KR102658528B1 - Method and system for story composition based on artificial intelligence - Google Patents

Abstract

An artificial intelligence-based story composition method is provided that allows users to easily construct sentences according to the topic of the story, thereby constructing a story by generating sentences that reflect the user's intention. The story composition method generates example sentences by combining a number of word data input from the user terminal, generates sentence data based on the affinity index between the characters included in it, and uses this to create a new story that reflects the user's intention. can be created.

Description

Method and system for story composition based on artificial intelligence}

The present invention relates to an artificial intelligence-based story composition method and story composition system that constructs a story by generating sentences to reflect the user's intention.

Stories such as fairy tales and novels are composed of a number of sentences with a certain connection structure according to the author's intention. These fairy tales and novels are provided to general readers in the form of paper books or e-books, or are widely used in the field of student education.

Meanwhile, it is quite difficult for ordinary people to compose a story based on appropriate arguments that reflect the same intentions as the author of the story. Accordingly, writing education is being provided to non-experts such as the general public and students to create their own new stories through existing stories such as fairy tales and novels.

However, existing writing education focuses only on having students write a story by presenting a topic or situation as data, and then modifying it through editing or presenting opinions.

Accordingly, existing writing education does not provide help to the general public who are not accustomed to composing multiple sentences with appropriate arguments depending on the topic or situation and structurally combining them to compose a story.

The present invention is intended to solve the above-mentioned problems, and is an artificial intelligence-based story composition method that allows users to easily construct sentences according to the topic of the story, thereby constructing a story by generating sentences that reflect the user's intention. The goal is to provide a story composition system for this purpose.

An artificial intelligence-based story composition method according to an embodiment of the present invention includes the steps of setting topic information and character information for a story in a story composition server; Classifying a plurality of word data transmitted from the user terminal to the story composition server according to the hexagram principle, and sorting the plurality of classified word data to generate one or more example sentences; and generating one or more sentence data by correcting the example sentence.

For this story composition method, the story composition system according to an embodiment of the present invention includes a user terminal and a story composition server connected to each other through a wired/wireless network.

In addition, the story composition server includes a setting module that sets topic information for the story to be composed, character information, and an affinity index between characters in the character information; And classifying and sorting a plurality of word data transmitted from the user terminal according to the hexadecimal principle, generating one or more example sentences from the plurality of sorted word data, and correcting the example sentences based on the set favorability index. It includes a sentence construction module that generates one or more sentence data from example sentences.

The artificial intelligence-based story composition method according to the present invention generates example sentences by combining a plurality of word data input from the user terminal, generates sentence data based on the affinity index between the characters included in the sentence, and uses this. This allows you to create a new story. Therefore, the present invention can easily and quickly construct a story through a large number of sentence data that reflects the user's intention.

In addition, the present invention allows the user to change the affinity between the characters in the story, so that a new story with a completely different meaning from the existing story can be created according to the user's selection.

1 is a diagram showing the schematic configuration of a story composition system according to an embodiment of the present invention.
Figure 2 is a diagram showing the configuration of the story composition server of Figure 1.
Figure 3 is a diagram showing a story composition method according to an embodiment of the present invention.
Figures 4 to 7 are diagrams showing various embodiments of the story composition method of the present invention.

The terms used in the specification and claims are general terms selected in consideration of their functions in various embodiments of the present invention. However, these terms may vary depending on the intention of technicians working in the field, legal or technical interpretation, and the emergence of new technologies. Additionally, some terms may be terms arbitrarily selected by the applicant. These terms may be interpreted as defined in this specification, and if there is no specific term definition, they may be interpreted based on the overall content of this specification and common technical knowledge in the relevant technical field.

In addition, the same reference numerals or symbols in each drawing attached to this specification indicate parts or components that perform substantially the same function. For convenience of explanation and understanding, different embodiments will be described using the same reference numbers or symbols. That is, even if all components having the same reference number are shown in multiple drawings, the multiple drawings do not represent one embodiment.

Additionally, in this specification and claims, terms including ordinal numbers such as 'first', 'second', etc. may be used to distinguish between components. These ordinal numbers are used to distinguish identical or similar components from each other, and the meaning of the term should not be interpreted limitedly due to the use of these ordinal numbers. For example, components combined with these ordinal numbers should not be interpreted as having a limited order of use or arrangement based on the number. If necessary, each ordinal number may be used interchangeably.

