and the weight of each dimension, the degree of influence of the character on the development of the plot is calculated by weighted sum or weighted average.

[0110] Optionally, set scoring rules for each dimension, set the score of each dimension to a certain range (such as 0-100), and calculate the score of each dimension according to the performance of the character in the plot text. For example, when a character's appearances reach M times, its character activity score is determined to be 90; when the number of plots involved reaches N, its plot participation score is determined to be 90 points; when the number of connected characters in the character relationship network reaches W, its character relationship network score is determined to be 90 points. Taking the weight of each dimension as the same and the weighted sum as an example, the score of the degree of influence of the character on the development of the plot is 270, where M, N, and W are all positive integers.

[0111] The preset influence threshold is a threshold value set by the user according to actual needs, or a threshold value that can screen out a certain number of main characters. Specifically, the preset proportion (such as 90%) of the maximum score value of the influence degree of the character on the development of the plot can be set as the preset influence degree threshold.

[0112] Optionally, the main role names may be added or deleted during the process of manually maintaining the role description information.

[0113] By comparing the character names extracted from the plot sub-text with the character names summarized in the character description information, it is possible to identify whether the characters appearing in the plot sub-text are main characters.

[0114] S305 . Determine a plurality of evaluation index values ​​of the plot subtext according to the character description information and the plot summary information, wherein the plurality of evaluation index values ​​correspond one-to-one to a plurality of preset evaluation indexes.

[0115] If it is determined based on the character description information that the characters appearing in the plot sub-text include main characters, a plurality of evaluation index values ​​of the plot sub-text are determined according to the character description information and the plot summary information.

[0116] Specifically, the multiple preset evaluation indicators include one or more of a character fate line change index, a character relationship line change index, a dramatic conflict index, a suspense setting index, and an emotion concentration index; the multiple evaluation indicator values ​​of the plot sub-text determined according to the character description information and the plot summary information include: determining the character fate line change index value, the character relationship line change index value, and the emotion concentration index value of the plot sub-text according to the character description information and the plot summary information, and determining the dramatic conflict index value and the suspense setting index value of the plot sub-text according to the plot summary information.

[0117] It is understandable that the above-mentioned preset evaluation indicators can be combined according to actual needs to evaluate the plot text.

[0118] The following is an introduction to each preset evaluation indicator and the method for determining the indicator value corresponding to each preset evaluation indicator.

[0119] The character fate line change index is used to evaluate the degree of change in the character's ultimate goal. The character description also includes each character's character fate line, which describes the series of events, choices, conflicts experienced by the character and how these experiences together shape the trajectory of his or her personality development, emotional changes, and final outcome.

[0120] The method for determining the change index value of the character's fate line is to determine whether there is a character's fate change node in the plot subtext according to the character's fate line and the plot summary information, and the character's fate change node is a node where the character's ultimate goal changes; if there is no character's fate change node in the plot subtext, then determine the character's fate change index value corresponding to the character's fate change index as a preset minimum index value; or, if there is a character's fate change node in the plot subtext, then determine the character's fate change index value according to the degree of change of the character's ultimate goal in the character's fate change node, wherein the character's ultimate goal refers to the task goal that the character is ultimately to complete, and the character's fate change index value is proportional to the degree of change of the ultimate goal. That is, the greater the change in the ultimate goal, the higher the character's fate change index value.

[0121] The role relationship line change index is used to evaluate the degree of change in the relationship between roles. The role description information includes the final relationship between each role, which is formed during the continuous development of the plot.

[0122] The method for determining the role relationship line change index value is to determine whether there is a role relationship change node in the plot sub-text based on the final relationship between the characters and the plot summary information; if there is no role relationship change node in the plot sub-text, that is, the relationship between the characters in the plot sub-text has not changed, then the task relationship line change index value is determined to be a preset minimum index value; or, if there is a role relationship change node in the plot sub-text, then the role relationship line change index value is determined based on the degree of influence of the role relationship change in the role relationship change node on the final relationship between the characters, wherein the role relationship line change index value is proportional to the degree of influence.

[0123] Optionally, before determining whether there is a role relationship change node in the plot subtext based on the final relationship between the roles and the plot summary information, it is necessary to determine whether the roles in the relationship combination corresponding to the role relationship are all main roles. If there are non-main roles, the role relationship line change index is directly determined to be the preset minimum index value; if the roles in the relationship combination are all main roles, it is further determined whether there is a role relationship change node in the plot subtext.

