Role fate track display method and device, equipment and medium

Abstract

The invention relates to a role fate track display method and device, equipment and a medium, and the method comprises the steps: obtaining a plot abstract text of a specified role based on a script text; according to the role intention information of the specified role and the at least one plot abstract text of the specified role, generating an event text of the specified role; inputting the role intention information of the specified role and the event text of the specified role into an influence value evaluation model for processing to obtain an influence value of each event text on the realization of the role intention of the specified role; and performing display processing on the influence value of each event text based on the time sequence information of each event text. According to the technical scheme, the role fate track can be accurately generated and displayed, and the development trend of the role is visually displayed.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology, and in particular to a method, device, equipment and medium for displaying a character's fate trajectory. Background Art

[0002] The script is the textual basis of artistic creation, and character creation is an important part of the script evaluation process.

[0003] Currently, the development trend of character growth is evaluated by manually reading the script, which cannot intuitively show the development trend of the character. Summary of the invention

[0004] In order to solve the above technical problems, the present disclosure provides a method, device, equipment and medium for displaying a character's destiny trajectory.

[0005] In a first aspect, an embodiment of the present disclosure provides a method for displaying a character's fate trajectory, comprising:

[0006] Inputting the script text into the plot generation model for processing to generate a plurality of plot summary texts corresponding to a plurality of plot scenes; the plurality of plot summary texts include at least one plot summary text of a designated character;

[0007] Acquiring role intention information of the designated role, and inputting the role intention information of the designated role and at least one plot summary text of the designated role into an event generation model for processing to generate an event text of the designated role;

[0008] Inputting the role intention information of the designated role and the event text of the designated role into the influence value evaluation model for processing, and obtaining the influence value of each event text on the designated role realizing the role intention;

[0009] Based on the time sequence information of each event text, the influence value of each event text on the designated character's realization of the character intention is displayed.

[0010] In a second aspect, the present disclosure provides a model training method, including:

[0011] Acquire training data; each piece of the training data includes sample character intention information, at least one sample plot summary text, and corresponding annotated sample event text, and an impact sample value of the corresponding annotated sample event text;

[0012] Inputting the sample character intention information and the at least one sample plot summary text into a pre-built initial event generation model, and generating a predicted event text through the initial event generation model;

[0013] Calculating a first current loss according to the sample event text and the predicted event text by using a preset loss function;

[0014] updating the model parameters of the initial event generation model according to the first current loss until the initial event generation model converges, thereby obtaining a trained event generation model;

[0015] Inputting the sample role intention information and the sample event text into a pre-built initial impact value evaluation model for processing to obtain an impact prediction value corresponding to the sample event text;

[0016] Calculating a second current loss according to the impact sample value and the impact prediction value by using a preset loss function;

[0017] The model parameters of the initial influence value assessment model are updated according to the second current loss until the initial influence value assessment model converges, thereby obtaining a trained influence value assessment model.

[0018] In a third aspect, an embodiment of the present disclosure provides a device for displaying a character's fate trajectory, including:

[0019] A first generation module is used to input the script text into the plot generation model for processing, and generate a plurality of plot summary texts corresponding to a plurality of plot scenes one by one; the plurality of plot summary texts include at least one plot summary text of a specified character;

[0020] A second generation module is used to obtain the role intention information of the specified role, and input the role intention information of the specified role and at least one plot summary text of the specified role into an event generation model for processing to generate an event text of the specified role;

[0021] A third generation module is used to input the role intention information of the designated role and the event text of the designated role into the impact value evaluation model for processing, so as to obtain the impact value of each event text on the designated role to realize the role intention;

[0022] The display module is used to display the influence value of each event text on the realization of the role intention of the designated role based on the time sequence information of each event text.

[0023] In a fourth aspect, an embodiment of the present disclosure provides an electronic device, comprising: a processor; a memory for storing executable instructions of the processor; the processor is used to read the executable instructions from the memory and execute the instructions to implement the method for displaying the character's destiny trajectory described in the first aspect above, or to implement the training method described in the second aspect above.

[0024] In a fifth aspect, an embodiment of the present disclosure provides a computer-readable storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by a processor, the method for displaying the character's destiny trajectory as described in the first aspect above is implemented, or the training method as described in the second aspect above is implemented.

[0025] Compared with the prior art, the technical solution provided by the disclosed embodiment has the following advantages: by processing the script text, the plot summary text of the character is obtained, and the event text of the character is generated according to the character's role intention information and the plot summary text. Then, the role intention information and the event text of the character are input into the influence value evaluation model for processing to obtain the influence value of each event text on the character's realization of the role intention, and based on the timing information of each event text, the influence value of each event text is displayed, thereby accurately determining the influence value of each event on the character's realization of the role intention, and constructing and displaying the character's destiny trajectory through the influence values ​​of each event, and accurately generating and displaying the character's destiny trajectory, and intuitively visualizing the character's development trend, so as to assist users in more intuitively understanding the character's development trend, and provide support for analyzing the characters of historical hit dramas. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] 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.