In this specification, singular expressions include plural expressions, unless the context clearly dictates otherwise. In the present application, terms such as 'comprise' or 'constitute' are intended to designate the presence of characteristics, numbers, steps, operations, components, parts, or combinations thereof described in the specification, but are intended to indicate the presence of one or more other It should be understood that this does not exclude in advance the possibility of the presence or addition of features, numbers, steps, operations, components, parts or combinations thereof.

Additionally, in an embodiment of the present invention, when a part is connected to another part, this includes not only direct connection but also indirect connection through other media. In addition, the fact that a part includes a certain component does not mean that other components are excluded, but that it may further include other components, unless specifically stated to the contrary.

FIG. 1 is a diagram showing the schematic configuration of a story composition system according to an embodiment of the present invention, and FIG. 2 is a diagram showing the configuration of the story composition server of FIG. 1.

Referring to the drawings, the story composition system of this embodiment may be configured to include a story composition server 100 and a user terminal 200 that can be connected and communicate with each other through a wired / wireless network 300.

The user terminal 200 can access the story composition server 100 through the network 300. The user terminal 200 may transmit story composition information including the topic, characters, situations, etc. of the story to be constructed to the story composition server 100.

Additionally, the user terminal 200 may receive one or more image data for a story to be composed from the story composition server 100 and transmit one or more word data to the story composition server 100 in response. Here, word data may be parts of speech such as nouns, verbs, adjectives, adverbs, and conjunctions that represent people, places, situations, times, actions, etc. for image data.

The story composition server 100 may extract one or more contextual image data corresponding to the topic of the story according to the story composition information transmitted from the user terminal 200 and transmit it to the user terminal 200.

Additionally, the story composition server 100 may generate one or more sentence data for the story from one or more word data transmitted from the user terminal 200.

The story composition server 100 stores sentence data sequentially according to its creation order, and combines the stored sentence data to construct a new story corresponding to the story topic.

This story composition server 100 may include a setting module 110, a sentence composition module 120, and a database 130.

In the database 130, data on each of existing stories, such as published or published fairy tales or novels, may be stored as learning data. Additionally, the database 130 may store story data generated by the sentence construction module 120, which will be described later.

Accordingly, the story composition server 100 of this embodiment can perform learning through machine learning or deep learning using learning data stored in the database 130. In addition, by generating one or more sentence data from a plurality of word data transmitted from the user terminal 200 according to the learning results, a new story corresponding to the theme of the preset story can be constructed.

As an artificial intelligence network model used for learning of the story composition server 100, various models such as RNN (Recurrent Neural Network), DNN (Deep Neural Network), and DRNN (Dynamic Recurrent Neural Network) can be used.

The setting module 110 of the story composition server 100 sets the subject of the story to be composed, that is, the title, based on the story composition information transmitted from the user terminal 200, and determines the characters and the affinity between the characters accordingly. can be set. The setting module 110 may include a topic and character setting unit 111 and a favorability setting unit 113.

The topic and character setting unit 111 can set the topic and characters of the story to be composed based on the story composition information. The topic and character setting unit 111 provides a plurality of story topics to the user terminal 200, and can set one selected from the user terminal 200 as the topic of the story to be composed.

In addition, the topic and character setting unit 111 provides a plurality of characters according to the set story topic to the user terminal 200, and can set one or more characters selected from the user terminal 200 as characters for the story topic. there is. The topic and character setting unit 111 can output the set story topic and characters as topic information and character information, respectively.

The likeability setting unit 113 can set the likeability index between characters from the topic information and character information provided by the topic and character setting unit 111. The likeability index can be set as a score for mutual likeability between characters, and can be set to a value between 0 and 100.

The likeability setting unit 113 may set a likeability index between characters according to the mutual likeability of character information by referring to story data corresponding to the topic information among the existing story data stored in the database 130 based on the topic information.

Additionally, according to another embodiment of the present invention, the likeability setting unit 113 may set the likeability index between people in the person information based on the likeability information transmitted from the user terminal 200.

For example, the user terminal 200 may select at least one character from among the plurality of characters provided by the topic and character setting unit 111 and transmit the selection result to the story composition server 100. At this time, the user terminal 200 can set the favorability information between at least one selected person and transmit it to the story composition server 100. Accordingly, the likeability setting unit 113 can set the likeability index between people based on the likeability information transmitted from the user terminal 200.

The sentence composition module 120 constructs one or more sentences for the story to be composed based on the topic information, character information, and favorability index set by the setting module 110, and combines the composed sentences to create a story corresponding to the topic information. can be created. The sentence composition module 120 may include an image extraction unit 121, a word classification unit 123, a connection item recommendation unit 125, and a sentence composition unit 127.

