Strong plot determination method, system, equipment and medium

Abstract

The invention relates to a strong plot determination method, system and device and a medium, and the method comprises the steps: splitting a script content into a plurality of sessions, and determining a target object based on the split sessions; inputting the plurality of sessions into the large language model one by one to determine plots corresponding to each session; and for any plot, determining a real-time attribute of a target object in the plot, and taking the plot with the real-time attribute meeting a preset condition as a strong plot. The strong plot points in the script can be quickly and efficiently determined, and the determined strong plot points are more objective compared with traditional current plot points.

Description

Technical Field

[0001] The present application relates to the field of computers, and in particular to a method, system, device and medium for determining strong plot points. Background Art

[0002] In the field of film and television work evaluation, plot points are an important indicator that highlights the changes in the script plot. They are usually an incident or event surrounding the characters in the script. The story content in a script is built up by a large number of plot points. However, when evaluating the script content, it is impossible to evaluate all the events corresponding to the plot points. Therefore, it is necessary to select representative plot points from the corresponding plot points as "strong plot points". Use this as the main evaluation basis to evaluate the script.

[0003] However, the traditional way to confirm plot points is to manually read and understand the script content. It takes a lot of time to read the script, especially for some high-quality scripts, it takes more time to extract the corresponding strong plot points. In addition, since manual reading and comprehension are subject to the cognition and value orientation of the corresponding personnel, it is difficult to obtain an objective result for different people's identification criteria of strong plot points. This will greatly affect the further evaluation of the script. For this reason, in some cases, a method of simultaneous evaluation by multiple people will be used to balance the influence of personal subjective bias. However, the method of multi-person evaluation will bring about diversity of results, reduce the efficiency of script evaluation, and increase the cost of script evaluation. Summary of the invention

[0004] The present application provides a method for determining strong plot points to solve the problems of lack of objectivity and low efficiency in traditional extraction of strong plot points.

[0005] In a first aspect of the present application, a method for determining a strong plot point is proposed, comprising:

[0006] Split the script content into multiple scenes, and determine the target audience based on the multiple scenes after the split;

[0007] Inputting the plurality of scenes into the large language model one by one to determine the plot point corresponding to each scene;

[0008] For any of the plot nodes, the real-time attribute of the target object in the plot node is determined, and the plot node whose real-time attribute meets the preset conditions is taken as a strong plot node.

[0009] In some embodiments of the present application, the script content is divided into multiple scenes, and the target object is determined based on the multiple scenes after the division, including:

[0010] Input the text content corresponding to each scene into the large language model to determine the objects that appear in each scene;

[0011] Determine the number of pages in the script corresponding to the text content of each scene;

[0012] Determine the number of script pages for each object based on the number of scenes in which each object appears;

[0013] The object whose script page number meets the preset threshold is taken as the target object.

[0014] In some embodiments of the present application, the plurality of scenes are input into the large language model one by one to determine the plot point corresponding to each scene, including:

[0015] For any of the sessions, input the text content corresponding to the session into the large language model to obtain the event corresponding to the session;

[0016] Based on all the scenes, obtain an event set corresponding to the script content;

[0017] The events in the event set are analyzed, and one or more scenes of the same event are taken as a plot node to obtain a plot node set corresponding to the script content.

[0018] In some embodiments of the present application, for any of the plot nodes, determining the real-time attribute of the target object in the plot node, and taking the plot node whose real-time attribute meets the preset conditions as a strong plot node includes:

[0019] For any of the plot nodes, determining an associated event in the plot node corresponding to the target object;

[0020] Determine first identity attribute information of other objects involved in the association event to the target object;

[0021] Acquire the second identity attribute information of the other object to the target object at a moment before the association event;

[0022] If the first identity attribute information is different from the second identity attribute information, the plot point is taken as a strong plot point.

[0023] In some embodiments of the present application, for any of the plot nodes, determining the real-time attribute of the target object in the plot node, and taking the plot node whose real-time attribute meets the preset conditions as a strong plot node includes:

[0024] For any of the plot nodes, determining an associated event in the plot node corresponding to the target object;

[0025] Determining a first value of the emotion or sentiment of the target object in the associated event;

[0026] Acquire a second value of the emotion or sentiment of the target object at a time before the associated event;

[0027] If the first value and the second value are different, the plot point is regarded as a strong plot point.

[0028] In some embodiments of the present application, for any of the plot nodes, determining the real-time attribute of the target object in the plot node, and taking the plot node whose real-time attribute meets the preset conditions as a strong plot node includes:

[0029] For any of the plot nodes, determining an associated event in the plot node corresponding to the target object;

[0030] Determine the non-target objects affected by the associated event and the number of the non-target objects;

[0031] The associated events whose number exceeds the first predetermined value are regarded as special events, and the plot nodes corresponding to the special events are regarded as strong plot nodes.

[0032] In some embodiments of the present application, for any of the plot nodes, determining the real-time attribute of the target object in the plot node, and taking the plot node whose real-time attribute meets the preset conditions as a strong plot node includes:

[0033] For any of the plot nodes, determining an associated event in the plot node corresponding to the target object;

[0034] When the number of target objects that meet the preset condition in the associated event exceeds a second threshold, the plot node corresponding to the associated event is taken as a strong plot node.

