Information processing device and advertising data generation method

Abstract

This invention provides an information processing device, an advertising data generation method, and a program that facilitate the creation of highly appealing manga advertisements. [Solution] In an advertising generation system in which a PC and a server are connected to each other via a network including a wireless communication network and a wired communication network, the server 14 includes a cutting unit 30 that cuts out panels from a manga, an evaluation unit 36 ​​that derives an evaluation value indicating the likelihood of each panel attracting attention based on the characteristic parts of each panel cut out by the cutting unit 30, and a banner generation unit 38 that generates advertising data using the panels selected based on the evaluation values. The evaluation unit 36 ​​derives an evaluation value for each panel based on evaluation rules predetermined for each characteristic part of the panel.

Description

Technical Field

[0001] The present invention relates to an information processing apparatus and an advertisement data generation method.

Background Art

[0002] In recent years, so-called web manga, which enables digital manga to be viewed on websites, has become widely popular.

[0003] Here, Patent Document 1 describes an information processing apparatus that extracts images of important scenes and characteristic scenes from a plurality of images based on content such as still image content and video content, and generates digest content with a manga-style layout conversion.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] Here, in order to increase the ability to attract customers to web manga, frames that form the web manga are extracted, and banners as advertisements are generated using the extracted frames. However, it has been labor-intensive to manually extract appropriate frames as advertisements from a large number of frames that form the manga.

[0006] Note that the information processing apparatus described in Patent Document 1 extracts scenes and generates digest content, but this digest content may not necessarily be able to extract scenes appropriate as advertisements for web manga.

[0007] Therefore, an object of the present invention is to provide an information processing apparatus and an advertisement data generation method that can easily create manga advertisements with high appeal. [Means for solving the problem]

[0008] The information processing device of the present invention comprises: a cutting unit that cuts out panels from a manga; an evaluation unit that derives an evaluation value indicating the likelihood of each panel attracting attention based on the characteristic portion of each panel cut out by the cutting unit; and a generation unit that generates advertising data using the panels selected based on the evaluation value.

[0009] The present invention provides an advertising data generation method comprising: a first step in which a cutting unit cuts out a panel from a manga; a second step in which an evaluation unit derives an evaluation value indicating the likelihood of the panel attracting attention based on the characteristic portion of the panel cut out by the cutting unit; and a third step in which a generation unit generates advertising data using the panel selected based on the evaluation value. [Effects of the Invention]

[0010] According to the present invention, highly appealing manga advertisements can be easily created. [Brief explanation of the drawing]

[0011] [Figure 1] This is a diagram illustrating the configuration of the manga advertisement generation system according to the first embodiment. [Figure 2] This is a functional block diagram of the server related to the manga advertisement generation process of the first embodiment. [Figure 3] This figure shows the derivation of the evaluation value for the piece in the first embodiment. [Figure 4] This flowchart shows the flow of the manga advertisement generation process in the first embodiment. [Figure 5] This is a functional block diagram of the server related to the manga advertisement generation process in the second embodiment. [Figure 6] This is a flowchart showing the flow of the manga advertisement generation process in the second embodiment. [Modes for carrying out the invention]

[0012] Embodiments of the present invention will be described below with reference to the drawings. The embodiments described below are merely examples of how the present invention can be implemented, and the present invention is not limited to the specific configurations described below. In implementing the present invention, specific configurations may be adopted as appropriate depending on the embodiment.

[0013] (First Embodiment) Figure 1 is a diagram showing the configuration of the manga advertisement generation system 10 of this embodiment. The manga advertisement generation system 10 comprises at least one personal computer (hereinafter referred to as "PC") 12 and a server 14. The PC 12 and the server 14 are connected to each other via a network 16, which includes a wireless communication network and a wired communication network.

[0014] PC12 is an information processing device such as a desktop computer, laptop computer, smartphone, or tablet terminal equipped with a display 13, and is operated by a worker to generate advertising data (hereinafter referred to as "banner") for web comics that can be viewed on a website. Specifically, the worker uses PC12 to select the comic for which to create a banner, and the server 14 executes the comic advertisement generation process for the selected comic. The banners in this embodiment are displayed on websites, SNS (Social Networking Service), etc.

[0015] Server 14 is an information processing device that stores data for multiple manga and performs manga advertisement generation processing for manga selected by the operator. The manga stored as data in Server 14 include, for example, image data obtained by scanning paper manga or manga generated as digital data. In this embodiment, the manga is formed from multiple panels and divided into pages. The manga is also defined by the volume number or chapter number of the collected volumes.

[0016] Manga data is associated with attribute data such as the manga's author, publication date, and genre. Manga genres include, for example, romance, fantasy, battle, and sports.

[0017] The manga advertisement generation process of this embodiment derives an evaluation value indicating the degree of easy attention for each frame based on the characteristic parts of each frame cut from the manga, and generates advertisement data using the frames selected based on the evaluation value. Thereby, the manga advertisement generation process of this embodiment can easily create advertisements for manga with high appeal power.

[0018] FIG. 2 is a functional block diagram regarding the manga advertisement generation process executed by the server 14.

[0019] The server 14 includes a storage unit 20, a communication unit 22, and an arithmetic unit 24.

[0020] The communication unit 22 transmits and receives data via the network 16 to and from the PC 12 and other information processing devices.

