JP7539685B2 - Method for presenting a story development to a user, a story development presentation device, a computer program thereof, a method for analyzing a story development, a story development analysis device, and a computer program thereof - Google Patents

Description

Article 30, paragraph 2 of the Patent Act applies March 27, 2020, https://sig-cc.org/? p=164 (Published online in the 3rd Comic Engineering Research Society Proceedings)

The present disclosure relates to a method for presenting a story development to a user, a story development presentation device, a computer program thereof, a method for analyzing a story development, a story development analysis device, and a computer program thereof.

Patent document 1 discloses a method for searching for scenes that meet certain conditions from among multiple scenes that combine multiple media in space and time, using temporal changes in information on the appearance time relationship of various media such as text, images, figures, and audio.

Japanese Patent Application Publication No. 8-137900

There are times when it is desirable to understand and analyze the story development of content such as manga, novels, and movies. To understand the story development, a written description that outlines the story itself, i.e. a "plot," is sometimes used.

However, a written explanation that outlines the story itself can sometimes lack an objective grasp of the story's development. Furthermore, carelessly presenting a text that explains the story can lead to spoilers for those who have not read or watched the film.

Therefore, it is desirable to be able to understand or analyze the development of a story without relying on text that explains the story itself.

The inventors came up with the idea that the change over time in the frequency of appearance of objects (such as characters) that appear in a story is a fundamental factor in the development of the story. Therefore, by using the change over time in the frequency of appearance of objects, it becomes possible to understand or analyze the development of a story without relying on text that explains the story itself. The present disclosure is based on this idea.

One aspect of the present disclosure is a method for presenting the development of a story to a user. The disclosed method is a method for presenting the development of a story in which a plurality of objects appear to a user, comprising: a processor reading time series data from a database in which the time series data indicating a change over time in the appearance of each of the plurality of objects is stored; the processor generating a graphical image indicating a change over time in the appearance of one or more objects included in the plurality of objects based on the read time series data; and the processor presenting the graphical image to the user as the development of the story.

Another aspect of the present disclosure is a story development presentation device. The disclosed device includes a database storing time-series data indicating the change over time in the appearance of each of a plurality of objects appearing in a story, and a processor configured to execute a process of generating a graphical image indicating the change over time in the appearance of one or more objects included in the plurality of objects based on the time-series data read from the database, and presenting the graphical image to a user as the development of the story.

Another aspect of the present disclosure is a computer program. The disclosed computer program causes a computer to execute a process of generating a graphical image showing a change over time in the appearance frequency of one or more objects included in a plurality of objects that appear in a story, based on time-series data showing a change over time in the appearance frequency of each of the plurality of objects that appear in the story read from a database storing the time-series data, and presenting the graphical image to a user as the story progresses.

Another aspect of the present disclosure is a method for analyzing the development of a story in which multiple objects appear. The disclosed method includes a processor reading time series data from a database in which the time series data indicates a change in the appearance frequency of each of the multiple objects over time and is non-negative matrix data, and the processor applying a non-negative matrix factorization algorithm to the time series data to analyze the story.

Another aspect of the present disclosure is a story development analysis device. The disclosed device includes a database that stores time-series data that indicates a change over time in the appearance frequency of each of a plurality of objects that appear in a story and is non-negative matrix data, and a processor configured to execute a process of applying a non-negative matrix factorization algorithm to the time-series data read from the database.

Another aspect of the present disclosure is a computer program. The disclosed computer program causes a computer to execute a process of applying a non-negative matrix factorization algorithm to time-series data read from a database storing time-series data that indicates a change over time in the appearance frequency of each of a plurality of objects that appear in a story and that is non-negative matrix data.

Further details are described in the embodiments below.

FIG. 1 is a schematic diagram of a system including a story development presentation device according to the first embodiment. FIG. 2 is a data structure diagram of time-series data showing the change over time in the appearance rate of each of a plurality of objects. FIG. 3 is a flowchart of the story development presentation process. FIG. 4 is a schematic diagram showing an example of a screen displayed on a user terminal. FIG. 5 is a schematic diagram showing another example of a screen displayed on the user terminal. FIG. 6 is a schematic diagram showing yet another example of a screen displayed on a user terminal. FIG. 7 is a data structure diagram of time-series data according to the second embodiment. FIG. 8 is a data structure diagram of the first non-negative matrix data (basis matrix). FIG. 9 is a data structure diagram of the second non-negative matrix data (activation matrix). FIG. 10 is a schematic diagram of a system including a story development analysis device according to the third embodiment. FIG. 11 is a flowchart of the story development analysis process according to the third embodiment. FIG. 12 is a schematic diagram showing an example of a screen displayed on a user terminal. FIG. 13 is a schematic diagram showing an example of a screen displayed on a user terminal.

<1. Overview of a method for presenting a story development to a user, a story development presentation device, a computer program for presenting a story development, a method for analyzing a story development, a story development analysis device, and a computer program for analyzing a story development>

(1) A method according to an embodiment is a method for presenting to a user the development of a story in which multiple objects appear. The method includes a processor reading out time series data from a database in which the time series data indicating the change over time in the appearance of each of the multiple objects is stored. The time series data indicating the change over time in the appearance of each of the multiple objects indicates a story development. The method includes the processor generating a graphical image indicating the change over time in the appearance of one or more objects included in the multiple objects based on the read time series data, and presenting the graphical image to the user as the development of the story. The appearance of objects is presented to the user by a graphical image, allowing the user to easily grasp the development of the story.

(2) It is preferable that the graphical images are presented to the user without accompanying text explaining the story. By showing the change in the frequency of appearance of objects over time without accompanying text explaining the story, it is possible to prevent spoilers while indicating the development of the story.

(3) Preferably, the method according to the embodiment further comprises the processor selecting the one or more objects from the plurality of objects. In this case, even if the number of objects included in the time-series data is huge, only the temporal change in the appearance frequency of the selected object is presented as a graphical image. Therefore, even if the number of objects is huge, a decrease in visibility can be prevented.

(4) The one or more objects may be selected based on data input by the user. In this case, the user can select the objects at will.

(5) The one or more objects may be selected based on the results of analyzing the time series data. In this case, the objects are selected according to the content of the time series data.

(6) The one or more objects may be selected based on the appearance frequency indicated by the time series data. In this case, the object is selected according to the appearance frequency indicated by the time series data.

(7) The time series data may be non-negative matrix data. It is preferable that the one or more objects are selected based on at least one of two non-negative matrices obtained from the time series data by a non-negative matrix factorization algorithm. In this case, the objects are selected based on the analysis results by the non-negative matrix factorization algorithm.