[0027] 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.

[0028] Figure 1 A flowchart of a method for displaying a character's fate trajectory provided by an embodiment of the present disclosure;

[0029] Figure 2 A schematic diagram of a process for determining an impact value of an event text provided by an embodiment of the present disclosure;

[0030] Figure 3 A schematic diagram of the shape of a positive trajectory;

[0031] Figure 4 This is a schematic diagram of the shape of another positive trajectory;

[0032] Figure 5 It is a schematic diagram of the shape of a flat trajectory;

[0033] Figure 6 This is a schematic diagram of the shape of a negative trajectory;

[0034] Figure 7 This is a schematic diagram of the shape of another negative trajectory;

[0035] Figure 8 A schematic diagram of another process for determining the impact value of an event text provided by an embodiment of the present disclosure;

[0036] Fig. 9 A flowchart of another method for displaying a character's fate trajectory provided by an embodiment of the present disclosure;

[0037] Fig.10 A schematic diagram showing the trajectory of a character's fate from an objective perspective is shown;

[0038] Fig.11 A schematic diagram showing the trajectory of a character's fate from a subjective perspective is shown;

[0039] Fig.12 A schematic diagram of the structure of a device for displaying a character's destiny trajectory provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0040] 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.

[0041] 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.

[0042] Figure 1 This is a flow chart of a method for displaying a character's fate trajectory provided in an embodiment of the present disclosure. The method provided in an embodiment of the present disclosure can be executed by a display device for a character's fate trajectory, which can be implemented using software and / or hardware and can be integrated on any electronic device with computing capabilities.

[0043] like Figure 1 As shown, the method for displaying the character's destiny trajectory provided by the embodiment of the present disclosure may include:

[0044] Step 101, input the script text into the plot generation model for processing, and generate a plurality of plot summary texts corresponding to a plurality of plot scenes.

[0045] The method of the disclosed embodiment is used to analyze and process the script text, generate the fate trajectory of each character in the script text, and display the fate trajectory of the character to intuitively visualize the development trend of the character.

[0046] In this embodiment, the plot generation model is used to analyze the plot summary of each plot scene in the script text. The input of the plot generation model is the script text, and the output is the plot summary text. The plot generation model can be implemented based on a large language model. The model structure and training method of the large language model can be implemented based on existing methods. For example, the training process includes: obtaining training data, each piece of training data includes a sample script text, and a corresponding annotated sample plot summary text, and then training the plot generation model based on the training data.

[0047] In actual applications, the script text includes multiple scenes, each scene corresponds to a plot scene. As an example, the script text is divided into script sub-texts corresponding to each plot scene. For each plot scene, the script sub-text corresponding to the plot scene is input into the plot generation model for processing to generate a plot summary text corresponding to the plot scene.

[0048] The multiple plot summary texts include at least one plot summary text of a specified role. Usually, the script text contains multiple roles, and the specified role is a role whose fate trajectory is currently to be determined among the multiple roles. Optionally, at least one plot summary text of the specified role is obtained from the multiple plot summary texts in the following manner: obtaining the role name of the specified role, matching the role name of the specified role in each plot summary text, and using the plot summary text that matches the role name as the plot summary text of the specified role.

[0049] Step 102, obtaining role intention information of a designated role, and inputting the role intention information of the designated role and at least one plot summary text of the designated role into an event generation model for processing to generate an event text of the designated role.

[0050] In this embodiment, the role intention information is used to represent the role's goals and / or the role's needs. The role intention information includes the role goal text and the role need text. For example, the role goal text is "maintaining and saving precarious family relationships" and the role need text is "recognition and respect, sense of security, sense of control, guilt and redemption."

[0051] The following describes how to obtain the role intention information.

[0052] As an example, the role intent information may be data obtained by artificial intelligence analysis, and obtaining the role intent information of a specified role includes: inputting the role biography text of the specified role into a large text model for processing to generate the role intent information of the specified role. In this example, for a role with a role biography text, the role biography text may be input into a pre-trained large text model for processing to generate the role intent information, and the model structure and training method of the large text model may be implemented based on existing methods.

[0053] As another example, the role intention information may be manually provided data, and obtaining the role intention information of a specified role includes: receiving text content input by a user as the role intention information of the specified role.

[0054] In this embodiment, the event generation model is used to perform semantic analysis and other processing on all plot summary texts of the specified role based on the role intention information, extract the events that affect the specified role to realize the role intention, and generate the event text of each event. The input of the event generation model is the role intention information and the plot summary text, and the output is the event text. The event generation model can be implemented based on the large language model. The model structure and training method of the large language model can be implemented based on the existing method. For example, the training process includes: obtaining training data, each training data includes sample role intention information, at least one sample plot summary text, and corresponding annotated sample event text, and then, training the event generation model based on the training data. Optionally, training the event generation model includes: inputting the sample role intention information and at least one sample plot summary text into a pre-constructed initial event generation model for processing to generate a predicted event text, and then, calculating the current loss using a preset loss function according to the sample event text and the predicted event text, and updating the model parameters of the initial event generation model according to the current loss until the initial event generation model converges to obtain a trained event generation model.