[0035] In a second aspect of the present application, a strong plot point determination system is also proposed, comprising:

[0036] An object determination module, used to split the script content into multiple scenes, and determine the target object based on the multiple scenes after the split;

[0037] A plot point determination module, used for inputting the plurality of scenes into the large language model one by one to determine the plot point corresponding to each scene;

[0038] The strong plot node determination module is used to determine the real-time attribute of the target object in any of the plot nodes, and take the plot node whose real-time attribute meets the preset conditions as a strong plot node.

[0039] The third aspect of the present application also proposes a computer device: at least one communication interface; at least one bus connected to the at least one communication interface; at least one processor connected to the at least one bus; at least one memory connected to the at least one bus, wherein the processor is configured to execute the method for determining strong plot nodes described in any one of the above embodiments.

[0040] The fourth aspect of the present application further proposes a computer storage medium storing computer executable instructions, wherein the computer executable instructions are used to execute the method for determining strong plot points in any one of the above-mentioned implementation modes.

[0041] Compared with the prior art, the above technical solution provided by the embodiment of the present application has the following advantages: the method provided by the embodiment of the present application determines the main characters according to the scenes of the script, and uses the main characters as clues to take the plot points where the attributes of the main characters change as strong plot points. Objective and fair strong plot points are extracted. And with the help of the processing power of the large language model, it can be implemented quickly and efficiently. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0043] 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.

[0044] One or more embodiments are exemplarily described by pictures in the corresponding drawings, and these exemplified descriptions do not constitute limitations on the embodiments. Elements with the same reference numerals in the drawings represent similar elements, and unless otherwise stated, the figures in the drawings do not constitute proportional limitations.

[0045] Figure 1 A flowchart of a method for determining a strong plot point provided in an embodiment of the present application;

[0046] Figure 2 A schematic diagram of a process for determining a target object provided in an embodiment of the present application;

[0047] Figure 3 A schematic diagram of a process for determining plot points in a script provided in an embodiment of the present application;

[0048] Figure 4 A schematic diagram of a process for determining a strong plot point based on a character's identity attribute provided in an embodiment of the present application;

[0049] Figure 5 A schematic diagram of a process for determining strong plot points based on character emotion changes provided in an embodiment of the present application;

[0050] Figure 6 A schematic diagram of a process for determining a strong plot point based on the impact of an event corresponding to a character on other characters provided in an embodiment of the present application;

[0051] Figure 7 A schematic diagram of a process for determining strong plot points based on relationships between characters provided in an embodiment of the present application;

[0052] Figure 8 A schematic diagram of the structure of a strong plot point determination system provided in an embodiment of the present application;

[0053] Fig. 9 A schematic diagram of the structure of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0054] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0055] The disclosure below provides many different embodiments or examples to implement different structures of the present invention. In order to simplify the disclosure of the present invention, the parts and settings of specific examples are described below. Of course, they are only examples, and the purpose is not to limit the present invention. In addition, the present invention can repeat reference numbers and / or letters in different examples. This repetition is for the purpose of simplification and clarity, and does not itself indicate the relationship between the various embodiments and / or settings discussed.

[0056] The present application provides a method for determining strong plot points to solve the problems of lack of objectivity and low efficiency in traditional extraction of strong plot points.

[0057] like Figure 1 As shown, in the first aspect of the present application, a method for determining a strong plot point is proposed, comprising:

[0058] Step S1, splitting the script content into multiple scenes, and determining the target object based on the multiple scenes after the split;

[0059] Step S2, inputting the plurality of scenes into the large language model one by one to determine the plot point corresponding to each scene;

[0060] Step S3: for any of the plot nodes, determine the real-time attribute of the target object in the plot node, and take the plot node whose real-time attribute meets the preset conditions as a strong plot node.

[0061] In the embodiment of the present application, the script is a text of multiple episodes, each episode includes multiple scenes, and each scene includes a scene number and information such as the location, time, and characters of the scene;

[0062] The target object represents a character in the script that meets the preset conditions, that is, it is defined as the main character in this application. The main character will be used as an example to explain the scheme of this application below;

[0063] A large language model refers to an artificial intelligence model with text processing capabilities. The text processing capabilities mainly involved in this application include the ability to recognize, understand, reason about text content, and generate text;

[0064] A plot point refers to one or more scenes in a script. The events described in these scenes are the same event. Generally, the time, scenes, and plots within the plot nodes are continuous. A strong plot point refers to an event that occurs around the main character, and certain attributes of the main character in the event meet predetermined conditions.

[0065] Real-time attributes refer to preset data used to evaluate a character's status, such as the character's emotions, i.e., psychological states such as happiness, sadness, excitement, and calmness; the character's personality, i.e., whether the character is a good person or a bad person, or an elite or an ordinary person, etc., and the recognition given by other people through the character's social activities; the character's own subjective judgment on external people, things, and objects, i.e., the character's outlook on life, values, and world view, etc., and the character's subjective evaluation of external things based on his or her experience and cognition.