(8) Preferably, the method according to the embodiment further comprises the processor selecting a partial time range of the entire time range of the time series data. The graphical image may be generated based on partial time series data in the partial time range of the time series data. In this case, even for a long story, a graphical image of the partial time series data in the selected partial time range is shown to the user, thereby preventing a decrease in visibility.

(9) The partial time range may be selected based on data input by the user. In this case, the user can arbitrarily select the partial time range.

(10) The partial time range may be selected based on the results of analyzing the time series data. In this case, the partial time range is selected according to the content of the time series data.

(11) The time series data may be non-negative matrix data. It is preferable that the partial time range is selected based on at least one of two non-negative matrices obtained from the time series data by a non-negative matrix factorization algorithm. In this case, the partial time range is selected based on the analysis results by the non-negative matrix factorization algorithm.

(12) The method according to the embodiment may further include the processor selecting the one or more objects from the plurality of objects, and the processor selecting a partial time range of a total time range of the time series data. The graphical image is preferably generated to indicate a temporal change in the appearance of the one or more objects in the partial time range.

(13) A story development presentation device according to an embodiment includes a database storing time-series data indicating a change over time in the appearance of each of a plurality of objects appearing in a story, and a processor configured to execute a process of generating a graphical image indicating a change over time in the appearance of one or more objects included in the plurality of objects based on the time-series data read from the database, and presenting the graphical image to a user as the development of the story.

(14) A computer program according to an embodiment causes a computer to execute a process of generating a graphical image showing a change over time in the appearance of one or more objects included in a plurality of objects based on time series data showing a change over time in the appearance of each of the plurality of objects appearing in a story read from a database storing the time series data, and presenting the graphical image to a user as the story progresses.

(15) A method according to an embodiment is a method for analyzing the development of a story in which multiple objects appear. The method includes a processor reading time-series data from a database in which time-series data indicating a temporal change in the appearance frequency of each of the multiple objects and being non-negative matrix data is stored, and the processor applying a non-negative matrix factorization algorithm to the time-series data in order to analyze the story. In this case, a novel method for analyzing the story development is obtained, that is, an analysis method using a non-negative matrix factorization algorithm. As described below, the present inventors have found that an analysis method using a non-negative matrix factorization algorithm is useful for analyzing the story development.

(16) A story development analysis device according to an embodiment includes a database storing time-series data that indicates a change over time in the appearance frequency of each of a number of objects that appear in a story and that is non-negative matrix data, and a processor configured to execute a process of applying a non-negative matrix factorization algorithm to the time-series data read from the database.

(17) A computer program according to an embodiment causes a computer to execute a process of applying a non-negative matrix factorization algorithm to time-series data read from a database storing time-series data that indicates a change over time in the appearance frequency of each of a number of objects that appear in a story and that is non-negative matrix data.

<2. Examples of a method for presenting a story development to a user, a story development presentation device, a computer program for presenting a story development, a method for analyzing a story development, a story development analysis device, and a computer program for analyzing a story development>

<2.1 First embodiment>

FIG. 1 shows a system 10 according to the first embodiment. The system 10 according to the first embodiment includes a story development presentation device 100. The story development presentation device 100 presents data showing the story development of content to a user via a network 400. The content is, for example, a manga, a novel, a movie, a television program, etc. The content is not limited to manga, etc., and any content with a story will suffice. Another example of content with a story is conversation data in a chat app. The flow of conversation in a chat app has a story, similar to conversations in a manga or novel.

In the first embodiment, the user is, for example, a reader of manga or novels, or a viewer of movies or television programs. The user can access the story development presentation device 100 using a user terminal 300. The user terminal 300 is, for example, a smartphone, a tablet, a desktop computer, or a laptop computer. The user terminal 300 includes a processor 310, a storage device 320, and an input/output device 330.

The system 10 includes a user terminal 300 connected to the story development presentation device 100 via a network 400. The network 400 is, for example, the Internet. The story development presentation device 100 is, for example, a server installed on the Internet.

The story development presentation device 100 is configured by a computer having a processor 110 and a storage device 120. The processor 110 is, for example, a CPU. The storage device 120 has, for example, a primary storage device and a secondary storage device. The primary storage device is, for example, a RAM. The secondary storage device is, for example, a hard disk drive (HDD) or a solid state drive (SDD).

The storage device 120 stores a computer program 121 for operating a computer as the story development presentation device 100. The computer program 121 comprises program code for causing the processor 110 to execute a story development presentation process 111. The story development presentation process 111 is a process that includes presenting data indicating the story development of the content to the user via the network 400. The story development presentation process 111 will be described later.

The story development presentation device 100 includes a database 200. The database 200 may be stored in the storage device 120, or may be stored in another computer (e.g., a database server on the Internet) connected to the computer including the processor 110 and the storage device 120.

The database 200 includes multiple content data 201, 202, and 203. If the content is a manga, the content data includes image data of the manga and work information such as the title of the work. If the content is a novel, the content data includes text data of the novel and work information such as the title of the work. If the content is a movie or television program, the content data includes video and audio data and work information such as the title of the work. In an embodiment, the content may be digital content, or may be content on paper or other media.

In FIG. 1, first content data 201 of the first content (title: xxx), second content data 202 of the second content (title: yyy), and third content data 203 of the third content (title: zzz) are shown.

The content data 201, 202, and 203 according to the embodiment further include time series data 201M, 202M, and 203M, respectively. Each of the time series data 201M, 202M, and 203M indicates the change over time in the appearance frequency of each of the multiple objects appearing in the content. For example, the first content data 201 includes first time series data 201M indicating the change over time in the appearance frequency of each of the multiple objects appearing in the first content. The second content data 202 includes second time series data 202M indicating the change over time in the appearance frequency of each of the multiple objects appearing in the second content. The third content data 203 includes third time series data 203M indicating the change over time in the appearance frequency of each of the multiple objects appearing in the third content.

An object is, for example, a person (character), an animal, another object, a concept, a catchphrase, another word, a sound, a landscape, or a background that appears in the content. An object may be an object represented as a diagram or an object in a video, or a word in a sentence, or an object or concept represented by a conversation, etc.

The appearance frequency indicates the degree to which an object appears in the content. The appearance frequency may be the number of times an object appears, or the percentage of appearance of an object. The appearance frequency may simply indicate whether or not an object appears.

"Time" in content need only be an indicator of the progress of the content's story, and does not have to be physical time. For example, if the content is a manga or novel, the page number, chapter number, and volume number are each indicators of the progress of the story, and can be treated as "time" in the content. If the content is a movie or television program, the elapsed time from the start of the content can be treated as "time" in the content. Also, if the story of the content unfolds while indicating a specific date and time, that date, etc. can be treated as "time" in the content.