[0055] As an example, taking the role goal text as "maintaining and saving the precarious family relationship" and the role demand text as "recognition and respect, sense of security, sense of control, guilt and redemption", the plot summary text includes the following four texts: "When Zhang was teaching, the students ignored him", "Zhang declined the invitation to dinner and went home to cook for his wife", "Zhang talked to the neighbor in the elevator and was under pressure", "Zhang went home and found that the orchid he had taken good care of was stuck with his father-in-law's cigarette butt". The above role intention information and the four plot summary texts were input into the event generation model for processing, and the following two event texts were generated: "The dilemma of teachers' work" and "Increased family pressure".

[0056] Step 103 , inputting the role intention information of the designated role and the event text of the designated role into the influence value evaluation model for processing, and obtaining the influence value of each event text on the designated role realizing the role intention.

[0057] In this embodiment, the impact value evaluation model is used to analyze the event text and the role intention information through semantic analysis and other processing methods to quantify the impact of the evaluation event on the role's realization of the role intention. The input of the impact value evaluation model is the role intention information and the event text. The impact value evaluation model can be implemented based on the large language model. The model structure and training method of the large language model can be implemented based on the existing method. By inputting the role intention information and the event text sequence of the specified role into the impact value evaluation model, the impact value corresponding to each event text is obtained after processing. For example, the training process includes: obtaining training data, each training data includes sample role intention information and sample event text, and the impact sample value of the corresponding annotated sample event text, and then, training the impact value evaluation model based on the training data. Optionally, training the impact value evaluation model includes: inputting the sample role intention information and the sample event text into the pre-constructed initial impact value evaluation model for processing, and obtaining the impact prediction value corresponding to the sample event text; calculating the current loss using a preset loss function according to the impact sample value and the impact prediction value; updating the model parameters of the initial impact value evaluation model according to the current loss until the initial impact value evaluation model converges to obtain the trained impact value evaluation model.

[0058] The following describes the implementation method of determining the impact value in conjunction with the impact value evaluation model.

[0059] In one embodiment of the present disclosure, Figure 2 As shown, Figure 2 A flow chart of determining the influence value of an event text provided by an embodiment of the present disclosure, wherein the role intention information of a specified role and the event text of the specified role are input into an influence value evaluation model for processing, and the influence value of each event text on the specified role realizing the role intention is obtained, including the following steps:

[0060] Step 201, inputting the role intention information of the designated role and the event text of the designated role into the impact value evaluation model for processing, and obtaining the trajectory shape parameters corresponding to the event text of the designated role and the position information of each event text on the trajectory.

[0061] In this embodiment, the input of the impact value evaluation model is the role intention information and the event text sequence, and the output is the trajectory shape parameters of the trajectory and the position information of each event text in the event text sequence on the trajectory.

[0062] The model training process of the influence value evaluation model is as follows: obtain training data, each of which includes sample role intention information, sample event text, and corresponding annotated trajectory shape parameters and location information of the sample event text on the trajectory, and then train the influence value evaluation model based on the training data. Among them, the trajectory shape parameters are used to characterize the shape of the trajectory.

[0063] Optionally, multiple track types are set according to the rules, and different types of character tracks are represented by the track types, and each track type has corresponding track shape parameters.

[0064] As an example, the trajectory types include positive type, flat type, negative type, and open type, as shown in the accompanying figure, Figure 3 This is a diagram of the shape of a positive trajectory, which is monotonically increasing, indicating that as the plot develops, the character is closer to the process of realizing the character's intention, such as achieving the character's goal or satisfying the character's needs. Figure 4 This is another shape diagram of a positive trajectory. This trajectory first decreases and then increases, indicating that as the plot develops, the character is further away from achieving the character's intention due to factors such as the environment and other characters, and then the character overcomes obstacles and gets closer to achieving the character's intention. Figure 5 It is a schematic diagram of a flat trajectory. The trajectory is always within a certain range, indicating that as the plot develops, the character's subjective events and objective events have no significant impact on the realization of the character's intentions. Figure 6 This is a diagram of the shape of a negative trajectory, which first rises and then falls. Figure 7 It is a schematic diagram of the shape of another negative trajectory, which is monotonically decreasing. The open trajectory is used to indicate that the trajectory does not belong to any of the above cases.

[0065] Step 202, obtaining a trajectory expression that conforms to the trajectory shape parameters, and determining the influence value of each event text according to the position information of each event text on the trajectory and the trajectory expression.

[0066] In this embodiment, after determining the trajectory shape parameters of the trajectory, a trajectory expression that conforms to the trajectory shape parameters is generated, and then, according to the position information of each event text on the trajectory, the mapping point corresponding to each event text on the trajectory is determined, and the mapping point of each event text is substituted into the trajectory expression to calculate the influence value of each event text. For example, taking the xy coordinate system as an example, a trajectory expression that conforms to the trajectory shape parameters is determined in the xy coordinate system, and the position information of each event text on the trajectory includes the x coordinate value corresponding to each event text. For each event text, the x coordinate value corresponding to the event text is substituted into the trajectory expression to calculate the y coordinate value, and the y coordinate value is determined as the influence value of the event text, thereby obtaining the influence value of each event text.