[0124] The dramatic conflict index is used to evaluate the contradictions and tensions between different characters or between characters and the environment in the plot subtext due to various differences or contradictions.

[0125] The method for determining the dramatic conflict index value is to determine whether the character has completed the task goal in the plot according to the plot summary information. If the character has not completed the task goal, the dramatic conflict index value is determined to be the preset maximum index value; if the character has completed the task goal, the dramatic conflict index value is determined according to the degree of tortuousness of the process of the character completing the task goal, and the dramatic conflict index value is proportional to the tortuousness. That is, the more tortuous the process of the character completing the task goal, the higher the dramatic conflict index value.

[0126] The suspense setting index is used to evaluate the effect of suspense setting in the plot subtext. Specifically, setting suspense usually requires setting expected results that are easy for the audience to guess but not completely certain for the known actions that have been explained.

[0127] The method for determining the suspense setting index value is to determine whether the plot subtext sets an expected result for a known action based on the plot summary information. If the expected result is not set for the known action, the suspense setting index is determined to be a preset minimum index value; if the plot subtext sets an expected result for the known action, the suspense setting index value is determined based on the uncertainty of the expected result, wherein the suspense setting index value is proportional to the uncertainty.

[0128] The emotional concentration index is used to evaluate the emotional trend of the character itself and between characters. The character description information includes the character's global emotional information, which is used to reflect the emotional changes between characters and emotional outburst points.

[0129] The method for determining the emotion concentration index value is to obtain the local emotion information of the character in the plot sub-text according to the plot summary information analysis, and determine the emotion concentration index value based on the contribution degree of the local emotion information to the global emotion information, wherein the emotion concentration index value is proportional to the contribution degree.

[0130] In summary, the character fate line change index value, character relationship line change index value, dramatic conflict index value, suspense setting index value and emotional concentration index value of the plot sub-text are obtained respectively.

[0131] Optionally, the definition of each indicator value may be to determine the indicator value within the range of a preset minimum indicator value and a preset maximum indicator value; or, to determine different indicator levels within the range of a preset minimum indicator value and a preset maximum indicator value, wherein each indicator level corresponds to a different indicator value, and the indicator level is first determined according to the above rules, and then the indicator value corresponding to the indicator level is further determined.

[0132] For example, for any indicator, the preset minimum indicator value is 0, the preset maximum indicator value is 2, and the indicator value can be any value in the interval [0,2]; or, the interval [0,2] is divided into three indicator levels, the indicator value corresponding to the first indicator level is 2, the indicator value corresponding to the second indicator level is 1, and the indicator value corresponding to the third indicator level is 0.

[0133] S306: Determine the weight corresponding to each of the preset evaluation indicators according to the type of the plot text.

[0134] The type of plot text determines the importance of the evaluation criteria of different dimensions, so different evaluation index weights are set for different plot text types. For example, for suspense dramas, the importance of suspense setting indicators is relatively high, so the weight of suspense setting indicators for suspense-type plot texts is higher.

[0135] S307: Calculate the plot level score of the plot subtext based on the multiple evaluation indicator values ​​and the weight corresponding to each of the preset evaluation indicators.

[0136] The product of the index value and the weight of each preset evaluation index is calculated respectively to obtain the score corresponding to each preset evaluation index, and the score corresponding to each preset evaluation index is further summed to obtain the plot level score of the plot sub-text.

[0137] S308: Determine that the plot level score of the plot subtext is a preset minimum score.

[0138] When the main character does not appear in the plot sub-text, it means that the dramatic function of the plot sub-text does not act on the main character. At this time, the plot level score of the plot sub-text is determined to be the preset minimum score.

[0139] S309: Determine whether to obtain the plot level score corresponding to each plot sub-text. If yes, execute S310; if no, execute S303.

[0140] S310: Determine a plot distribution trend of the plot text according to the temporal positions of the multiple plot sub-texts in the plot text and the plot grade score of each of the plot sub-texts.

[0141] S311, calculating the similarity between the plot distribution trend and a preset plot distribution index to obtain an evaluation result of the plot text.

[0142] The disclosed embodiment sets a variety of preset evaluation indicators and sets different indicator weights according to different script types, so that the evaluation of the plot text is adapted to the type of the plot text and its rationality is improved.