The frequency of appearance is, for example, the number of times an item appears, the percentage of appearance, or whether or not it appears per unit period in the story flow of the content. The unit period may be, for example, one page, one episode, one volume, or a unit length based on the form of the medium in which the content is published, or it may be a unit time length such as one minute, ten minutes, or one hour.

The number of appearances in a unit period indicates how many times a particular object appeared in the unit period. The frequency of appearance in a unit period is the ratio of the number of times a particular object appeared in the unit period to the total number of times all objects appeared in the unit period. The presence or absence of appearance in a unit period indicates whether or not a particular object appeared in the unit period.

Figure 2 shows an example of the data structure of the first time series data 201M of the first content. The other time series data 202M, 203M have a similar data structure, although they are not shown. Here, the first content is a manga.

The first time series data 201M comprises time information T and object information S. The time information T is an index showing the progress (progress position) of the story of the first content, and is represented by a combination of a page number, an episode number, and a volume number. That is, the time information T comprises a page number T1, which is the first time information, an episode number T2, which is the second time information, and a volume number T3, which is the third time information. Here, the unit period for appearance is one page, which is the smallest time unit in the time information T. In FIG. 2, time progresses from t1 to tx.

The object information S indicates the appearance of each of the multiple objects. The object information S includes the appearance of all objects that appear in the first content. In FIG. 2, the objects shown are character A, character B, character C, character D, building E, building F, tool G, and tool H.

The appearance degree is associated with each unit period (each page). For example, the first line P1 in FIG. 2 shows the appearance degree (number of appearances) of each object in the unit period of Volume 1, Chapter 1, and Page 1 of the first content. Specifically, the first line p1 in FIG. 2 shows that in Volume 1, Chapter 1, and Page 1, Character A appeared twice, Character B once, Character C once, Character D zero times, Building E zero times, Building F zero times, Tool G zero times, and Tool H one time. In this way, in the first line P1 in FIG. 2, the appearance degree of each object in the unit period is associated with the unit period of Volume 1, Chapter 1, and Page 1 of the first content. Similarly, the appearance degree of each object in the other unit periods is associated with each other.

Since the first time series data 201M is configured as described above, it indicates the change over time in the appearance frequency of each of multiple objects A, B, C, D, E, F, G, and H that appear in the first content. Such time series data 201M, 202M, and 203M not only indicate the change over time in the appearance frequency of the objects, but also quantitatively represent the development of the story. Therefore, the time series data 201M, 202M, and 203M are useful for understanding and analyzing the development of the story.

Figure 3 shows an example of a story development presentation process 111 using time series data. The story development presentation process 111 is executed by the processor 110 of the story development presentation device 100. The story development presentation device 100 according to the embodiment generates a graphical image indicating a story development from time series data read in response to a query input by a user. The story development presentation device 100 presents the graphical image to the user as information indicating the story development of the content.

By presenting the time series data to the user as graphical images, the user can easily recognize the story development, which is indicated by the temporal change in the appearance frequency of objects, through the graphical images. The presentation of the story development through graphical images allows for a more visual and intuitive understanding than the presentation of the story development through text or numerical values.

In the story development presentation process 111 according to the first embodiment, first, the processor 110 of the story development presentation device 100 executes a search process (step S111). The search process is a process for searching for content that the user is looking for. In the search process, the story development presentation device 100 accepts a content search request from the user terminal 300 and executes a search in the database 200. Hereinafter, the searched content may be referred to as the target content.

The content search request includes, for example, a query for the search. The query includes search data used to search for the content. The search data is data for identifying the content, for example, data indicating the title of the work. The query may be a document written in a natural language.

The processor 110 obtains specific content data that matches the search data as a search result in the database 200. That is, the processor 110 reads out the content data of the target content from the database 200. The read out content data includes the time series data described above. The content data including the time series data is stored in the storage device 120 (step S112).

The processor 110 generates image data (graphical image) that graphically represents the time series data from the stored time series data (step S115). In the embodiment, the graphical image is image data of a graph that shows the change in appearance over time (see FIG. 4).

The time series data stored in database 200 includes data showing the change over time in the appearance of all objects in the corresponding content. However, from the standpoint of user visibility or ease of understanding of the story development, it may be better to show the change over time in the appearance of a smaller number of objects rather than showing the change over time in the appearance of all objects in the target content.

The processor 110 according to the embodiment selects objects to be used in generating the graphical image in order to narrow down the objects shown in the graphical image (step S113). The objects selected in step S113 are called target objects. In the embodiment, the graphical image is generated as image data of a graph showing the change over time in the appearance frequency of the target object. Note that all objects may be selected as target objects.

The selection of the target object may be automatic selection in which the processor 110 executes a predetermined selection algorithm, or manual selection in which the target object is selected by user input from the terminal 300. Automatic selection reduces the burden on the user, while manual selection allows the user to make a free selection according to their interests. The result of the search process in step S111 or the search data may be used for the automatic selection of the target object.

The automatic selection of the target object can be based on the results of analyzing the time series data of the target content. The analysis of the time series data is, for example, an analysis of the appearance frequency of each object. The processor 110 can analyze the appearance frequency of each object and select the target object based on the magnitude of the appearance frequency of the object or the degree of change in the appearance frequency. The analysis of the time series data may be performed by non-negative matrix factorization, as described in the second embodiment below.

The selection criteria for the target object are, for example, that the appearance frequency of the object is greater than a threshold value, or that the object's appearance frequency is within a predetermined top number (for example, within the top three). An object with a high appearance frequency can be said to be a main object in the target content. By showing the change over time in the appearance frequency of the main object, it is possible to show the development of a story centered around the main object.

Other selection criteria for target objects are that the appearance rate of the object is less than a threshold value, or that the magnitude of the appearance rate of the object is within a certain lower number (for example, within the bottom three). Since an object with a low appearance rate may be a key object that occupies an important position in the development of the story, it is also useful to show the change in the appearance rate of an object with a low appearance rate over time.

The time series data stored in database 200 also includes data showing the change over time in appearance over the entire time range of the corresponding content (from the start to the end of the content). However, from the perspective of user visibility or ease of understanding of the story development, it may be better to show the change over time in appearance over a shorter time range (a part of the entire time range) rather than showing the change over time in appearance over the entire time range.

The processor 110 according to the embodiment selects a time range to be used in generating the graphical image in order to narrow down the time range shown in the graphical image to a partial time range (step S114). The time range selected in step S114 is called the target time range. Data showing the temporal change in appearance in the target time range in the time series data is called partial time series data. In the embodiment, the graphical image is generated from the partial time series data as image data of a graph showing the temporal change in appearance of the object in the target time range. Note that the entire time range of the target content may be selected as the target time range. The entire time range indicates, for example, all volumes, all chapters, and all pages of the content. The partial time range that is the target time range is, for example, one or more specific volume numbers, one or more chapter numbers in a specific volume, or one or more specific page numbers.