[0067] In one embodiment of the present disclosure, Figure 8 As shown, Figure 8 A flowchart of another method of determining the influence value of an event text provided by an embodiment of the present disclosure is provided. The role intention information of a specified role and the event text of the specified role are input into the influence value evaluation model for processing to obtain the influence value of each event text on the specified role realizing the role intention, including the following steps:

[0068] Step 801, inputting the role intention information of the specified role and the event text of the specified role into the influence value evaluation model for processing, obtaining the influence type of each event text on the specified role's realization of the role intention, and the confidence level of each event text belonging to the influence type.

[0069] In this embodiment, the input of the impact value evaluation model is the role intention information and the event text, and the output is the impact type of each event text and the confidence that each event text belongs to the impact type.

[0070] The model training process of the influence value evaluation model is as follows: obtain training data, each piece of training data includes sample role intention information, sample event text, and the impact type of the corresponding annotated sample event text, and then train the influence value evaluation model based on the training data. Among them, the impact type includes positive impact and negative impact. Positive impact means that the event has a positive impact on the role, making the role closer to achieving the role intention, and negative impact means that the event has a negative impact on the role, making the role further away from achieving the role intention.

[0071] Step 802: Determine the impact value change between every two event texts that are adjacent in time sequence according to the impact type and confidence level of each event text.

[0072] Optionally, when the impact type of the current event text is positive, the confidence is proportional to the increase in the impact value, wherein the increase in the impact value is the increase in the impact value of the current event text relative to the impact value of the previous event text. When the impact type of the current event text is negative, the confidence is proportional to the decrease in the impact value, wherein the decrease in the impact value is the decrease in the impact value of the current event text relative to the impact value of the previous event text.

[0073] Optionally, multiple change levels are preset, and the confidence interval corresponding to each change level is set, and the impact value change amount corresponding to each change level is set. After obtaining the impact type and confidence of each event text, the confidence corresponding to the current event text is matched with multiple confidence intervals to determine a target confidence interval containing the confidence corresponding to the current event text, obtain a target change level corresponding to the target confidence interval, and use the impact value change amount corresponding to the target change level as the impact value change amount between the current event text and the previous event text.

[0074] Step 803: Determine the influence value of each event text according to the influence value variation between every two event texts that are adjacent in time sequence.

[0075] In this embodiment, based on the influence value change between every two event texts adjacent in time sequence, by presetting the influence value of the first event text, the influence value of each event text can be determined in time sequence, thereby obtaining the influence value of each event text.

[0076] As an example, taking the subjective perspective as an example, combined with the inner needs of Zhang's character, it can be determined that the impact types of event text 1 "Zhang was ignored by students" and event text 2 "Zhang was ignored by his parents-in-law" are both negative impacts, while the impact type of event text 3 "Zhang killed his parents-in-law" is positive. The confidence of event 3 is 0.8, and the confidence of event 2 is 0.2. The impact value change is determined through the impact type and confidence of the event text, and then the impact value of each event text is calculated based on the impact value change.

[0077] Step 104 , based on the time sequence information of each event text, display processing is performed on the impact value of each event text on the specified character's realization of the character's intention.

[0078] In this embodiment, the timing information of the event text is used to indicate the timing position of the event text in the script text. According to the timing information of each event text and the impact value corresponding to each event text, the event point of each event text is generated in the view, and then, the event points of each event text are connected in sequence according to the timing information of each event text, the character fate trajectory of the specified character is generated, and the character fate trajectory is displayed. Optionally, the display mode of the corresponding event point in the character fate trajectory can also be adjusted according to the proportion of the corresponding event of each event text in the plot, and the display mode includes but is not limited to shape, color, etc.

[0079] According to the technical solution of the embodiment of the present disclosure, by processing the script text, the plot summary text of the character is obtained, and the event text of the character is generated according to the character intention information and the plot summary text. Then, the character intention information and the event text of the character are input into the influence value evaluation model for processing to obtain the influence value of each event text on the realization of the character intention by the character. Based on the timing information of each event text, the influence value of each event text is displayed. Thus, the influence value of each event on the realization of the character intention can be accurately determined, and the character fate trajectory is formed and displayed through the influence value of each event. The character fate trajectory can be accurately generated and displayed, and the development trend of the character is intuitively visualized to assist users to understand the development trend of the character more intuitively, and provide support for analyzing the characters of historical hit dramas. For example, it can be used in the following scenarios: classifying the character type by the character fate trajectory; reflecting the development rhythm of the character in the overall story progress; combining the fate trajectory of the likable characters in the historical hit dramas to establish a structured reference, such as the character trough should appear in the 5% area.