The image extraction unit 121 may extract one or more image data for each situation from the database 130 based on subject information. The image extraction unit 121 may transmit the extracted image data to the user terminal 200.

The word classification unit 123 may classify and sort one or more word data transmitted from the user terminal 200 and generate one or more example sentences from the sorted word data. The word classification unit 123 can classify and sort one or more word data based on the hexadecimal principle.

The connected item recommendation unit 125 may recommend and output one or more connected items to be included in the example sentence generated by the word classification unit 123. Connected items can take the form of parts of speech such as verbs, adjectives, adverbs, and conjunctions.

The connected item recommendation unit 125 may recommend one or more connected items based on the favorability index set between the characters in the example sentence. The connection item recommendation unit 125 can recommend one of a positive connection item and a negative connection item depending on the size of the favorability index.

For example, when the mutual favorability index between the characters in the example sentence is 50 or more, the connected item recommendation unit 125 may recommend and output one or more positive connected items. On the other hand, if the mutual favorability index between the characters in the example sentence is less than 50, the connected item recommendation unit 125 may recommend and output one or more negative connected items.

The sentence composition unit 127 may correct the example sentence generated by the word classification unit 123 based on one or more connected items output from the connected item recommendation unit 125. The sentence composition unit 127 may generate and output one or more corrected example sentences.

The sentence composition unit 127 may transmit one or more generated corrected example sentences to the user terminal 200. Then, one corrected example sentence selected in the user terminal 200 can be generated as sentence data.

The sentence composition module 120 can sequentially store sentence data generated in the sentence composition unit 127 according to its creation order. Then, when the composition of the story according to the preset topic information is completed, that is, when a plurality of sentence data for the story to be constructed is generated, the sentence composition module 120 generates a story using the plurality of stored sentence data. You can.

Here, the sentence composition module 120 can store a plurality of sentence data and stories generated therefrom in the database 130. Additionally, the sentence construction module 120 can perform learning using the stored story as learning data.

As described above, the sentence construction module 120 of this embodiment generates one or more example sentences from a plurality of word data transmitted from the user terminal 200, and connects one or more based on the favorability index between the characters in the example sentences. Items are recommended, and one or more sentence data can be generated by correcting example sentences using the connection items. And, using sentence data, a new story that reflects the user's intention can be constructed according to preset topic information.

Figure 3 is a diagram showing a story composition method according to an embodiment of the present invention, and Figures 4 to 7 are diagrams showing various embodiments of the story composition method according to the present invention.

Hereinafter, the story composition method of this embodiment will be described in detail with reference to FIGS. 1 and 2 described above along with FIGS. 3 to 7.

Referring to the drawing, the story composition server 100 may transmit the topic of each of a plurality of stories, for example, the story title and information about the characters of the story, to the user terminal 200. The user terminal 200 can select a story to be composed among the topics of each of the plurality of stories received and select characters for the selected story. The setting module 110 of the story composition server 100 can set topic information and character information for the story to be composed based on the selection result of the user terminal 200 (S10).

Next, the setting module 110 can set the favorability index between the characters according to the character information from the content of the existing story corresponding to the topic information stored in the database 130 (S20).

This likability index can be set to a value between 0 and 100 between characters, and a relatively higher value can mean that the likability of each person is higher.

For example, as shown in Figure 4, [Kongjwi Patjwi] may be set as subject information, and [Kongjwi], [Patjwi], [Father], and [Wonnim] may be set as person information.

The setting module 110 can set the affinity index between the above-mentioned characters from the existing Kongjwi Patjwi story stored in the database 130. At this time, the highest affinity between [Kongjwi] and [Father] and [Kongjwi] and [Wonnim] can be set to the highest mutual favorability index, for example, 100. Additionally, the affinity index can be set to be relatively low between [Kongjwi] and [Patjwi], [Patjwi] and [Kongjwi], and [Wonnim] and [Patjwi].

Next, the story composition server 100 combines a plurality of word data transmitted from the user terminal 200 to generate one or more example sentences, and from this, one or more sentence data for the set story topic can be generated ( S30).

First, the story composition server 100 may extract one or more contextual image data for the story according to the topic information and transmit it to the user terminal 200 (S31).

The user terminal 200 may transmit a plurality of word data to the story composition server 100 in response to the received image data for each situation. The story composition server 100 may generate one or more example sentences from a plurality of word data transmitted from the user terminal 200 (S32).

The sentence composition module 120 of the story composition server 100 can receive a plurality of word data from the user terminal 200. The word classification unit 123 of the sentence construction module 120 may classify a plurality of word data according to the hexadecimal principle and sort the classified word data to generate one or more example sentences.