The selection of the target time range may be automatic selection in which the processor 110 executes a predetermined selection algorithm, or may be manual selection in which the user inputs from the terminal 300. Automatic selection reduces the burden on the user, while manual selection allows the user to freely select according to their interests. The results of the search process in step S111 or the search data may be used for the automatic selection of the target time range.

The automatic selection of the target time range can be based on the results of analyzing the time series data of the target content. The analysis of the time series data is, for example, an analysis of the appearance degree of each object. The processor 110 can analyze the appearance degree of each object and select the target time range based on the magnitude of the appearance degree of the object or the degree of change in the appearance degree. The object to be analyzed may be the target object, or may include objects other than the target object. The analysis of the time series data may be performed by non-negative matrix factorization, as described in the second embodiment described later.

The target time range is selected, for example, as a time range including a time when the target object appears more frequently than a threshold value. Also, the target time range is selected as a time range including a time when the degree of change in the appearance rate of the target object becomes greater than a threshold value. By selecting a time range in which the target object appears more frequently, an appropriate story development is presented. Also, since the timing when the appearance rate of the target object changes (increases or decreases) can be an important timing in the story development, selecting a time range in which the degree of change in the appearance rate of the target object is large can result in an appropriate story development presentation.

In FIG. 3, after the time series data is read in step S112, the target object is selected (step S113) and the target time range is selected (step S114), but the reverse may be done. In other words, after the target object and the target time range are selected, only data indicating the temporal change in the appearance frequency of the target object in the target time range may be read out from the time series data stored in database 200.

In step S115, the processor 110 generates image data (graphical image) of a graph showing the temporal change in the appearance frequency of the target object in the target time range.

In step S116, the processor 110 transmits image data constituting the graphical image together with other necessary data to the user terminal 300. The user terminal 300 receives data including the image data for the graphical image. The user terminal 300 displays the graphical image, etc. on a display (input/output device 330) provided in the user terminal 300 as data showing the story development of the content to be presented (step S131).

Then, the processor 110 can accept user operations on the user terminal 300 and update the graphical images, etc. displayed on the user terminal 300 at any time (step S117).

Figure 4 shows an example of a screen 500 displayed on the display of the user terminal 300. The screen 500 is a graphical user interface (GUI) screen. On the screen 500, a graphical image 512 of time series data and the like are displayed, and user operations are accepted.

The screen 500 has a display area 501 for data received from the story development presentation device 100. The display area 501 has a graphical image 512 of the time series data. The graphical image 512 shows the story development in terms of changes in the appearance frequency of objects in the target content.

In addition to the graphical image 512, the display area 501 has a work name display section 510, a target object name display section 511, and a seek bar 516.

The work name display section 510 is an area for displaying the work name of the target content. The name display section 511 is an area for displaying the name of the target object in the target content. The name of the target object is, for example, the character name. FIG. 4 shows a case where three objects, character A, character B, and character C, have been selected as target objects. In the display area 501 shown in FIG. 4, character A is given the reference symbol "O-1", character B is given the reference symbol "O-2", and character C is given the reference symbol "O-3".

The graphical image 512 shown in FIG. 4 is configured as a graph showing the temporal variation of the appearance frequency of the target object in the target time range. The horizontal axis of the graph, which is the graphical image 512, shows time, and the vertical axis shows the appearance frequency (appearance ratio) of each target object. Here, multiple (three) target objects have been selected, and the target time range is from episode 1 to episode 20.

The graphical image 512 shown in FIG. 4 shows the percentage of appearances of each target object relative to the total number of appearances of all selected target objects in each time range (range of one episode). In other words, it shows the percentage of appearances of each target object when the total number of appearances of the three target objects in each time range (range of one episode) is 100. For example, in episode 9, out of the three characters A, B, and C, the appearance percentage of character A (O-1) was 53.77%, the appearance percentage of character B (O-2) was 46.23%, and the appearance percentage of character C (O-3) was 0%.

In the graphical image 512 shown in FIG. 4, the change over time in the appearance ratio of each target object is shown by the change over time in the area each target object occupies on the graph. This makes it easy to visually and intuitively understand the change over time in the appearance ratio. Moreover, in the graphical image 512 shown in FIG. 4, the change over time in the appearance ratio is shown only for the multiple selected target objects, so that the change in the story development due to the multiple selected target objects can be visually and intuitively grasped.

The story development presentation device 100 according to the embodiment allows the user to visually and intuitively grasp changes in the story development, making it possible to search for a specific point (a specific scene) in the story without looking at the content directly. For example, even if the user cannot remember where a specific scene is in which Character A and Character B appear together, the user can refer to the graphical image 512 to find such a scene from a point where both Character A and Character B make a high appearance.

The screen 500 shown in FIG. 4 presents the user with a graphical image 512 of the appearance rate without accompanying text (e.g., a synopsis) explaining the story of the target content. Therefore, it is possible to present the user with the development of the story while preventing spoilers from being revealed to the user.

For example, a user may use the user terminal 300 to read chapter 18 of content viewed online (e.g., a manga that can be read online), and mistakenly skip chapter 19 and try to read chapter 20. If the server providing the content detects the possibility of such an error, the story development presentation device 100 can display on the user terminal 300 a graphical image 512 showing the change over time in the frequency of appearance of each object in chapter 19 as a warning of the possibility of an error.

If a user who sees graphical image 512 has already read episode 19, the user can easily recall the story of episode 19 by using the information indicated by graphical image 512 (the proportion of each object that appears, and its change over time) as a hint. Conversely, if the user has not read episode 19, even if the user sees the information indicated by graphical image 512, he or she will not recall the story of episode 19. In this way, by using graphical image 512, the user can confirm whether a specific part of the content has already been read or viewed.

If a method of checking the synopsis of episode 19 were adopted to check whether episode 19 has already been read, the user would find out the synopsis of episode 19 in advance even though the user has not read it, resulting in so-called spoilers. However, according to this embodiment, it is possible to check whether a specific part of the content has been read or viewed while preventing spoilers.

In the above case, the server providing the content may itself have the function of the story development presentation device 100, or the server providing the content and the story development presentation device 100 may be connected via a network. When the story development presentation device 100 detects the need to check whether the content has been read or viewed, it transmits a graphical image 512 of the part requiring confirmation to the user terminal 300.

In addition, in an embodiment, in order to allow a more accurate understanding of the appearance ratio at each point in time on the horizontal axis of the graph, the screen 500 can display the appearance ratio of each target object as a numerical value. To display the appearance ratio numerically, the screen 500 includes a seek bar 516 and a selection line R. The seek bar 516 includes a slider 516A that can be moved along the horizontal axis direction of the graph.