[0066] In some embodiments of the present application, characters or main characters are not limited to people, but also include some items that have a strong influence on the characters in the play. For example, the corresponding real-time attributes of treasures or treasures in certain types of scripts are determined according to their influence on the characters or things in the script. For example, some treasures or life forms in fantasy or science fiction scripts have some human-like intelligence or cognitive abilities. In this regard, on the one hand, they can be treated as characters and given the real-time attributes of the characters. On the other hand, the corresponding special real-time attributes can be determined based on the abilities they possess or the effects they set on other characters in the script.

[0067] In step S1, the script is split into scenes, the characters appearing in each scene are obtained, the number of scenes in which each character appears is further counted, and the number of scenes is used as a reference for determining whether the corresponding character is a main character. For example, in some embodiments of the present application, the characters whose number of scenes reaches a predetermined value are regarded as main characters. Alternatively, in some embodiments, the number of appearances of the characters is sorted, and the characters with the highest number of appearances are regarded as main characters.

[0068] In step S2, the script content corresponding to each scene, that is, the corresponding text content, is input into the large language model, and the large language model understands the text content of each scene and outputs the plot points existing in the corresponding scene.

[0069] In some embodiments of the present application, when using a large language model to output the plot points of a scene, different large language models have inconsistent understandings of the plot points. It is necessary to customize different prompt words for different large language models, and provide cases consisting of certain script content and corresponding plot points to prompt the large language model to infer the corresponding plot nodes of each scene according to the text logic in the case.

[0070] In step S3, after the plot point is obtained, the real-time attributes of the main characters in the plot point are further determined, and then it is judged whether the real-time attributes of the main characters meet the preset conditions. If they meet the conditions, the corresponding plot point is regarded as a strong plot point. Specifically, for example, in a plot point, a main character has an emotional breakdown due to a series of events, and the plot point is regarded as a strong plot point.

[0071] In some embodiments of the present application, the real-time attributes of the main characters in the plot nodes can be determined by extracting the corresponding real-time attributes from the text content corresponding to the plot nodes output by the large language model in step S2 through keyword matching.

[0072] In addition, in some embodiments, the real-time attributes of the main characters in the plot nodes can also input the text content corresponding to the plot nodes into the large language model, and the large language model gives the corresponding real-time attributes. For different real-time attributes, when the large language model is difficult to accurately output, it can be supplemented by prompt word engineering, giving specific examples, so that the large language model can give the real-time attributes of the main characters in the corresponding plot nodes according to the logical content in the examples. That is, different prompt words need to be provided for different real-time attributes.

[0073] like Figure 2 As shown, in some embodiments of the present application, the script content is divided into multiple scenes, and the target object is determined based on the multiple scenes after the division, including:

[0074] Step S21, inputting the text content corresponding to each scene into the large language model to determine the objects appearing in each scene;

[0075] Step S22, determining the number of script pages corresponding to the text content corresponding to each scene in the script;

[0076] Step S23, determining the number of script pages of each object according to the number of scenes in which each object appears;

[0077] Step S24: The object whose script page number meets the preset threshold is taken as the target object.

[0078] As mentioned above, objects, i.e. characters, are not limited to people. Some objects in some fantasy scripts can also be regarded as "main characters". Therefore, if the characters appearing in a scene are determined only by the introduction at the beginning of the corresponding scene of the script, the problem of missing characters is inevitable. For this reason, in this embodiment, the text content corresponding to the scene is input into the large language model, and the large language model outputs the characters appearing in each scene after understanding the text content. Furthermore, different prompt words are formulated according to the needs for character recognition in different types of scripts.

[0079] After determining the characters that appear in each scene corresponding to the script, all the characters that appear in the script can be determined. Further, from the perspective of the characters, determine which scenes each character appears in, and determine the number of script pages for the corresponding scenes. The number of script pages of the scenes in which the characters appear is used as the number of script pages of the characters in the script, which is used to indicate the proportion of the corresponding characters in the script, or the weight of the characters. Based on this, the characters whose script pages exceed the predetermined value are regarded as the main characters. For example, character A appears in 3 scenes, and the page numbers of these 3 scenes are 9 pages, 5 pages, and 8 pages respectively. Then the number of script pages corresponding to character A is 9+5+8=24 pages. If the threshold is 10 pages, character A is regarded as the main character. If the threshold is 50, character A is just an ordinary character and cannot be regarded as the main character.

[0080] In some embodiments of the present application, when determining the main character, it is also possible to determine whether the main character is the main character based on the proportion of the number of script pages corresponding to the character to the total number of script pages. For example, with 5% as a restriction condition, the characters with more than 5% of the script pages are considered the main characters.

[0081] like Figure 3 As shown, in some embodiments of the present application, the plurality of scenes are input into the large language model one by one to determine the plot point corresponding to each scene, including:

[0082] Step S31: for any of the sessions, input the text content corresponding to the session into a large language model to obtain an event corresponding to the session;

[0083] Step S32, based on all the scenes, obtaining an event set corresponding to the script content;

[0084] Step S33, analyzing the events in the event set, taking one or more scenes of the same event as a plot node, and obtaining a plot node set corresponding to the script content.

[0085] In this embodiment, when determining the plot points, the text content of each scene is first input into the large language model, and the event corresponding to each scene is obtained through the large language model. Then, the text content of all scenes is input into the large language model, and then all events in the script are obtained to form an event set of the script.