Referring to FIG. 5, when the topic information is set to [Kongjwi and Patjwi], the word data corresponding to the image data from the user terminal 200 is [Kongjwi], [Father], [Once upon a time], [A village], [Together] and [Live] can be transmitted.

The word classification unit 123 can classify multiple word data according to the hexadecimal principle. And, by sorting the classified word data, one example sentence can be created: [Once upon a time, Kongjwi's father lived together in a village].

In this way, since the word classification unit 123 generates example sentences by combining a plurality of word data input from the user terminal 200, the generated example sentences may be awkward sentences that do not conform to grammar.

Accordingly, the connected item recommendation unit 125 may recommend and output one or more connected items based on a preset favorability index (S33). At this time, the connection item may be recommended as a connection item with a positive meaning or a connection item with a negative meaning depending on the size of the favorability index.

As shown in Figures 4 and 5, the mutual favorability index between [Kongjwi] and [father] is set to 100. The connected item recommendation unit 125 determines that Kongjwi and the father in the example sentence are characters, and based on the preset affinity index between [Kongjwi] and [father], connects items with positive meaning, such as [Osondoson], [Happily]. ] can be recommended and printed.

Subsequently, the sentence composition unit 127 may correct the example sentence generated by the word classification unit 123 using one or more connection items recommended by the connection item recommendation unit 125 to generate a corrected example sentence. (S34).

As shown in Figure 5, the sentence composition unit 127 uses [Oson Doson] and [Happy] recommended in the connected item recommendation unit 125 to create two corrected example sentences, namely, [Kongjwi and Father in a Village Once upon a time] [Osondoson lived together] and [Kongjwi and his father lived happily together in a village once upon a time] can be created respectively.

Next, the sentence construction unit 127 transmits the generated plurality of corrected example sentences to the user terminal 200, and can generate sentence data corresponding to the image for each situation according to the selection of the user terminal 200 (S35) ).

The sentence composition unit 127 displays the connected item parts, that is, the parts of [Ohson Doson] and [Happy] among a number of corrected example sentences, separately, for example, by displaying different font colors or different sizes or thicknesses to display the user terminal (200). ) can be transmitted. Accordingly, the user can easily recognize the words recommended by the story composition server 100, that is, the connected items.

Next, the sentence composition module 120 can sequentially store the generated sentence data. Then, when the composition of the story according to the topic, that is, topic information, is completed, that is, when multiple sentence data of the story are generated, a new story can be created by combining them (S40).

The generated story can be stored in the database 130 and used as learning data for the sentence construction module 120. Additionally, the story may be transmitted to the user terminal 200 and displayed to the user.

As described above, the story composition method according to this embodiment generates example sentences from a plurality of word data transmitted from the user terminal 200, and corrects the example sentences based on the likeability index between the characters in the example sentences. The above sentence data can be generated. And, by combining these sentence data, a new story can be constructed according to the story topic set by the user.

In other words, the present invention can generate sentence data with various meanings according to the favorability index between people using only a plurality of word data input through the user terminal 200. Accordingly, the present invention can easily construct a large number of sentence data reflecting the user's intention, and create a new story by combining them.

Meanwhile, as described above, the setting module 110 can set the favorability index between characters from the contents of the existing story stored in the database 130.

In addition, the setting module 110 can set the affinity index between characters based on the favorability information between characters transmitted together with the character of the story selected by the user terminal 200.

First, referring to FIG. 6, the topic information may be set to [Kongjwi Patjwi] according to the selection of the user terminal 200.

The setting module 110 can set an affinity index of 10 for [Kongjwi]'s [Patjwi] and an affinity index of 0 for [Patjwi]'s [Kongjwi] based on the existing Kongjwi Patjwi story stored in the database 130.

Subsequently, [Kongjwi], [Patjwi], [field], [work], and [all day] may be transmitted from the user terminal 200 to the story composition server 100 as word data corresponding to the image data.

The word classification unit 123 of the sentence construction module 120 can generate an example sentence of [working in the field all day] by classifying and sorting a large number of word data according to the six and four principles.

Additionally, the connection item recommendation unit 125 may recommend negative connection items such as [sarcasm] and [sarcasm] based on a preset favorability index.

Accordingly, the sentence composition unit 127 uses the linking item to create two corrective example sentences: [Patjwi was making fun of Kongjwi, who worked in the field all day] and [Patjwi was making fun of Kongjwi, who worked in the field all day]. can be created. In addition, one sentence selected by the user terminal 200 among the generated corrected example sentences can be generated as sentence data.