The user can use the GUI function to move slider 516A to select each point in time (episode number) on the horizontal axis of the graph. In FIG. 4, episode 9 is selected. Selection line R moves in conjunction with slider 516A. The user can also move the selection line using the GUI function. While seek bar 516 is located outside the graph, selection line R is located on the graph, making it possible to more clearly indicate which episode number is selected.

Screen 500 can display display sections 513, 514, and 515 that indicate the appearance frequency (appearance ratio) at the time (episode number) selected. Display sections 513, 514, and 515 can have the shape of a speech bubble, for example. FIG. 4 shows a first display section 513, a second display section 514, and a third display section 515 that indicate the appearance ratio in the selected 9th episode. First display section 513 shows the appearance ratio of character A (O-1) as a number (53.77%). Second display section 514 shows the appearance ratio of character B (O-2) as a number (46.23%). Third display section 515 shows the appearance ratio of character C (O-3) as a number (0%). This shows that in volume 9 of the target content, characters A and B appear frequently, and character C does not appear.

Figure 5 shows the display area 501 of the screen 500 after a user operation has been performed to move the slider 516A to the position for episode 13. In episode 13, the appearance rate of character A shown by the first display unit 513 is 58.77%, the appearance rate of character B shown by the second display unit 514 is 23.16%, and the appearance rate of character C shown by the third display unit 515 is 21.35%. It can be seen that in episode 13, in addition to characters A and B, character C appears. Also, according to the graphical image 512 shown in Figure 5, character C did not appear until episode 12, but began to appear from episode 13.

In this way, by presenting the user with image data showing the change in appearance over time, it is possible to present the user with the episode number (time of appearance) in which the target object appears, without showing the story itself. Also, even if the time-series data contains a huge number of objects, only the change in appearance over time of the selected target object is presented as image data, improving visibility.

Figure 6 shows another example of the display area 501 on the screen 500. Figure 6 shows a case where an object other than a character has been selected as the target object. Figure 6 shows a case where three objects, a character D, a building E, and a tool G, have been selected as the target objects. In the display area 501 shown in Figure 6, the character D is given the reference symbol "O-1", the building E is given the reference symbol "O-2", and the tool G is given the reference symbol "O-3".

In this way, according to the embodiment, the time series data of the appearance frequency of a target object that is automatically selected or arbitrarily selected by the user can be displayed as a graphical image 512, making it easy to present the development of a story from the perspective of an arbitrary object.

<2.2 Second embodiment>

In the second embodiment, at least one of the selection of the target object (step S113 in FIG. 3) and the selection of the target time range (step S114 in FIG. 3) is performed based on the result of applying the non-negative matrix factorization (NMF) algorithm to the time-series data 201M of the target content. In other words, in the second embodiment, the analysis of the time-series data 201M in the first embodiment is performed by NMF. Note that the points not specifically described in the second embodiment are the same as those in the first embodiment.

Figure 7 shows the data structure of time-series data 201M used in the story development presentation process 111 according to the second embodiment. In the second embodiment, the time-series data 201M is treated as matrix data.

In the time series data 201M shown in FIG. 7, each of the multiple rows indicates the change in the appearance frequency of each object over time, and each of the multiple columns indicates "time" in the content. In FIG. 7, the objects include character A, character B, character C, character D, building E, building F, tool G, and tool H, as in the first embodiment. For simplicity, time is shown as an integer of 1 or more, but as with the time information T in the first embodiment, it may be represented by a combination of page number, episode number, and volume number. In the following, the numbers indicating time represent the episode number.

For example, for character A, the appearance degree in episode 1 is represented by m _A1 , the appearance degree in episode 2 is represented by m _A2 , the appearance degree in episode 3 is represented by m _A3 , the appearance degree in episode 4 is represented by m _A4 , the appearance degree in episode 5 is represented by m _A5 , and the appearance degree in episode tx is represented by m _Ax . This represents the change in the appearance degree of character A over time. The change in the appearance degree of character A over time is represented as an x-dimensional vector. Similarly, the change in the appearance degree of other objects over time is represented.

In the second embodiment, the time series data 201M is treated as matrix data having a size of the number of objects x the number of times (1 to tx). The matrix elements in the matrix data indicate the appearance frequency of the above-mentioned m _A1 and the like.

The time series data 201M treated as matrix data is created as non-negative matrix data in which all matrix elements (such as appearance frequency m _A1 ) are non-negative. The time series data 201M being non-negative matrix data makes it possible to apply the non-negative matrix factorization (NMF) algorithm.

The NMF algorithm can obtain two nonnegative matrices, a first nonnegative matrix W and a second nonnegative matrix H, from a given nonnegative matrix X. In other words, the NMF algorithm is an algorithm for approximating a given nonnegative matrix X by the product of two nonnegative matrices, the first nonnegative matrix W and the second nonnegative matrix H.

When a given non-negative matrix X is an N×M matrix, the first non-negative matrix W is an N×r matrix, and the second non-negative matrix H is an r×M matrix. r is a value given in advance and is preferably an integer equal to or greater than 2. The first non-negative matrix W is a basis matrix having r basis vectors. The second non-negative matrix H is an activation matrix indicating the weight of each basis vector.

Figure 8 shows an example of first non-negative matrix data W (basis matrix) obtained by applying the NMF algorithm to time series data 201M, which is non-negative matrix data. Here, r = 3. Therefore, the first non-negative matrix data W shown in Figure 8 has three basis vectors F1, F2, and F3. Here, the multiple basis vectors F1, F3, and F3 are called topics.

The inventors have discovered that by applying NMF to time-series data that indicates the temporal change in the appearance frequency of each of multiple objects that appear in a story, it is possible to decompose and extract multiple topics F1, F2, and F3 that are important in the development of a story. The extracted topics F1, F2, and F3 are important factors that form the basis of the development of the story of the content. By applying NMF to the time-series data 201M, it is possible to easily obtain story development analysis results that are not easily obtained by humans directly reading or watching the content.

The multiple topics F1, F2, and F3 can also be interpreted as the main story and sub-stories that make up a story. This applies when multiple stories are mixed together within a single piece of content. For example, in a manga, one or more episodes may contain complete one-off stories, but a central storyline that runs through the entire content may be set. In such cases, the story progresses with a mixture of one-off stories and central stories within a single piece of content. By using NMF, multiple stories mixed together within a single piece of content can be extracted as topics F1, F2, and F3.

The first topic F1 (basis vector) shown in Figure 8 indicates the appearance frequency of each object in the first topic F1, with character B having the highest appearance frequency among all objects. Therefore, it can be seen that the first topic F1 is a topic that mainly focuses on character B.