[0086] Furthermore, the events in the event set are analyzed to confirm which events belong to the same event, that is, an event may be the same event in multiple scenes. For example, taking a war drama as an example, a battle may be divided into multiple scenes. When confirming events in scenes, since the text content of the scenes is input into the large language model alone, the events output by the large language model must be independent and multiple. However, these multiple events should actually be one event, but because the content input into the large model is segmented, multiple events appear. Therefore, it is necessary to analyze the events in the event set and merge the same events into one event. If the same event belongs to the same scene, the corresponding scene is determined as a plot node. If the same event involves multiple scenes, multiple scenes are regarded as the same plot node. In this way, all the plot nodes of the script are obtained to form a plot node set. In this way, based on the principle of event identity, the redundant scenes output due to scene splitting or text cutting when input into the large model are effectively aggregated.

[0087] In some embodiments of the present application, when analyzing the events of the event set, the characters and behaviors in the events can be used as the judgment principle for whether they are the same event. Further, multiple events are input into the large language model, and the large language model determines whether the input multiple events are the same event based on this principle. If the large language model outputs the same event, the scenes corresponding to the multiple events are further determined. If they are the same scene, the corresponding scenes are used as a plot point. If multiple scenes are involved, the multiple scenes are used as a plot point.

[0088] like Figure 4 As shown, in some embodiments of the present application, for any of the plot nodes, determining the real-time attribute of the target object in the plot node, and taking the plot node whose real-time attribute meets the preset conditions as a strong plot node includes:

[0089] Step S41, for any of the plot nodes, determining an associated event in the plot node corresponding to the target object;

[0090] Step S42: determining first identity attribute information of other objects involved in the association event to the target object;

[0091] Step S43, obtaining the second identity attribute information of the other object to the target object at the moment before the association event;

[0092] Step S44: If the first identity attribute information is different from the second identity attribute information, the plot point is regarded as a strong plot point.

[0093] In this embodiment, the identity attribute information represents the character's personality, background identity, and other information, and is an attribute that represents the character's identity in the play. It is usually formed by other characters' understanding of a certain character, and is a label given to a certain character by others on their social attributes. In this application, the plot point where the label of the main character changes is regarded as a strong plot point. The target object represents the main character, and the non-target object represents other characters relative to the main character.

[0094] In an embodiment of the present application, after the plot points are determined, for each plot point, the associated events related to the main characters in the plot point are determined based on the main characters that have been determined.

[0095] Further, determine whether the occurrence of the related event leads to a change in other characters' understanding of the main character's identity attributes. Specifically, first confirm other characters' understanding of the main character before the related event occurs, that is, the first identity attribute information, and then determine other characters' understanding of the main character after the related event occurs, that is, the second identity attribute information. If the first identity attribute information and the second identity attribute information are different, the corresponding plot point is regarded as a strong plot point. Specifically, for example, in a certain plot point, other people originally have a positive impression of a main character as a good person, but in a sudden event, his behavior or certain things reveal his true face as a bad person, then this plot point is regarded as a strong plot point.

[0096] In some embodiments of the present application, the confirmation of the identity attributes of characters in a plot point generally requires obtaining the recognition of the characters by other characters before and after the corresponding event occurs. If it is not possible to accurately obtain the recognition of the identity attributes of the main character by other characters before the corresponding event occurs in a plot point, the recognition of the identity attributes of the main character by other characters in the plot points before the plot point can be used as the first identity attribute.

[0097] In some embodiments of the present application, the confirmation of the identity attributes of the main characters can be identified by a large language model, that is, the text content corresponding to the plot node is input into the large language model, and the large model outputs the recognition of the identity attributes of other characters by each character in the plot node. For those who do not clearly reflect their recognition of the identity attributes of a certain character in the plot node, the previous identity attributes of the character are used. If it is the first plot node, the identity attributes of the corresponding character are set to the default values. When the text content of subsequent plot points is input into the large language model to output the identity attributes of the corresponding characters, if the character attributes given by the large language model change, the default values ​​are replaced by the changed character attributes.

[0098] Furthermore, when identifying the identity attributes of a person through a large language model, a corresponding prompt word project is appropriately formulated according to the capabilities of the large language model used, supplemented by the large language model reference, to accurately obtain the identity attributes of the person.

[0099] Furthermore, in some embodiments of the present application, when determining a plot point through a large language model, the identity attributes corresponding to each character in the plot point are confirmed so that they can be directly obtained for subsequent use.

[0100] like Figure 5 As shown, in some embodiments of the present application, for any of the plot nodes, determining the real-time attribute of the target object in the plot node, and taking the plot node whose real-time attribute meets the preset conditions as a strong plot node includes:

[0101] Step S51, for any of the plot nodes, determining an associated event in the plot node corresponding to the target object;

[0102] Step S52: determining a first value of the emotion or sentiment of the target object in the associated event;

[0103] Step S53, obtaining a second value of the emotion or feeling of the target object at a time before the associated event;

[0104] Step S54: If the first value and the second value are different, the plot point is regarded as a strong plot point.

[0105] In this embodiment, when determining whether a plot point is a strong plot point, the change in the mood or emotion of the main character is used to confirm whether the corresponding plot point is used as a strong plot point. Specifically, first confirm the associated event corresponding to the main character in the plot point. Then determine the mood or emotional state of the main character in the associated event, for example, it can be any one of happy, sad, excited, and calm. Further, obtain the mood or emotional state of the main character before the associated event, which can also be any one of the states of happy, sad, excited, and calm. If the mood or emotional state of the main character before and after the event is inconsistent, the corresponding plot node is used as a strong plot node.