[0080] Based on the above embodiments, Fig. 9 A flow chart of another method for displaying a character's fate trajectory provided by an embodiment of the present disclosure is shown as follows: Fig. 9 As shown, after inputting the role intention information of the designated role and at least one plot summary text of the designated role into the event generation model for processing, the method further includes the following steps:

[0081] Step 901, generating event texts for designated roles and determining the event type of each event text.

[0082] In this embodiment, the role intention information is used to represent the role's goal and / or the role's needs. The role intention information includes the role goal text and the role need text. For example, the role goal text is "maintaining and saving precarious family relationships", and the role need text is "recognition and respect, sense of security, sense of control, guilt and redemption". The event generation model is used to perform semantic analysis and other processing on all plot summary texts of the specified role based on the role intention information, extract events that affect the specified role to realize the role intention, so as to generate event texts for each event, and determine the event type of each event text.

[0083] The training process of the event generation model is as follows: obtain training data, each of which includes sample character intention information, at least one sample plot summary text, and the corresponding annotated sample event text and the event type of the sample event text, and then train the event generation model based on the training data. Among them, the event type includes active events and passive events. Active events refer to events that the character actively promotes in order to achieve the user's intention, including the behavior of the character due to changes in his inner needs. Passive events refer to events that affect the character and prompt the character to take action.

[0084] As an example, the event types corresponding to the event texts “difficulties in teachers’ work” and “increased family pressure” are both passive events.

[0085] Step 902: According to the event type of the event text, the event text and the role intention information matching the event type are input into the influence value evaluation model for processing to obtain the influence value of each event text.

[0086] In one embodiment of the present disclosure, for each first event text whose event type is an active event, the role target text of the specified role and the first event text of the specified role are input into the influence value evaluation model for processing to obtain the first influence value corresponding to each first event text.

[0087] In one embodiment of the present disclosure, for each second event text whose event type is a passive event, the role requirement text of the specified role and the second event text of the specified role are input into the influence value evaluation model for processing to obtain the second influence value corresponding to each second event text.

[0088] Step 903: Display the first impact value corresponding to each first event text, and / or display the second impact value corresponding to each second event text.

[0089] In this embodiment, the trend chart reflecting the trajectory of the character's destiny is divided into a subjective perspective and an objective perspective. The subjective perspective is used to represent the subjective emotions of the character in the path of satisfying his or her inner needs, and the objective perspective is used to represent the progress of objective events in the process of the character achieving external goals. From the subjective perspective, the focus is on the distance between the character's passive events and the character's needs, and from the objective perspective, the focus is on the distance between the character's active events and the character's goals.

[0090] In one embodiment of the present disclosure, the first impact value corresponding to each first event text is displayed, including: generating an event point of each first event text in the view according to the timing information of each first event text and the first impact value corresponding to each first event text, and then connecting the event points of each first event text in sequence according to the timing information of each first event text, generating the first character fate trajectory of the specified character in an objective perspective, and displaying the first character fate trajectory. Fig.10 , Fig.10 A schematic diagram showing the trajectory of a character's fate from an objective perspective is shown.

[0091] In one embodiment of the present disclosure, the second impact value corresponding to each second event text is displayed, including: generating an event point of each second event text in the view according to the timing information of each second event text and the second impact value corresponding to each second event text, and then connecting the event points of each second event text in sequence according to the timing information of each second event text, generating the second character fate trajectory of the specified character from the subjective perspective, and displaying the second character fate trajectory. Fig.11 , Fig.11 A schematic diagram showing the fate trajectory of a character from a subjective perspective.

[0092] In the disclosed embodiment, by determining the event text of active events and determining the event text of passive events, the influence value of each event text is determined in combination with the character goal text and the character requirement text, and then the character's fate trajectory from an objective perspective is generated through the influence value of the active event, and the character's fate trajectory from an intuitive perspective is generated through the influence value of the passive event, so as to meet the demand for intuitively displaying the character's development trend in multiple dimensions.

[0093] Based on the above embodiment, the present disclosure also proposes a model training method, which includes the following steps:

[0094] Acquire training data; each piece of training data includes sample character intention information, at least one sample plot summary text, and corresponding annotated sample event text, and an impact sample value of the corresponding annotated sample event text;

[0095] Inputting sample character intention information and at least one sample plot summary text into a pre-built initial event generation model, generating a predicted event text through the initial event generation model; calculating a first current loss based on the sample event text and the predicted event text through a preset loss function; updating model parameters of the initial event generation model based on the first current loss until the initial event generation model converges, thereby obtaining a trained event generation model;

[0096] The sample role intention information and sample event text are input into a pre-built initial influence value assessment model for processing to obtain the influence prediction value corresponding to the sample event text; the second current loss is calculated according to the influence sample value and the influence prediction value through the preset loss function; the model parameters of the initial influence value assessment model are updated according to the second current loss until the initial influence value assessment model converges, thereby obtaining the trained influence value assessment model.