The second topic F2 (basis vector) indicates the appearance frequency of each object in the second topic F2, with character A having the highest appearance frequency among all objects. Therefore, it can be seen that the second topic F2 is a topic that mainly focuses on character A.

The third topic F3 (basis vector) indicates the appearance frequency of each object in the third topic F3, with character C appearing the most among all objects, and character B also appearing to some extent. Therefore, it can be seen that the third topic is centered on characters B and C, with character C being more important than character B.

Figure 9 shows an example of second non-negative matrix data H (activation matrix) obtained by applying the NMF algorithm to time-series data 201M, which is non-negative matrix data. Here, since r = 3, the second non-negative matrix data H indicates the weights (activation levels) of the three topics F1, F2, and F3 at each time. In other words, the second non-negative matrix data H (activation matrix) indicates the change over time in the degree of weight each of topics F1, F2, and F3 has at each time (number of episodes), in other words, the change over time in the degree of activation.

In the story development presentation process 111 of the second embodiment, the processor 110 selects a target object using the first non-negative matrix W (step S113 in FIG. 3). For example, the processor 110 automatically selects one of the multiple topics F1, F2, and F3.

The selection of topics F1, F2, and F3 may be manual selection by the user with reference to the NMF results displayed on the user terminal 300. In this case, the NMF results are displayed on the user terminal 300 prior to the user's selection (see the third embodiment described below).

Then, the processor 110 selects, from the first non-negative matrix W, one or more objects that have a high appearance frequency in the selected topic. Note that one or more objects that have a low appearance frequency may also be selected. Through the above process, objects related to the selected topic are automatically selected as target objects.

In addition, in the story development presentation process 111 of the second embodiment, the processor 110 selects the target time range using the second non-negative matrix H (step S114 in FIG. 3). For example, the processor 110 refers to the second non-negative matrix H and selects, as the target time range, a time range that includes a time during which the weight (activation level) of the topic selected as described above is greater than a threshold value. The selection of the target time range may be manual selection by a user operation that refers to the NMF results displayed on the user terminal 300.

The graphical image 512 generated in the second embodiment can show the change over time in the appearance of the target object selected using the NMF results in the target time range selected using the NMF results. In other words, in the second embodiment, the story development in the selected specific topic is presented.

<2.3 Third embodiment>

Figure 10 shows a system 10 according to the third embodiment. In the third embodiment, the points that are not specifically described are the same as those in the first and second embodiments.

The system 10 according to the third embodiment includes a story development analysis device 100A. The story development analysis device 100A analyzes the story development of the content and generates data indicating the story development as the analysis result. The story development analysis device 100A according to the third embodiment analyzes the story development of the content using non-negative matrix factorization (NMF) as an example. The story development analysis device 100A also operates as a story development presentation device that presents the data indicating the story development obtained as the analysis result to the user via the network 400.

In the second embodiment, the analysis results obtained by applying NMF to time series data were used to select a target object or a target time range, but in the third embodiment, the analysis results obtained by applying NMF to time series data are used as data indicating the story development. Story development analysis device 100A generates a graphical image indicating the story development from the analysis results obtained by NMF. Story development analysis device 100A presents the graphical image to the user as information indicating the story development of the content.

In the third embodiment, the user is a person involved in the production of the content, such as a content author or editor. The user may also be a reader of manga or novels, or a viewer of movies or television programs. The story development analysis device 100A presents the results of the story development analysis to the user, thereby supporting the user in coming up with ideas for or changing the story development.

The system 10 includes a user terminal 300 connected to the story development analysis device 100A via a network 400. A user can access the story development analysis device 100A using the user terminal 300. The user terminal 300 includes a processor 310, a storage device 320, and an input/output device 330.

The story development analysis device 100A is configured by a computer including a processor 110 and a storage device 120. The storage device 120 stores a computer program 121A for operating the computer as the story development analysis device 100A (story development presentation device). The computer program 121A includes program code for causing the processor 110 to execute a story development analysis process 111A (story development presentation process). The story development analysis process 111A is a process that includes analyzing the story development of the content. The story development analysis process 111A will be described later.

The story development analysis device 100A includes a database 200. The database 200 includes multiple content data 201, 202, and 203. As in FIG. 1, FIG. 10 shows first content data 201 of a first content (work name: xxx), second content data 202 of a second content (work name: yyy), and third content data 203 of a third content (work name: zzz). The content data 201, 202, and 203 according to the embodiment further include time series data 201M, 202M, and 203M, respectively. Each of the time series data 201M, 202M, and 203M indicates the change over time in the appearance frequency of each of multiple objects that appear in the content.

The data structure of each of the time series data 201M, 202M, and 203M according to the third embodiment is similar to the data structure of the time series data 201M according to the second embodiment (see FIG. 7). Each of the time series data 201M, 202M, and 203M according to the third embodiment is non-negative matrix data, similar to the second embodiment.

Figure 11 shows an example of story development analysis processing 111A using time series data. Story development analysis processing 111A is executed by processor 110 of story development analysis device 100A. Story development analysis device 100A according to the embodiment applies the NMF algorithm to time series data 201M, 202M, and 203M, which are non-negative matrix data, in order to analyze the story development.

In the story development analysis process 111A according to the third embodiment, the processor 110 first executes a search process (step S311). As in the first embodiment, the search process is a process of searching the database 200 for the content (target content) that the user is looking for.

The processor 110 obtains data of specific content that matches the search data as a search result in the database 200. That is, the processor 110 reads data of the target content from the database 200. The read content data includes the time series data described above. The content data including the time series data is stored in the storage device 120 (step S312).

The processor 110 applies the NMF algorithm to the stored time series data (see FIG. 7) (step S313). As a result of the analysis using NMF, the processor 110 obtains the first non-negative matrix data W shown in FIG. 8 and the second non-negative matrix data H shown in FIG. 9. As described above, the first non-negative matrix W is a basis matrix having r basis vectors. The second non-negative matrix H is an activation matrix indicating the weight of each basis vector.

The processor 110 generates image data (graphical image) indicating the story development of the target content from the first non-negative matrix data W and the second non-negative matrix data H (step S314). The processor 110 transmits the image data constituting the graphical image to the user terminal 300 together with other necessary data.

The user terminal 300 receives data including image data for a graphical image. The user terminal 300 displays the graphical image, etc., on a display (input/output device 300) provided in the user terminal 300 as data showing the analysis results of the story development of the content to be presented.

Figure 12 shows an example of a screen 500A displayed on the display of the user terminal 300. Screen 500A is a graphical user interface (GUI) screen. Screen 500A displays the analysis results and accepts user operations.

Screen 500A has a work title display section 510, a topic display section 601, and a story development display section 602.

The topic display section 601 displays one or more topics included in the target content. The topic display section 601 also displays the legend for the graph displayed in the story development display section 602.