[0143] Based on the above embodiment, determining multiple evaluation index values ​​of the plot sub-text based on the character description information and the plot summary information includes: inputting the character description information and the plot summary information into a large language model, and using multiple preset prompt word instructions to make the large language model output multiple evaluation index values ​​of the plot sub-text, and the preset prompt word instructions correspond one-to-one to the multiple evaluation index values.

[0144] Figure 4 This is a diagram of an evaluation method for plot text provided by an embodiment of the present disclosure. Figure 4 As shown, the terminal device includes an algorithm layer and a model layer. The user inputs the character description information and the plot summary information into the terminal device, and the algorithm layer determines whether the main character appears in the plot sub-text. If so, the model layer (large language model) is called to input the character description information and the plot summary information into the large language model. The large language model analyzes the character description information and the plot summary information in combination with preset prompt word instructions to obtain multiple indicator values.

[0145] Among them, the preset prompt word instructions are used to guide the large model to perform analysis corresponding to each preset evaluation indicator, such as "Please give the emotional concentration index value based on the character description information and plot summary information."

[0146] Furthermore, after calculating the plot level score of each of the plot sub-texts according to the character description information and the plot summary information, the method further includes: verifying the plot level score of each of the plot sub-texts.

[0147] Since the large language model is unstable when generating results, it is necessary to further verify the plot level score after obtaining it, and when the verification result shows that the large language model needs to be updated, re-analysis is performed based on the updated large language model.

[0148] Among them, a possible implementation method is to input the character description information and the plot summary information into the large language model again to obtain a first verification reference score for the plot sub-text; if the difference between the first verification reference score and the plot level score is greater than a first preset threshold, adjust the parameters of the large language model to reduce the difference between the first verification reference score and the plot level score to obtain an updated large language model; input the character description information and the plot summary information into the updated large language model to obtain a second verification reference score for the plot sub-text; calculate the average of the plot level score, the first verification reference score, and the second verification reference score to obtain the verified plot level score.

[0149] Among them, due to the randomness of the large language model when generating results, the first verification reference score may be different from the previously obtained plot level score. If the difference between the two is within a reasonable range (that is, less than or equal to the first preset threshold), the previously obtained plot level score is considered to be credible and can be directly used in the visualization processing of the evaluation results; or the average of the previously obtained plot level score and the first verification reference value is used as the verified plot level score for the visualization processing of the evaluation results.

[0150] If the difference between the first verification reference score and the plot level score is greater than the first preset threshold, the previously obtained plot level score is considered unreliable and needs to be processed. The processing method is to adjust the parameters of the large language model to reduce the difference between the first verification reference score and the plot level score, and obtain an updated large language model; further input the character description information and the plot summary information into the updated large language model again to obtain a second verification reference score, and use the average of the plot level score, the first verification reference score and the second verification reference score as the verified plot level score, which is used in the visualization of the evaluation results.

[0151] Another possible implementation method is to obtain plot level scores of multiple plot sub-texts; if the variance of the plot level scores of the multiple plot sub-texts is greater than a second preset threshold, the temperature parameter of the large language model is updated to obtain an updated large language model; for any one plot sub-text, the character description information and the plot summary information are input into the updated large language model to obtain a third verification reference value.

[0152] The obtained plot level scores of the multiple plot sub-texts may be the plot level scores of multiple consecutive plot sub-texts, for example, the plot level scores corresponding to the first plot sub-text to the first plot sub-text currently completed evaluation.

[0153] If the variance of these plot level scores is greater than the second preset threshold, it means that the output result of the large language model is too random, and the plot level scores output by it need to be verified. Further, by adjusting the temperature parameter of the large language model, when the plot level scores are generally high, such as when the mean is greater than the third preset threshold, the temperature parameter is lowered; conversely, when the plot level scores are generally low, such as when the mean is less than the fourth preset threshold, the temperature parameter is increased.

[0154] Further, the character description information and the plot summary information are input into the updated large language model to obtain a third verification reference value, and the average of the plot level score and the third verification reference value is used as the verified plot level score.

[0155] The disclosed embodiment further improves the accuracy of the plot level score by verifying the plot level score, thereby greatly improving the accuracy of the plot text evaluation result.