[0106] Further, in some embodiments of the present application, when determining whether a plot point is a strong plot point by using the emotion or emotional changes of a character, the recognition of the change of the emotion or emotional state of the character can be achieved through a large language model. And when the plot point is output through the large language model, the emotion or emotional state of each character in the plot point is output at the same time.

[0107] Furthermore, in some embodiments of the present application, the text content corresponding to the plot point can be separately input into the large language model again and supplemented with the corresponding prompt words, and the large language model gives the corresponding character's emotion or emotional state. Output the emotion or emotional state of each character in the plot point. This prevents different types of target results from affecting the accuracy of the output content of the large language model.

[0108] In addition, in some embodiments of the present application, when determining the emotional or emotional state of a character in a plot point, other artificial intelligence models can also be used to output the emotional or emotional state of the character, thereby reducing the cost of using a large language model. Specifically, first, based on the main characters previously confirmed in the plot point, further determine the text content related to the description or behavior of the main characters in the text content corresponding to the plot point. Further, input the text content related to the main characters into the corresponding artificial intelligence model, and the corresponding artificial intelligence model outputs the emotional state of the corresponding text content. And use the emotional state as the emotional state of the corresponding main character.

[0109] like Figure 6 As shown, in some embodiments of the present application, for any of the plot nodes, determining the real-time attribute of the target object in the plot node, and taking the plot node whose real-time attribute meets the preset conditions as a strong plot node includes:

[0110] Step S61: for any of the plot nodes, determine the associated event corresponding to the target object in the plot node;

[0111] Step S62: determining the non-target objects affected by the associated event and the number of the non-target objects;

[0112] Step S63: The associated events whose number exceeds the first predetermined value are regarded as special events, and the plot nodes corresponding to the special events are regarded as strong plot nodes.

[0113] In this embodiment, when determining whether a corresponding plot point is a strong plot point, it can also be determined based on the degree of influence of the associated event corresponding to the main character on other characters.

[0114] Specifically, for any plot node, first determine the associated events related to the main characters in the plot node. For example, input the text content corresponding to the plot node into the large language model, and use the corresponding prompt words to let the large language model confirm the events in the input text content and the characters involved in the events. Further determine whether the character of the event output by the large language model is the main character. If it is the main character, then the event is used as the associated event related to the main character in the plot node.

[0115] Furthermore, when determining other characters corresponding to the associated event, it should be noted that in this embodiment, the target object refers to the corresponding main character, and the non-target object refers to other characters relative to the main character. In some cases, other characters include some of the main characters. For example, there are main characters A and B in a certain plot point. If the main character A participates in a certain event, the main character B does not directly participate, or is a bystander, that is, does not participate in a certain event. However, the event not only affects other non-main characters, but also affects the main character B. In this case, the main character B is also counted in the number of non-target objects affected by the associated event.

[0116] Further, when determining the number of other non-main characters affected by the associated events of the main characters, a large language model can be used to perform this process. Based on the principle of first determining the impact of the events associated with the main characters on other non-main characters, and then determining the number of impacts. First, the text content of the plot point corresponding to the associated event is input into the large language model, and the large language model confirms the scope of influence of the associated event, that is, whether it only affects the people involved in the event, or affects everyone in the script. For example, in political film and television dramas, the introduction of a certain policy, that is, in a broad sense, affects everyone, then it can be regarded as affecting all the characters in the script. In some other types of scripts, the impact of a certain event is generally other minor characters related to the main character, or other main characters. Therefore, in some embodiments of the present application, corresponding prompt word projects are formulated for cases with different impact ranges, that is, event cases with different impact ranges are used as reference text content, and the large language model is based on event cases with different impact ranges. Confirm which case type the event in the winning plot node input into the large language model belongs to, and finally output the corresponding type. For example, as mentioned above, in a prompt word project of the all-type, at least one text content involving an event or story that will affect all people must be provided as a case. If the prompt word project of the partial type is provided, at least one text content involving an event or story that will affect some people must also be provided as a case. After inputting into the large language model, the large language model is asked to confirm which type of event the input plot point corresponds to.

[0117] Furthermore, if the type output by the large language model is all types, the number of all characters (including main characters and non-main characters) in the script is used as the number of characters affected by the corresponding associated event in the plot node. If it is a partial type, the number of characters involved or appearing in the plot node is used as the number of characters affected by the corresponding associated event.

[0118] In addition, in some embodiments, if the things or objects occurring in the corresponding associated event also involve characters in other plot nodes or scenes, the characters that will be affected in these other scenes or plot nodes will also be counted as the number of characters affected by the associated event in the plot node.

[0119] Finally, after determining the number of characters affected by the associated events in the plot node, it is determined whether the number of affected characters exceeds the corresponding preset value. If it exceeds the corresponding preset value, the plot node is regarded as a strong plot node.

[0120] In some embodiments of the present application, the preset value of the number of characters used to measure the influence of the associated event can be set according to a certain ratio based on the number of characters in the script. For example, it can be set to 20%, that is, if it exceeds 20% of the total number of characters in the script, the corresponding plot point can be regarded as a strong plot point.