The topics displayed in the topic display section 601 are generated from the first non-negative matrix data W. The first non-negative matrix data W shown in FIG. 8 has three basis vectors F1, F2, and F3. Therefore, the three topics F1, F2, and F3 are displayed in the topic display section 601.

The topic display unit 601 shows objects in each of the topics F1, F2, and F3. The topic display unit 601 shown in FIG. 12 shows that the main object in the first topic F1 is character B, the main object in the second topic F2 is character A, and the main objects in the third topic F3 are character B and character C. As the main objects in each of the topics F1, F2, and F3 are displayed in the topic display unit 601, the user can understand the main objects contributing to each of the topics F1, F2, and F3.

The processor 110 determines the main object to be displayed in the topic display unit 601 based on the first non-negative matrix W. The main object is determined based on the values of the matrix elements of the first non-negative matrix W. The values of the matrix elements of the first non-negative matrix W indicate the contribution of the object to each topic. Therefore, the processor 110 can determine, for each basis vector (topic), for example, the object with the highest contribution, or the object with a contribution greater than a threshold, as the main object in that topic.

The story development display section 602 shows the change over time in the activation level (hereinafter also referred to as "activation value") of each of the topics F1, F2, and F3 in the form of a graph, which is a graphical image. In the story development display section 602 of FIG. 10, the horizontal axis shows the number of episodes as an example of time, and the vertical axis shows the activation level. In the story development display section 602 of FIG. 10, the change over time in the activation level of each of the three topics F1, F2, and F3 is shown as image data of a line graph that visualizes the change.

In the story development display section 602 of FIG. 12, the first topic F1, in which character B is the main object, and the second topic F2, in which character A is the main object, show a stable activation level over time. Therefore, a user observing the story development display section 602 can understand that character A and character B are the main objects (so-called protagonists) who always appear in the target content. Furthermore, since topics F1 and F2 related to characters A and B in the target content show a stable activation level over time, it can be seen that characters A and B, who are the protagonists, appear stably in the target content regardless of the ups and downs of the story development.

Furthermore, in the story development display section 602, the third topic F3 shows that it fluctuates greatly over time, resulting in ups and downs. Therefore, a user observing the story development display section 602 can understand that the ups and downs in the story development are created by the interaction between character C (main) and character B (sub-character).

In other words, the user can objectively understand that while characters A and B are the main characters, the key to the story development lies in the intertwining of characters C (main) and B (sub-character). In this way, the user can easily grasp the story development pattern of the target content. Such a story development pattern may not be easily obtained by a human reading or watching the content directly, but the story development pattern can be easily obtained by analysis using the NMF algorithm.

Moreover, because the analysis results of the story development obtained by the NMF algorithm are numerical data, they are useful for examining the similarity of story development between different contents. For example, if the creator wants the content to be produced to resemble the story of another piece of content, or if the creator does not want the content to resemble the story of another piece of content, the creator user can consider the story development of the content to be restricted by referring to the story development of the other piece of content obtained by the story development analysis device 100A.

The story development analysis results obtained by the NMF algorithm are numerical data that quantitatively represent the story development of the content. Therefore, it is possible to search for other content that is similar to a certain piece of content by using the identity or similarity between the analysis results of the content.

The screen 500A may include a display showing the first non-negative matrix data W shown in FIG. 8 and the second non-negative matrix H shown in FIG. 9 as numerical data. In this case, the user can check the details of the analysis results.

Furthermore, the story development analysis device 100A according to the third embodiment also has a function of displaying a value indicating how to adjust the appearance rate of objects (characters) in the target content in order to review or modify the story development. This function allows a user, such as a creator, to simulate what would happen if the story development were changed. In addition, the user can obtain hints for suggesting story developments.

To assist in adjusting the appearance of objects, the processor 110 according to the third embodiment can execute the processes from step S315 onwards in the story development analysis process 111A. Here, the adjusted appearance of objects is obtained by changing the activation value (activation level) of each topic obtained by NMF.

Figure 13 shows a screen 500A that includes a function to assist in adjusting the appearance degree of an object. Similar to Figure 12, this screen 500A includes a work name display section 510, a topic display section 601, and a story development display section 602. Furthermore, the screen 500A shown in Figure 13 includes an object contribution degree display section 701, a story development adjustment section 702, and an adjustment result display section 703.

The object contribution display section 701 is an area that displays numerical data indicating the first non-negative matrix W. By referring to the object contribution display section 701 in addition to the story development display section 602, the user can grasp the analysis results by the NMF algorithm in more detail.

The story development adjustment unit 702 includes a selection unit 702A for selecting a topic to be adjusted, a selection unit 702B for selecting the number of episodes to be adjusted (target time range), and an input unit 702C for inputting an activation value. The topic selection unit 702A selects a topic whose activation value (activation level) is to be adjusted from among multiple topics F1, F2, and F3 obtained by NMF. The topic selection unit 702A displays the extracted topics F1, F2, and F3 selectable using radio buttons or the like.

The processor 110 of the story development analysis device 100A receives data indicating the selected target topic F3 from the user terminal 300. Based on the data received from the user terminal 300, the processor 110 selects, for example, the third topic F3 as the target topic for which the activation value is to be changed in the second non-negative matrix H (step S315).

In the episode number selection unit 702B, the target episode number for which the appearance rate of the object in the target content is to be adjusted is input by the user, for example as a numerical value. The processor 110 accepts data indicating the selected target episode number (target time range) from the user terminal 300. Based on the data accepted from the user terminal 300, the processor 110 selects, for example, the third episode as the episode number (time range) for which the activation value is to be changed in the second non-negative matrix H (step S316).

In the activation value input section 702C, the user inputs the changed activation value, for example as a numerical value. The user inputs an arbitrary value as the changed activation value. The changed activation value is set, for example, to a value (for example, 0.7) different from the activation value (for example, 0) of the target topic obtained by NMF.

The processor 110 receives the changed activation value from the user terminal 300. The processor 110 generates a changed second non-negative matrix H by changing only the activation value of the target episode (episode 3) of the target topic F3 in the second non-negative matrix H from 0 (see FIG. 9) before the change to a changed value (0.7). The processor 110 executes a matrix operation to multiply the first non-negative matrix W shown in FIG. 8 by the changed second non-negative matrix. This matrix operation updates the appearance rate of each object in the target episode, episode 3, in the time series data 201M (non-negative matrix data) shown in FIG. 7 (step S317). Note that in the updated time series data 201M, there is no change in the appearance rate of each object other than the target episode, so the operation to update the appearance rate of objects other than the target episode may be omitted in the matrix operation.