[0097] Optionally, the sample role intention information and the sample event text are input into a pre-built initial influence value assessment model for processing to obtain an influence prediction value corresponding to the sample event text, including: inputting the sample role intention information and the sample event text into the initial influence value assessment model, and determining the trajectory shape parameters corresponding to the sample event text and the position information of each sample event text on the trajectory through the initial influence value assessment model; obtaining an expression that conforms to the trajectory shape parameters through the initial influence value assessment model, and determining the influence prediction value of each sample event text based on the position information and expression of each sample event text on the trajectory.

[0098] Optionally, the sample role intention information and the sample event text are input into a pre-built initial influence value assessment model for processing to obtain an influence prediction value corresponding to the sample event text, including: inputting the sample role intention information and the sample event text into the initial influence value assessment model, determining the influence type of the sample event text and the confidence that each sample event text belongs to the influence type through the initial influence value assessment model; determining the influence prediction value change between every two sample event texts that are adjacent in time sequence according to the influence type and confidence of each sample event text through the initial influence value assessment model; determining the influence prediction value of each sample event text according to the influence prediction value change between every two sample event texts that are adjacent in time sequence through the initial influence value assessment model.

[0099] Fig.12 A schematic diagram of the structure of a device for displaying a character's fate trajectory provided by an embodiment of the present disclosure, such as Fig.12 As shown, the display device of the character's destiny trajectory includes: a first generating module 1201, a second generating module 1202, a third generating module 1203, and a display module 1204.

[0100] The first generation module 1201 is used to input the script text into the plot generation model for processing, and generate a plurality of plot summary texts corresponding to a plurality of plot scenes one by one; the plurality of plot summary texts include at least one plot summary text of a specified character;

[0101] The second generation module 1202 is used to obtain the role intention information of the specified role, and input the role intention information of the specified role and at least one plot summary text of the specified role into the event generation model for processing to generate an event text of the specified role;

[0102] The third generation module 1203 is used to input the role intention information of the specified role and the event text of the specified role into the impact value evaluation model for processing, so as to obtain the impact value of each event text on the specified role to realize the role intention;

[0103] The display module 1204 is used to display the impact value of each event text on the specified role's realization of the role intention based on the time sequence information of each event text.

[0104] In one embodiment of the present disclosure, the second generating module 1202 is specifically used to:

[0105] The character biography text of the specified character is input into the big text model for processing to generate the character intention information of the specified character.

[0106] In one embodiment of the present disclosure, the second generating module 1202 is specifically used to:

[0107] Inputting the role intention information of the specified role and the event text of the specified role into the influence value evaluation model for processing, obtaining the trajectory shape parameters corresponding to the event text of the specified role and the position information of each event text on the trajectory;

[0108] Get an expression that meets the trajectory shape parameters, and determine the influence value of each event text based on the position information of each event text on the trajectory and the expression.

[0109] In one embodiment of the present disclosure, the device further includes:

[0110] A first training module is used to obtain first training data; each piece of first training data includes sample character intention information, sample event text, and corresponding annotated trajectory shape parameters and position information of the sample event text on the trajectory;

[0111] Based on the first training data, an influence value evaluation model is trained.

[0112] In one embodiment of the present disclosure, the second generating module 1202 is specifically used to:

[0113] Inputting the role intention information of the specified role and the event text of the specified role into the influence value evaluation model for processing, obtaining the influence type of each event text on the realization of the role intention of the specified role and the confidence level of each event text belonging to the influence type;

[0114] According to the impact type and confidence of each event text, determine the impact value change between every two event texts that are adjacent in time sequence;

[0115] The influence value of each event text is determined according to the influence value change between every two event texts that are adjacent in time sequence.

[0116] In one embodiment of the present disclosure, the device further includes:

[0117] A second training module is used to obtain second training data; each piece of second training data includes sample role intention information, sample event text, and an impact type of the corresponding annotated sample event text;

[0118] Based on the second training data, an influence value evaluation model is trained.

[0119] In one embodiment of the present disclosure, the second generating module 1202 is further configured to:

[0120] The event type for which each event text is generated.

[0121] In one embodiment of the present disclosure, the device further includes:

[0122] A third training module is used to obtain third training data; each piece of third training data includes sample character intention information, at least one sample plot summary text, and corresponding annotated sample event text and event type of the sample event text;

[0123] Based on the third training data, an event generation model is trained.

[0124] In one embodiment of the present disclosure, the role intention information includes the role target text, and the third generation module 1203 is specifically used to:

[0125] For each first event text whose event type is an active event, the role target text of the specified role and the first event text of the specified role are input into the influence value evaluation model for processing to obtain the first influence value corresponding to each first event text.

[0126] In one embodiment of the present disclosure, the display module 1204 is specifically used for:

[0127] Generating an event point of each first event text in the view according to the time sequence information of each first event text and the first impact value corresponding to each first event text;

[0128] Connecting the event points of each first event text in sequence according to the time sequence information of each first event text, generating the first character fate trajectory of the designated character from an objective perspective;

[0129] Displays the fate trajectory of the first character.