[0121] like Figure 7 As shown, in some embodiments of the present application, for any of the plot nodes, determining the real-time attribute of the target object in the plot node, and taking the plot node whose real-time attribute meets the preset conditions as a strong plot node includes:

[0122] Step S71: for any of the plot nodes, determine the associated event corresponding to the target object in the plot node;

[0123] Step S72: When the number of target objects that meet the preset conditions in the associated event exceeds a second threshold, the plot node corresponding to the associated event is taken as a strong plot node.

[0124] In this embodiment, the target object of the preset condition refers to the target object with contradictions. When determining whether the corresponding plot point satisfies a strong plot point according to the real-time attributes of the main characters, whether the corresponding plot point is a strong plot point can be determined according to the contradiction relationship between the main characters.

[0125] Specifically, the text content corresponding to the plot node can be first input into the corresponding large language model, and the large language model determines the associated events associated with the main characters. At the same time, the behavior of the main characters in the associated events is confirmed. Then the behavior of the main characters is input into the large language model, and the large language model determines whether there is a contradiction in the behavior between the main characters. If the result of the large language model output is that there is a contradiction, the relationship between the corresponding main characters is identified as a contradictory relationship. Further, after the large language model determines the relationship between all the main characters, the number of main characters with contradictions is counted. If the number exceeds a preset threshold, the corresponding plot node is regarded as a strong plot node. In some embodiments of the present application, the number of thresholds can be flexibly set as needed, for example, 2 are set, that is, as long as there is a contradiction between two main characters, the corresponding plot node is considered to be a strong plot node.

[0126] Furthermore, when the large language model is used to confirm that there is a contradiction in the relationship between the main characters in the event, corresponding prompt word cases can be formulated according to different types of contradictions, so that the large language model can confirm whether there is a contradiction in the behavior of the characters in the text content of the input plot point according to the cases in the prompt words. The types of contradictions include: emotional contradictions, interest contradictions, physical conflicts, ideological contradictions, and value contradictions. Conflicts are the mutual consumption or confrontation caused by the behavior or language of characters in social activities. Furthermore, for different types of contradictions, appropriate cases are selected as part of the prompt words for reference by the large language model, so that the large language model can make more accurate judgments on the relationship between characters.

[0127] The above technical solution provided by the embodiment of the present application has the following advantages over the prior art: the method provided by the embodiment of the present application determines the main characters according to the scenes of the script, and uses the main characters as clues to take the plot points where the attributes of the main characters change as strong plot points. Objective and fair strong plot points are extracted. And with the help of the processing power of the large language model, it can be achieved quickly and efficiently. In some embodiments, the strong plot points are determined by mixing the large language model with other models, effectively reducing the cost of using the large language model.

[0128] In some embodiments of the present application, as described above, strong plot nodes are further rated according to the number of changes in the real-time attributes of the main characters in the corresponding plot nodes. If only one real-time attribute changes, it is rated as a one-star strong plot node, if two real-time attributes change, it is rated as a two-star strong node, and so on, for progressive rating. For example, if in addition to the contradictions between the main characters at a certain strong plot node, the identity background and other attributes of one or more characters change, then the strong plot node is rated as a two-star strong plot node. If only one real-time attribute changes, it is rated as a one-star strong plot node.

[0129] like Figure 8 As shown, in the second aspect of the present application, a strong plot point determination system is also proposed, comprising:

[0130] The object determination module 1 is used to divide the script content into multiple scenes and determine the target object based on the divided multiple scenes;

[0131] A plot point determination module 2, used for inputting the plurality of scenes into the large language model one by one to determine the plot point corresponding to each scene;

[0132] The strong plot node determination module 3 is used to determine the real-time attribute of the target object in any of the plot nodes, and to take the plot nodes whose real-time attributes meet the preset conditions as strong plot nodes.

[0133] like Fig. 9 As shown, the third aspect of the present application further proposes a computer device: at least one communication interface 912; at least one bus 914 connected to the at least one communication interface 912; at least one processor 911 connected to the at least one bus 914; at least one memory 913 connected to the at least one bus, wherein the processor 911 is configured to execute the steps of the method for determining a strong plot point described in any one of the above embodiments:

[0134] Split the script content into multiple scenes, and determine the target audience based on the multiple scenes after the split;

[0135] Inputting the plurality of scenes into the large language model one by one to determine the plot point corresponding to each scene;

[0136] For any of the plot nodes, the real-time attribute of the target object in the plot node is determined, and the plot node whose real-time attribute meets the preset conditions is taken as a strong plot node.

[0137] In some embodiments of the present application, the script content is divided into multiple scenes, and the target object is determined based on the multiple scenes after the division, including:

[0138] Input the text content corresponding to each scene into the large language model to determine the objects that appear in each scene;

[0139] Determine the number of pages in the script corresponding to the text content of each scene;

[0140] Determine the number of script pages for each object based on the number of scenes in which each object appears;

[0141] The object whose script page number meets the preset threshold is taken as the target object.