The processor 110 transmits the updated appearance frequency of each object in the target episode, episode 3, to the user terminal 300 as the adjustment result (step S318). The user terminal 300 receives the adjustment result and displays the adjustment result in the adjustment result display section 703 of the screen 500A. The adjustment result display section 703 displays the appearance frequency (number of appearances) of each object for setting the activation value of the third topic F3 in the target episode, episode 3, to 0.7. The user can understand the appearance frequency of the objects after adjustment by referring to the adjustment result display section 703.

<3. Notes>

The present invention is not limited to the above embodiment, and various modifications are possible.

10: System 100: Story development presentation device 100A: Story development analysis device 110: Processor 111: Story development presentation processing 111A: Story development analysis processing 120: Storage device 121: Computer program 121A: Computer program 200: Database 201: First contents data 201M: First time series data 202: Second contents data 202M: Second time series data 203: Third contents data 203M: Third time series data 300: User terminal 310: Processor 320: Storage device 330: Input/output device 400: Network 500: Screen 500A: Screen 501: Display area 510: Work name display section 511: Name display section 512: Graphical image 513: First display section 514: Second display section 515: Third display section 516 : Seek bar 516A : Slider 601 : Topic display section 602 : Story development display section 701 : Object contribution display section 702 : Story development adjustment section 702A : Selection section 702B : Selection section 702C : Input section 703 : Adjustment result display section A : Character B : Character C : Character D : Character E : Building F : Building G : Tool H : Tool F1 : First topic F2 : Second topic F3 : Third topic R : Selection line S : Object information T : Time information

Claims (14)

1. A method for presenting a user with a story development in which a plurality of objects appear, comprising:
a processor reading out time series data from a database storing time series data indicating a change over time in the appearance frequency of each of the plurality of objects;
the processor generates a graphical image showing a change over time in appearance of one or more objects included in the plurality of objects based on the read time-series data;
the processor presents the graphical images to the user as the story unfolds ;
the processor further comprising selecting the one or more objects from among the plurality of objects;
the time series data is non-negative matrix data,
the one or more objects are selected based on at least one of two non-negative matrices obtained from the time series data by a non-negative matrix factorization algorithm;
How the story unfolds is presented to the user. 2. The method of claim 1, wherein the graphical images are presented to the user without accompanying text explaining the story. The processor further comprises selecting a partial time range of an entire time range of the time series data;
The method for presenting the development of a story to a user according to claim 1 or 2 , wherein the graphical image is generated based on partial time-series data in the partial time range of the time-series data. The method of claim 3 , wherein the portion of the time range is selected based on data input by the user. The method for presenting a story development to a user according to claim 3 , wherein the partial time range is selected based on a result of analyzing the time series data. 1. A method for presenting a user with a story development in which a plurality of objects appear, comprising:
a processor reading out time series data from a database storing time series data indicating a change over time in the appearance frequency of each of the plurality of objects;
the processor generates a graphical image showing a change over time in appearance of one or more objects included in the plurality of objects based on the read time-series data;
The processor presents the graphical images to the user as the story unfolds.
Prepare for this.
The processor further comprises selecting a partial time range of an entire time range of the time series data;
the graphical image is generated based on partial time series data in the partial time range of the time series data;
the time series data is non-negative matrix data,
The partial time range is selected based on at least one of two non-negative matrices obtained from the time series data by a non-negative matrix factorization algorithm.
How the story unfolds is presented to the user. the processor selecting a partial time range of the entire time range of the time series data;
Further comprising:
3. A method for presenting a storyline to a user according to claim 1 or claim 2, wherein the graphical images are generated to indicate a temporal change in the appearance of the one or more objects within the portion of time. A database storing time-series data showing changes in the appearance frequency of each of a plurality of objects appearing in a story over time;
a processor configured to execute a process of generating graphical images showing a change over time in appearance of one or more objects included in the plurality of objects based on the time-series data read from the database, and presenting the graphical images to a user as the story progresses;
Equipped with
the processor is configured to select the one or more objects from among the plurality of objects;
the time series data is non-negative matrix data,
the one or more objects are selected based on at least one of two non-negative matrices obtained from the time series data by a non-negative matrix factorization algorithm;
A device that presents story development. A database storing time-series data showing changes in the appearance frequency of each of a plurality of objects appearing in a story over time;
a processor configured to execute a process of generating graphical images showing a change over time in appearance of one or more objects included in the plurality of objects based on the time-series data read from the database, and presenting the graphical images to a user as the story progresses;
Equipped with
The processor is configured to select a portion of a time range of an entire time range of the time series data;
the graphical image is generated based on partial time series data in the partial time range of the time series data,
the time series data is non-negative matrix data,
The partial time range is selected based on at least one of two non-negative matrices obtained from the time series data by a non-negative matrix factorization algorithm.
A device that presents story development. A computer program for causing a computer to execute a process of generating graphical images showing a change over time in appearance of one or more objects included in a plurality of objects, based on time-series data showing a change over time in appearance of each of the plurality of objects appearing in a story read from a database storing the time-series data, and presenting the graphical images to a user as the story progresses ,
the processing includes selecting the one or more objects from among the plurality of objects;
the time series data is non-negative matrix data,
the one or more objects are selected based on at least one of two non-negative matrices obtained from the time series data by a non-negative matrix factorization algorithm;
Computer program. A computer program for causing a computer to execute a process of generating graphical images showing a change over time in appearance of one or more objects included in a plurality of objects, based on time-series data showing a change over time in appearance of each of the plurality of objects appearing in a story read from a database storing the time-series data, and presenting the graphical images to a user as the story progresses,
The process includes selecting a partial time range from an entire time range of the time series data;
the graphical image is generated based on partial time series data in the partial time range of the time series data;
the time series data is non-negative matrix data,
The partial time range is selected based on at least one of two non-negative matrices obtained from the time series data by a non-negative matrix factorization algorithm.
Computer program. 1. A method for analyzing a story development involving multiple objects, comprising:
a processor reads out time-series data from a database storing time-series data indicating a change over time in the appearance frequency of each of the plurality of objects and being non-negative matrix data;
The processor obtains two non-negative matrices as an analysis result of the development of the story by applying a non-negative matrix factorization algorithm to the time series data.
A method for analyzing the development of a story. A database storing time-series data that indicates a change over time in the appearance frequency of each of a plurality of objects that appear in a story and is non-negative matrix data;
a processor configured to execute a process of obtaining two non-negative matrices as an analysis result of the development of the story by applying a non-negative matrix factorization algorithm to the time-series data read from the database;
A story development analysis device comprising: A computer program for causing a computer to execute a process of obtaining two non-negative matrices as an analysis result of the development of a story by applying a non-negative matrix factorization algorithm to time series data read from a database in which the time series data indicates a temporal change in the appearance frequency of each of a number of objects that appear in a story and is non-negative matrix data.

Images