[0156] Based on the above embodiment, the method further includes: predicting the broadcast data of the film and television works corresponding to the plot text according to the multiple plot level scores corresponding to the multiple plot sub-texts.

[0157] Specifically, the broadcast data represents the average effective viewing time per person per minute.

[0158] The effective viewing time per minute is the ratio of the total viewing time to the number of viewers. The number of viewers in that minute is counted in units of minutes. Users who play at normal speed are counted as 1 viewer, and users who play at N times the speed are counted as Number of viewers. Users who skip this minute are counted as 0 viewers.

[0159] Furthermore, when the plot text is truly adapted into a film or television work, and there is no rewriting or enhancement of the picture rendering during the film or television adaptation process, the actual broadcast data of the film or television work is counted and compared with the multiple plot level scores corresponding to the multiple plot sub-texts, and the large language model is further updated based on the comparison results to make the plot level score curve closer to the actual broadcast data curve.

[0160] Figure 5 A schematic diagram of the comparison between actual broadcast data and plot level scores provided in an embodiment of the present disclosure, wherein the horizontal axis represents the temporal position after the plot or scene is aligned with the number of minutes of the film and television work, the vertical axis on the left represents the plot level score, and the vertical axis on the right represents the average effective viewing time per person, the upper curve is the actual broadcast data curve, and the lower curve is the plot level score curve.

[0161] According to the trend of the plot level score curve and the actual broadcast data curve, it can be seen that at locations with higher plot level scores, the average effective viewing time per person is longer; at locations with lower plot level scores, the average effective viewing time per person is shorter.

[0162] The disclosed embodiment predicts the broadcast effect of the film and television works corresponding to the plot text based on the plot level score, so as to help the drama producers understand the evaluation results of the plot text in a visual form and provide scientific reference opinions; at the same time, the large language model is further updated according to the comparison results between the plot level score and the actual broadcast data, so as to continuously improve the scientificity and accuracy of the plot text evaluation.

[0163] Figure 6 The schematic diagram of the structure of the plot text evaluation device provided in the embodiment of the present disclosure. The plot text evaluation device may be the terminal device described in the above embodiment, or the plot text evaluation device may be a component or assembly in the terminal device. The plot text evaluation device provided in the embodiment of the present disclosure may execute the processing flow provided in the plot text evaluation method embodiment, such as Figure 6As shown, the evaluation device 60 for plot text includes: a division module 61, an acquisition module 62, a first calculation module 63, a determination module 64, and a second calculation module 65; wherein the division module 61 is used to divide the plot text into a plurality of plot sub-texts corresponding to a plurality of plot plots one by one; the acquisition module 62 is used to obtain the character description information corresponding to the plot text and the plot summary information corresponding to each of the plot sub-texts; the first calculation module 63 is used to calculate the plot level score of each of the plot sub-texts according to the character description information and the plot summary information; the determination module 64 is used to determine the plot distribution trend of the plot text according to the time sequence positions of the plurality of plot sub-texts in the plot text and the plot level score of each of the plot sub-texts; the second calculation module 65 is used to calculate the similarity between the plot distribution trend and a preset plot distribution index to obtain the evaluation result of the plot text.

[0164] Optionally, the acquisition module 62 includes a first acquisition unit 621 and a second acquisition unit 622. The first acquisition unit 621 is used to input the plot text and the character analysis guide into the large language model to obtain the character description information; the second acquisition unit 622 is used to input the plot sub-text and the plot analysis guide into the large language model to obtain the plot summary information.

[0165] Optionally, the first calculation module 63 includes a first determination unit 631, a second determination unit 632, and a calculation unit 633; ​​the first determination unit 631 is used to determine multiple evaluation index values ​​of the plot sub-text according to the character description information and the plot summary information, and the multiple evaluation index values ​​correspond one-to-one to multiple preset evaluation indicators; the second determination unit 632 is used to determine the weight corresponding to each of the preset evaluation indicators according to the type of the plot text; the calculation unit 633 is used to calculate the plot level score of the plot sub-text based on the multiple evaluation index values ​​and the weight corresponding to each of the preset evaluation indicators.

[0166] Optionally, the character description information includes the name of the main character, and the degree of influence of the main character on the development of the plot is greater than a preset influence threshold, and the first determination unit 631 is specifically used to extract the characters appearing in the plot sub-text according to the plot summary information; if it is determined based on the character description information that the characters appearing in the plot sub-text include the main character, then multiple evaluation index values ​​of the plot sub-text are determined based on the character description information and the plot summary information; or, if it is determined based on the character description information that the characters appearing in the plot sub-text do not include the main character, then the plot level score of the plot sub-text is determined to be a preset minimum score.