[0142] In some embodiments of the present application, the plurality of scenes are input into the large language model one by one to determine the plot point corresponding to each scene, including:

[0143] For any of the sessions, input the text content corresponding to the session into the large language model to obtain the event corresponding to the session;

[0144] Based on all the scenes, obtain an event set corresponding to the script content;

[0145] The events in the event set are analyzed, and one or more scenes of the same event are taken as a plot node to obtain a plot node set corresponding to the script content.

[0146] In some embodiments of the present application, for any of the plot nodes, determining the real-time attribute of the target object in the plot node, and taking the plot node whose real-time attribute meets the preset conditions as a strong plot node includes:

[0147] For any of the plot nodes, determining an associated event in the plot node corresponding to the target object;

[0148] Determine first identity attribute information of other objects involved in the association event to the target object;

[0149] Acquire the second identity attribute information of the other object to the target object at a moment before the association event;

[0150] If the first identity attribute information is different from the second identity attribute information, the plot point is taken as a strong plot point.

[0151] In some embodiments of the present application, for any of the plot nodes, determining the real-time attribute of the target object in the plot node, and taking the plot node whose real-time attribute meets the preset conditions as a strong plot node includes:

[0152] For any of the plot nodes, determining an associated event in the plot node corresponding to the target object;

[0153] Determining a first value of the emotion or sentiment of the target object in the associated event;

[0154] Acquire a second value of the emotion or sentiment of the target object at a time before the associated event;

[0155] If the first value and the second value are different, the plot point is regarded as a strong plot point.

[0156] In some embodiments of the present application, for any of the plot nodes, determining the real-time attribute of the target object in the plot node, and taking the plot node whose real-time attribute meets the preset conditions as a strong plot node includes:

[0157] For any of the plot nodes, determining an associated event in the plot node corresponding to the target object;

[0158] Determine the non-target objects affected by the associated event and the number of the non-target objects;

[0159] The associated events whose number exceeds the first predetermined value are regarded as special events, and the plot nodes corresponding to the special events are regarded as strong plot nodes.

[0160] In some embodiments of the present application, for any of the plot nodes, determining the real-time attribute of the target object in the plot node, and taking the plot node whose real-time attribute meets the preset conditions as a strong plot node includes:

[0161] For any of the plot nodes, determining an associated event in the plot node corresponding to the target object;

[0162] When the number of target objects that meet the preset condition in the associated event exceeds a second threshold, the plot node corresponding to the associated event is taken as a strong plot node.

[0163] The fourth aspect of the present application further provides a computer storage medium storing computer executable instructions, wherein the computer executable instructions are used to execute the steps of the video annotation method in any one of the above embodiments:

[0164] Split the script content into multiple scenes, and determine the target audience based on the multiple scenes after the split;

[0165] Inputting the plurality of scenes into the large language model one by one to determine the plot point corresponding to each scene;

[0166] For any of the plot nodes, the real-time attribute of the target object in the plot node is determined, and the plot node whose real-time attribute meets the preset conditions is taken as a strong plot node.

[0167] In some embodiments of the present application, the script content is divided into multiple scenes, and the target object is determined based on the multiple scenes after the division, including:

[0168] Input the text content corresponding to each scene into the large language model to determine the objects that appear in each scene;

[0169] Determine the number of pages in the script corresponding to the text content of each scene;

[0170] Determine the number of script pages for each object based on the number of scenes in which each object appears;

[0171] The object whose script page number meets the preset threshold is taken as the target object.

[0172] In some embodiments of the present application, the plurality of scenes are input into the large language model one by one to determine the plot point corresponding to each scene, including:

[0173] For any of the sessions, input the text content corresponding to the session into the large language model to obtain the event corresponding to the session;

[0174] Based on all the scenes, obtain an event set corresponding to the script content;

[0175] The events in the event set are analyzed, and one or more scenes of the same event are taken as a plot node to obtain a plot node set corresponding to the script content.

[0176] In some embodiments of the present application, for any of the plot nodes, determining the real-time attribute of the target object in the plot node, and taking the plot node whose real-time attribute meets the preset conditions as a strong plot node includes:

[0177] For any of the plot nodes, determining an associated event in the plot node corresponding to the target object;

[0178] Determine first identity attribute information of other objects involved in the association event to the target object;

[0179] Acquire the second identity attribute information of the other object to the target object at a moment before the association event;

[0180] If the first identity attribute information is different from the second identity attribute information, the plot point is taken as a strong plot point.

[0181] In some embodiments of the present application, for any of the plot nodes, determining the real-time attribute of the target object in the plot node, and taking the plot node whose real-time attribute meets the preset conditions as a strong plot node includes:

[0182] For any of the plot nodes, determining an associated event in the plot node corresponding to the target object;

[0183] Determining a first value of the emotion or sentiment of the target object in the associated event;

[0184] Acquire a second value of the emotion or sentiment of the target object at a time before the associated event;

[0185] If the first value and the second value are different, the plot node is regarded as a strong plot node.

[0186] In some embodiments of the present application, for any of the plot nodes, determining the real-time attribute of the target object in the plot node, and taking the plot node whose real-time attribute meets the preset conditions as a strong plot node includes:

[0187] For any of the plot nodes, determining an associated event in the plot node corresponding to the target object;

[0188] Determine the non-target objects affected by the associated event and the number of the non-target objects;

[0189] The associated events whose number exceeds the first predetermined value are regarded as special events, and the plot nodes corresponding to the special events are regarded as strong plot nodes.