[0130] In one embodiment of the present disclosure, the role intention information includes role requirement text, and the third generation module 1203 is specifically used to:

[0131] For each second event text whose event type is a passive event, the role requirement text of the specified role and the second event text of the specified role are input into the influence value evaluation model for processing to obtain the second influence value corresponding to each second event text.

[0132] In one embodiment of the present disclosure, the display module 1204 is specifically used for:

[0133] Generating an event point of each second event text in the view according to the time sequence information of each second event text and the second impact value corresponding to each second event text;

[0134] Connecting the event points of each second event text in sequence according to the time sequence information of each second event text, and generating the second character fate trajectory of the designated character from a subjective perspective;

[0135] Displays the fate trajectory of the second character.

[0136] The device for displaying the fate trajectory of a character provided in the embodiment of the present disclosure can execute any method for displaying the fate trajectory of a character provided in the embodiment of the present disclosure, and has the corresponding functional modules and beneficial effects of the execution method. For the contents not described in detail in the embodiment of the device of the present disclosure, reference can be made to the description in any method embodiment of the present disclosure.

[0137] The present disclosure also provides an electronic device, which includes one or more processors and a memory. The processor may be a central processing unit (CPU) or other forms of processing units with data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device to perform desired functions. The memory may include one or more computer program products, and the computer program product may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and / or cache memory (cache), etc. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on a computer-readable storage medium, and the processor may run the program instructions to implement the method of the above embodiment of the present disclosure and / or other desired functions. Various contents such as input signals, signal components, noise components, etc. may also be stored in the computer-readable storage medium.

[0138] In one example, the electronic device may further include: an input device and an output device, and these components are interconnected through a bus system and / or other forms of connection mechanisms. In addition, the input device may also include, for example, a keyboard, a mouse, etc. The output device may output various information to the outside, including determined distance information, direction information, etc. The output device may include, for example, a display, a speaker, a printer, and a communication network and a remote output device connected thereto, etc. In addition, according to the specific application, the electronic device may also include any other appropriate components such as a bus, an input / output interface, etc.

[0139] In addition to the above methods and devices, the embodiments of the present disclosure may also be a computer program product, which includes computer program instructions. When the computer program instructions are executed by a processor, the processor executes any method provided by the embodiments of the present disclosure.

[0140] The computer program product may be written in any combination of one or more programming languages ​​to write program code for performing the operations of the disclosed embodiments, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0141] In addition, the embodiments of the present disclosure may also be a computer-readable storage medium on which computer program instructions are stored. When the computer program instructions are executed by a processor, the processor executes any method provided by the embodiments of the present disclosure.

[0142] Computer readable storage media can adopt any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium can include, for example, but is not limited to, a system, device or device of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0143] 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.

[0144] 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 displaying a character's fate trajectory, characterized in that: The method comprises: Inputting the script text into the plot generation model for processing to generate a plurality of plot summary texts corresponding to a plurality of plot scenes; the plurality of plot summary texts include at least one plot summary text of a designated character; Acquiring role intention information of the designated role, and inputting the role intention information of the designated role and at least one plot summary text of the designated role into an event generation model for processing to generate an event text of the designated role; Inputting the role intention information of the designated role and the event text of the designated role into the influence value evaluation model for processing, and obtaining the influence value of each event text on the designated role realizing the role intention; Based on the time sequence information of each event text, the influence value of each event text on the designated character's realization of the character intention is displayed.

2. The method according to claim 1, characterized in that The obtaining the role intention information of the specified role includes: The character biography text of the designated character is input into the big text model for processing to generate the character intention information of the designated character.

3. The method according to claim 1, characterized in that The step of inputting the role intention information of the designated role and the event text of the designated role into the impact value evaluation model for processing to obtain the impact value of each event text on the designated role realizing the role intention includes: Inputting the role intention information of the designated role and the event text of the designated role into the impact value evaluation model for processing, obtaining the trajectory shape parameters corresponding to the event text of the designated role and the position information of each event text on the trajectory; An expression that conforms to the trajectory shape parameters is obtained, and the influence value of each event text is determined according to the position information of each event text on the trajectory and the expression.

4. The method according to claim 1, characterized in that The step of inputting the role intention information of the designated role and the event text of the designated role into the impact value evaluation model for processing to obtain the impact value of each event text on the designated role realizing the role intention includes: Inputting the role intention information of the designated role and the event text of the designated role into the influence value evaluation model for processing, obtaining the influence type of each event text on the designated role realizing the role intention, and the confidence level of each event text belonging to the influence type; Determine, according to the impact type and the confidence of each event text, the impact value change between every two event texts that are adjacent in time sequence; The influence value of each event text is determined according to the influence value change between each two event texts that are adjacent in time sequence.

5. The method according to claim 1, characterized in that After inputting the role intention information of the designated role and at least one plot summary text of the designated role into the event generation model for processing, the method further includes: The event type for each of the event texts generated.