As such, in this embodiment, the setting module 110 sets the mutual favorability index between [Kongjwi] and [Patjwi] from the existing Kongjwi and Patjwi stories stored in the database 130, so that the sentences generated by the sentence construction unit 127 The meaning of sentence data does not differ significantly from the sentences of existing stories that correspond to it.

On the other hand, referring to Figure 7, the topic information is set to [Kongjwi and Patjwi] according to the selection of the user terminal 200, and the mutual affinity information between [Kongjwi] and [Patjwi] is sent from the user terminal 200 to the story composition server (100). ) can be transmitted.

The setting module 110 can change the favorability index for [Kongjwi] from the existing favorability index to 80 based on the favorability information transmitted from the user terminal 200 and set it.

Subsequently, [Kongjwi], [Patjwi], [field], [work], and [all day] may be transmitted from the user terminal 200 to the story composition server 100 as word data corresponding to the image data. In addition, the word classification unit 123 of the sentence construction module 120 can classify and sort a large number of word data according to the six and four principles to generate an example sentence of [Kongjwi and Patjwi work in the field all day].

The connection item recommendation unit 125 can recommend positive connection items, such as [help], [pity], etc., which have a completely different meaning from the one previously described in FIG. 6, based on the changed favorability index.

Accordingly, the sentence construction unit 127 uses the connection item to generate two corrected example sentences, [Patjwi helped Kongjwi, who worked in the field all day] and [Patjwi felt sorry for Kongjwi, who worked in the field all day]. You can. In addition, one sentence selected by the user terminal 200 among the generated corrected example sentences can be generated as sentence data.

As such, in this embodiment, the setting module 110 changes and sets the mutual liking index between [Kongjwi] and [Patjwi] based on the liking information transmitted from the user terminal 200, thereby sending the message to the sentence composition unit 127. The sentence data generated is significantly different in meaning from the sentences in the existing story that correspond to it.

In other words, the present invention can change the affinity index between the characters of the story according to the user's intention, thereby creating a new story that reflects the user's intention compared to the existing story.

The story composition method of the present invention described above can be implemented as a computer-readable recording medium storing a program for performing it, or as a program or application stored and driven on the recording medium. Accordingly, the story composition method of the present invention is mounted on the story composition server 100 or the user terminal 200 in the form of a program or application to perform this, and when run, one or more devices are used to compose a new story. Operations such as generating sentence data and a story using it are performed.

Above, an embodiment of the present invention has been described, but those skilled in the art can add, change, delete or add components without departing from the spirit of the present invention as set forth in the patent claims. The present invention may be modified and changed in various ways, and this will also be included within the scope of rights of the present invention.

In addition, the above-described terms are terms defined in consideration of functions in the present invention, and may vary depending on the intention or custom of the user or operator. Therefore, definitions of these terms should be made based on the content throughout this specification.

100: Story composition server 110: Setting module
111: Topic and character setting section 113: Likeability setting section
120: Sentence composition module 121: Image extraction unit
123: Word classification unit 125: Connection item recommendation unit
127: Sentence structure 130: Database
200: User terminal 300: Wired/wireless network

Claims (10)

In the artificial intelligence-based story composition method,
A step where the user terminal transmits story composition information including the topic, characters, and situation of the story to the story composition server,
The story composition server extracts one or more contextual image data corresponding to the topic of the story based on the story composition information and transmits it to the user terminal;
The user terminal transmitting one or more word data corresponding to the image data to the story composition server,
The story composition server generating a plurality of story topics and characters based on the story composition information and transmitting information about the plurality of story topics and characters to the user terminal;
Setting the selected topic and characters received from the user terminal by the story composition server as topic information and character information;
Setting a favorability index between characters based on the topic information and character information set by the story composition server;
The story composition server classifies one or more word data received from the user terminal according to the hexadecimal principle and sorts the classified word data to generate one or more example sentences;
A step where the story composition server generates a connection item with a positive meaning or a connection item with a negative meaning based on the size of the favorability index;
The story composition server generates a corrected example sentence by correcting the example sentence using one or more extracted connection items, and transmitting the generated corrected example sentence to the user terminal, and
The story composition server generating one sentence selected by the user terminal among the correction example sentences as sentence data.
Includes,
The step of setting the favorability index is,
A story composition method in which the story composition server sets a favorability index between characters according to the mutual favorability of character information based on story data corresponding to topic information among the story data stored in the database.


delete delete delete delete delete delete delete delete delete