[0190] In some embodiments of the present application, for any of the plot nodes, determining the real-time attribute of the target object in the plot node, and taking the plot node whose real-time attribute meets the preset conditions as a strong plot node includes:

[0191] For any of the plot nodes, determining an associated event in the plot node corresponding to the target object;

[0192] When the number of target objects that meet the preset condition in the associated event exceeds a second threshold, the plot node corresponding to the associated event is taken as a strong plot node.

[0193] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0194] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a general hardware platform, and of course, by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the relevant technology can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0195] It should be understood that the terms used herein are only for the purpose of describing specific example embodiments and are not intended to be limiting. Unless the context clearly indicates otherwise, the singular forms "one", "an" and "said" as used herein may also be meant to include plural forms. The terms "include", "comprise", "contain", and "have" are inclusive, and therefore specify the existence of stated features, steps, operations, elements and / or parts, but do not exclude the existence or addition of one or more other features, steps, operations, elements, parts, and / or combinations thereof. The method steps, processes, and operations described herein are not interpreted as necessarily requiring them to be performed in the specific order described or illustrated, unless the execution order is clearly indicated. It should also be understood that additional or alternative steps may be used.

[0196] The foregoing is merely a specific embodiment of the present invention, which enables those skilled in the art to understand or implement the present invention. 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 invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A method for determining strong plot points, characterized in that: include: Split the script content into multiple scenes, and determine the target audience based on the multiple scenes after the split; Inputting the plurality of scenes into the large language model one by one to determine the plot point corresponding to each scene; For any of the plot nodes, the real-time attribute of the target object in the plot node is determined, and the plot node whose real-time attribute meets the preset conditions is taken as a strong plot node.

2. The method according to claim 1, characterized in that The step of dividing the script content into multiple scenes and determining the target object based on the multiple scenes after the division includes: Input the text content corresponding to each scene into the large language model to determine the objects that appear in each scene; Determine the number of pages in the script corresponding to the text content of each scene; Determine the number of script pages for each object based on the number of scenes in which each object appears; The object whose script page number meets the preset threshold is taken as the target object.

3. The method according to claim 1, characterized in that The step of inputting the plurality of scenes into the large language model one by one to determine the plot point corresponding to each scene includes: For any of the sessions, input the text content corresponding to the session into the large language model to obtain the event corresponding to the session; Based on all the scenes, obtain an event set corresponding to the script content; The events in the event set are analyzed, and one or more scenes of the same event are taken as a plot node to obtain a plot node set corresponding to the script content.

4. The method according to claim 1, characterized in that: The step of determining the real-time attribute of the target object in any of the plot nodes and taking the plot node whose real-time attribute meets a preset condition as a strong plot node includes: For any of the plot nodes, determining an associated event in the plot node corresponding to the target object; Determine first identity attribute information of other objects involved in the association event to the target object; Acquire the second identity attribute information of the other object to the target object at a moment before the association event; If the first identity attribute information is different from the second identity attribute information, the plot point is taken as a strong plot point.

5. The method according to claim 1, characterized in that The step of determining the real-time attribute of the target object in any of the plot nodes and taking the plot node whose real-time attribute meets a preset condition as a strong plot node includes: For any of the plot nodes, determining an associated event in the plot node corresponding to the target object; Determining a first value of the emotion or sentiment of the target object in the associated event; Acquire a second value of the emotion or sentiment of the target object at a time before the associated event; If the first value and the second value are different, the plot point is regarded as a strong plot point.

6. The method according to claim 1, characterized in that The step of determining the real-time attribute of the target object in any of the plot nodes and taking the plot node whose real-time attribute meets a preset condition as a strong plot node includes: For any of the plot nodes, determining an associated event in the plot node corresponding to the target object; Determine the non-target objects affected by the associated event and the number of the non-target objects; The associated events whose number exceeds the first predetermined value are regarded as special events, and the plot nodes corresponding to the special events are regarded as strong plot nodes.

7. The method according to claim 1, characterized in that The step of determining the real-time attribute of the target object in any of the plot nodes and taking the plot node whose real-time attribute meets a preset condition as a strong plot node includes: For any of the plot nodes, determining an associated event in the plot node corresponding to the target object; When the number of target objects that meet the preset condition in the associated event exceeds a second threshold, the plot node corresponding to the associated event is taken as a strong plot node.

8. A system for determining strong plot points, characterized in that: include: An object determination module, used to split the script content into multiple scenes, and determine the target object based on the multiple scenes after the split; A plot point determination module, used for inputting the plurality of scenes into the large language model one by one to determine the plot point corresponding to each scene; The strong plot node determination module is used to determine the real-time attribute of the target object in any of the plot nodes, and take the plot node whose real-time attribute meets the preset conditions as a strong plot node.

9. A computer device: at least one communication interface; at least one bus connected to the at least one communication interface; at least one processor connected to the at least one bus; at least one memory connected to the at least one bus, wherein: The processor is configured to execute the method for determining strong plot points according to any one of claims 1 to 7.

10. A computer storage medium storing computer executable instructions, wherein the computer executable instructions are used to execute the method for determining strong plot points according to any one of claims 1 to 7.

Images