6. The method according to claim 5, characterized in that The role intention information includes a role target text, and the role intention information of the designated role and the event text of the designated role are input into the influence value evaluation model for processing to obtain the influence value of each event text on the designated role to realize the role intention, including: For each first event text whose event type is an active event, the role target text of the designated role and the first event text of the designated role are input into the influence value evaluation model for processing to obtain a first influence value corresponding to each first event text.

7. The method according to claim 6, characterized in that The display processing of the influence value of each event text on the realization of the role intention by the designated role based on the time sequence information of each event text includes: generating an event point of each first event text in a view according to the time sequence information of each first event text and the first impact value corresponding to each first event text; Connecting the event points of each of the first event texts in sequence according to the time sequence information of each of the first event texts to generate a first character fate trajectory of the designated character from an objective perspective; Display the fate trajectory of the first character.

8. The method according to claim 5, characterized in that The role intention information includes a role requirement text, and the role intention information of the designated role and the event text of the designated role are input into the impact value evaluation model for processing to obtain the impact value of each event text on the designated role realizing the role intention, including: For each second event text whose event type is a passive event, the role requirement text of the designated role and the second event text of the designated role are input into the impact value evaluation model for processing to obtain a second impact value corresponding to each second event text.

9. The method according to claim 8, characterized in that The display processing of the influence value of each event text on the realization of the role intention by the designated role based on the time sequence information of each event text includes: generating an event point of each second event text in a view according to the time sequence information of each second event text and the second impact value corresponding to each second event text; Connecting the event points of each second event text in sequence according to the time sequence information of each second event text, to generate the second character fate trajectory of the designated character from a subjective perspective; Displays the fate trajectory of the second character.

10. A model training method, characterized in that: The method comprises: Acquire training data; each piece of the training data includes sample character intention information, at least one sample plot summary text, and corresponding annotated sample event text, and an impact sample value of the corresponding annotated sample event text; Inputting the sample character intention information and the at least one sample plot summary text into a pre-built initial event generation model, and generating a predicted event text through the initial event generation model; Calculating a first current loss according to the sample event text and the predicted event text by using a preset loss function; updating the model parameters of the initial event generation model according to the first current loss until the initial event generation model converges, thereby obtaining a trained event generation model; Inputting the sample role intention information and the sample event text into a pre-built initial impact value evaluation model for processing to obtain an impact prediction value corresponding to the sample event text; Calculating a second current loss according to the impact sample value and the impact prediction value by using a preset loss function; The model parameters of the initial influence value assessment model are updated according to the second current loss until the initial influence value assessment model converges, thereby obtaining a trained influence value assessment model.

11. The method according to claim 10, characterized in that The step of inputting the sample role intention information and the sample event text into a pre-built initial impact value evaluation model for processing to obtain an impact prediction value corresponding to the sample event text includes: Inputting the sample role intention information and the sample event text into the initial impact value evaluation model, and determining the trajectory shape parameters corresponding to the sample event text and the position information of each sample event text on the trajectory through the initial impact value evaluation model; An expression conforming to the trajectory shape parameters is obtained through the initial impact value evaluation model, and the impact prediction value of each sample event text is determined according to the position information of each sample event text on the trajectory and the expression.

12. The method according to claim 10, characterized in that The step of inputting the sample role intention information and the sample event text into a pre-built initial impact value evaluation model for processing to obtain an impact prediction value corresponding to the sample event text includes: Inputting the sample role intention information and the sample event text into the initial impact value evaluation model, and determining the impact type of the sample event text and the confidence that each sample event text belongs to the impact type through the initial impact value evaluation model; Determine the change in the impact prediction value between every two sample event texts that are adjacent in time sequence according to the impact type and the confidence of each sample event text through the initial impact value evaluation model; The impact prediction value of each sample event text is determined by the initial impact value evaluation model according to the impact prediction value change between each two sample event texts that are adjacent in time sequence.

13. A device for displaying the fate trajectory of a character, characterized in that: include: A first generation module is used to input the script text into the plot generation model for processing, and generate a plurality of plot summary texts corresponding to a plurality of plot scenes one by one; The plurality of plot summary texts include at least one plot summary text for a specified character; A second generation module is used to obtain the role intention information of the specified role, and input the role intention information of the specified role and at least one plot summary text of the specified role into an event generation model for processing to generate an event text of the specified role; A third generation module is used to input the role intention information of the designated role and the event text of the designated role into the impact value evaluation model for processing, so as to obtain the impact value of each event text on the designated role to realize the role intention; The display module is used to display the influence value of each event text on the realization of the role intention of the designated role based on the time sequence information of each event text.

14. An electronic device, characterized in that: include: processor; a memory for storing instructions executable by the processor; The processor is used to read the executable instructions from the memory and execute the instructions to implement the method for displaying the character's destiny trajectory described in any one of claims 1-9 above, and / or to implement the training method described in any one of claims 10-12 above.

15. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, which, when executed by a processor, implements the method for displaying the character's destiny trajectory described in any one of claims 1-9, and / or implements the training method described in any one of claims 10-12.

Images