[0167] Optionally, the multiple preset evaluation indicators include one or more of a character fate line change index, a character relationship line change index, a dramatic conflict index, a suspense setting index, and an emotion concentration index. The first determination unit 631 is specifically used to determine the character fate line change index value, the character relationship line change index value, and the emotion concentration index value of the plot sub-text according to the character description information and the plot summary information, and determine the dramatic conflict index value and suspense setting index value of the plot sub-text according to the plot summary information.

[0168] Optionally, the first determination unit 631 is specifically used to input the character description information and the plot summary information into a large language model, and enable the large language model to output multiple evaluation index values ​​of the plot sub-text through multiple preset prompt word instructions, and the preset prompt word instructions correspond one-to-one to the multiple evaluation index values.

[0169] Optionally, the first calculation module 63 further includes a verification unit 634, which is used to verify the plot level score of each of the plot sub-texts.

[0170] Optionally, the verification unit 634 is used to input the character description information and the plot summary information into the large language model again to obtain a first verification reference score for the plot sub-text; if the difference between the first verification reference score and the plot level score is greater than a first preset threshold, adjust the parameters of the large language model to reduce the difference between the first verification reference score and the plot level score to obtain an updated large language model; input the character description information and the plot summary information into the updated large language model to obtain a second verification reference score for the plot sub-text; calculate the average of the plot level score, the first verification reference score, and the second verification reference score to obtain the verified plot level score.

[0171] Optionally, the verification unit 634 is used to obtain plot level scores of multiple plot sub-texts; if the variance of the plot level scores of the multiple plot sub-texts is greater than a second preset threshold, the temperature parameter of the large language model is updated to obtain an updated large language model; for any plot sub-text, the character description information and the plot summary information are input into the updated large language model to obtain a third verification reference value; and the average of the plot level score and the third verification reference value is used as the verified plot level score.

[0172] Figure 6 The plot text evaluation device of the illustrated embodiment can be used to implement the technical solution of the above method embodiment, and its implementation principle and technical effect are similar, which will not be repeated here.

[0173] Figure 7Schematic diagram of the structure of an electronic device provided in an embodiment of the present disclosure. The electronic device may be a terminal device as described in the above embodiment. The electronic device provided in an embodiment of the present disclosure may execute the processing flow provided in the embodiment of the method for evaluating a plot text, such as Figure 7 As shown, the electronic device 70 includes: a memory 71, a processor 72, a computer program and a communication interface 73; wherein the computer program is stored in the memory 71 and is configured so that the processor 72 executes the above-mentioned plot text evaluation method.

[0174] In addition, an embodiment of the present disclosure further provides a computer-readable storage medium on which a computer program is stored. The computer program is executed by a processor to implement the method for evaluating a plot text described in the above embodiment.

[0175] In addition, an embodiment of the present disclosure further provides a computer program product, which includes a computer program or instructions. When the computer program or instructions are executed by a processor, the method for evaluating a plot text as described above is implemented.

[0176] It should be noted that, in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0177] The above description is only a specific embodiment of the present disclosure, so that those skilled in the art can understand or implement the present disclosure. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present disclosure. Therefore, the present disclosure will not be limited to the embodiments described herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for evaluating a plot text, characterized in that: The method comprises: dividing the plot text into a plurality of plot subtexts corresponding one-to-one to the plurality of plot plots; Obtaining character description information corresponding to the plot text and plot summary information corresponding to each of the plot sub-texts; Calculating a plot level score for each of the plot subtexts based on the character description information and the plot summary information; Determining a plot distribution trend of the plot text according to the temporal positions of the multiple plot sub-texts in the plot text and the plot grade score of each of the plot sub-texts; The similarity between the plot distribution trend and a preset plot distribution index is calculated to obtain an evaluation result of the plot text.

2. The method according to claim 1, characterized in that The step of obtaining the character description information corresponding to the plot text and the plot summary information corresponding to each plot sub-text includes: Inputting the plot text and the role analysis guide into a large language model to obtain the role description information; The plot subtext and the plot analysis guide are input into a large language model to obtain the plot summary information.

3. The method according to claim 1, characterized in that The step of calculating the plot level score of each plot subtext according to the character description information and the plot summary information includes: Determine a plurality of evaluation index values ​​of the plot subtext according to the character description information and the plot summary information, wherein the plurality of evaluation index values ​​correspond one-to-one to a plurality of preset evaluation indexes; Determine the weight corresponding to each of the preset evaluation indicators according to the type of the plot text; The plot level score of the plot subtext is calculated based on the multiple evaluation index values ​​and the weight corresponding to each of the preset evaluation indexes.

4. The method according to claim 3, characterized in that The character description information includes the name of the main character, and the degree of influence of the main character on the development of the plot is greater than a preset influence degree threshold. The determining of the multiple evaluation index values ​​of the plot subtext according to the character description information and the plot summary information includes: Extracting the characters appearing in the plot subtext according to the plot summary information; If it is determined based on the character description information that the characters appearing in the plot subtext include main characters, then a plurality of evaluation index values ​​of the plot subtext are determined based on the character description information and the plot summary information; or, If it is determined based on the character description information that the characters appearing in the plot sub-text do not include a main character, then the plot level score of the plot sub-text is determined to be a preset minimum score.

5. The method according to claim 3, characterized in that: The plurality of preset evaluation indicators include one or more of a character fate line change indicator, a character relationship line change indicator, a dramatic conflict indicator, a suspense setting indicator, and an emotion concentration indicator; the plurality of evaluation indicator values ​​of the plot subtext are determined according to the character description information and the plot summary information, including: The character fate line change index value, the character relationship line change index value, and the emotion concentration index value of the plot sub-text are determined according to the character description information and the plot summary information, and the dramatic conflict index value and suspense setting index value of the plot sub-text are determined according to the plot summary information.

6. The method according to claim 3, characterized in that: The determining of a plurality of evaluation index values ​​of the plot subtext according to the character description information and the plot summary information comprises: The character description information and the plot summary information are input into a large language model, and the large language model is made to output a plurality of evaluation index values ​​of the plot subtext through a plurality of preset prompt word instructions, wherein the preset prompt word instructions correspond one to one to the plurality of evaluation index values.

7. The method according to claim 6, characterized in that After calculating the plot level score of each plot subtext according to the character description information and the plot summary information, the method further includes: The plot level score of each of the plot sub-texts is verified.

8. The method according to claim 7, characterized in that The checking of the plot level score of each plot sub-text includes: Inputting the character description information and the plot summary information into the large language model again to obtain a first verification reference score of the plot subtext; If the difference between the first verification reference score and the plot level score is greater than a first preset threshold, adjusting the parameters of the large language model so as to reduce the difference between the first verification reference score and the plot level score, thereby obtaining an updated large language model; Inputting the character description information and the plot summary information into the updated large language model to obtain a second verification reference score for the plot subtext; An average of the plot level score, the first verification reference score, and the second verification reference score is calculated to obtain a verified plot level score.

9. The method according to claim 7, characterized in that: The checking of the plot level score of each plot sub-text includes: Get the plot level scores of multiple plot subtexts; If the variance of the plot level scores of the plurality of plot sub-texts is greater than a second preset threshold, updating the temperature parameter of the large language model to obtain an updated large language model; For any plot sub-text, input the character description information and the plot summary information into the updated large language model to obtain a third verification reference value; An average value of the plot level score and the third verification reference value is used as the verified plot level score.

10. A plot text evaluation device, characterized in that: include: a division module for dividing the plot text into a plurality of plot subtexts corresponding one-to-one to a plurality of plot plots; An acquisition module, used to acquire character description information corresponding to the plot text and plot summary information corresponding to each plot sub-text; A first calculation module, configured to calculate a plot level score of each of the plot subtexts according to the character description information and the plot summary information; A determination module, configured to determine a plot distribution trend of the plot text according to the temporal positions of the plurality of plot sub-texts in the plot text and the plot grade score of each of the plot sub-texts; The second calculation module is used to calculate the similarity between the plot distribution trend and a preset plot distribution index to obtain an evaluation result of the plot text.

11. An electronic device, characterized in that: include: Memory; processor; as well as Computer programs; The computer program is stored in the memory and is configured to be executed by the processor to implement the method according to any one of claims 1 to 9.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 9 is implemented.

Images