[0021] The storage unit 20 stores programs for executing the manga advertisement generation process, manga data, and various other programs and data.

[0022] The arithmetic unit 24 is a computer such as a CPU (Central Processing Unit), and includes a cutting unit 30, an extraction unit 32, a trimming unit 34, an evaluation unit 36, and a banner generation unit 38. Note that these functions are realized by executing a program stored in the storage unit 20 included in the server 14.

[0023] The cutting unit 30 reads out the manga data stored in the storage unit 20 and cuts out frames from the manga. The manga to be read out is selected by an operator using the PC 12. Information such as the number of episodes, volumes, pages of the cut manga, and numbers (hereinafter referred to as "frame numbers") sequentially assigned from the beginning of each episode is associated with the frames cut by the cutting unit 30. The frame number is information indicating the time series of the frames.

[0024] The cutting unit 30 of this embodiment cuts out panels from the manga that are included in at least one of a predetermined page, chapter, or volume, as the target for deriving the evaluation value. As a result, the manga advertisement generation process of this embodiment evaluates only specific panels that are appropriate as advertisements. These specific panels are, for example, panels that contain exciting scenes in the manga, such as the first 5 pages or the last 5 pages of each chapter, or panels from chapters 1 to 3 or volume 1. The pages, chapters, or volumes from which the cutting unit 30 cuts out panels are predetermined.

[0025] The manga advertisement generation process of this embodiment can increase the sense of anticipation for the manga among viewers of the banner displayed on a website, etc., by generating a banner using such specific panels. Furthermore, by performing evaluation only on specific panels, the overall processing load is reduced compared to performing evaluation on all panels.

[0026] The extraction unit 32 extracts feature portions for each frame cut out by the cutting unit 30. In this embodiment, the extraction unit 32 extracts feature portions for frames that do not meet the cutoff criteria. The feature portions for each frame are extracted by image recognition. Feature portions include, for example, the proportion of the page occupied by the frame, the aspect ratio of the frame, the presence or absence of characters in the frame, the presence or absence of main characters in the frame, the orientation of the characters' faces in the frame, the characters' expressions, the proportion of the frame occupied by the characters' faces, the presence or absence of stylization of the characters in the frame, the amount of text in the frame, and so on.

[0027] The cutoff unit 34 selects frames that satisfy predetermined conditions (hereinafter referred to as "cutoff conditions") and will not derive an evaluation value from them. The cutoff conditions will be described in detail later.

[0028] The evaluation unit 36 ​​derives an evaluation value indicating the likelihood of each frame attracting attention, based on the characteristic parts of each frame extracted by the extraction unit 32. In this embodiment, the evaluation unit 36 ​​derives an evaluation value for each frame based on predetermined evaluation rules for each characteristic part of the frame. In this way, since the evaluation unit 36 ​​of this embodiment derives an evaluation value for each frame based on predetermined evaluation rules, it is easy to evaluate the frame as an advertisement.

[0029] The banner generation unit 38 generates a banner, which is advertising data, using frames selected based on evaluation values. In this embodiment, the manga advertising generation process displays frames on the PC 12's display 13 based on evaluation values, and the banner generation unit 38 generates a banner using frames selected by the operator from among the frames displayed on the display 13. The frames displayed on the PC 12's display 13 are, for example, frames with relatively high evaluation values. As a result, the banner generation unit 38 generates a banner using frames with relatively high evaluation values.

[0030] The banner generated by the banner generation unit 38 displays, for example, the title of the manga, the URL of the website where the manga can be viewed (Uniform Resource Locator), and at least one selected panel.

[0031] Furthermore, if multiple panels are selected to be used as a banner, the selected panels will be displayed in chronological order based on their panel numbers in GIF format (hereinafter referred to as "GIF image"). By generating a banner that includes a GIF image using multiple panels, viewers of the banner displayed on a website, etc., can predict the general flow of the manga, resulting in a more appealing advertisement.

[0032] Next, referring to Figure 3, the derivation of evaluation values ​​for panels by the manga advertisement generation process of this embodiment will be explained. Figure 3 is an example of a one-page manga used to derive evaluation values ​​for each panel. The genre of the manga shown in Figure 3 is romance, as an example.

[0033] As described above, the manga advertisement generation process of this embodiment does not derive an evaluation value for panels that satisfy the cutoff conditions. An example of the cutoff conditions is shown below. Note that the following cutoff conditions are just examples, and not all of the following conditions need to be included in the cutoff conditions; other conditions may also be included.

[0034] Disqualification condition 1: The number of characters in the frame is insufficient. Disqualification condition 2: Even if a character is included in the panel, the character's facial expression is not depicted. For example, the character is facing away. Disqualification condition 3: The frame does not fit within a certain aspect ratio. Disqualification condition 4: No lines. Disqualification condition 5: The area of ​​the dialogue (speech bubble) is equal to or greater than the specified proportion of the panel. Disqualification condition 6: The entire character's face does not fit within the frame. Disqualification condition 7: The character included in the frame is not a main character. Disqualification condition 8: The faces of the characters included in the frame are stylized or distorted.

[0035] In the example shown in Figure 3, frame 40A satisfies cutoff condition 4, so no evaluation value is derived for it. Similarly, frame 40B satisfies cutoff condition 1, so no evaluation value is derived for it. On the other hand, frames 40C and 40D do not satisfy any of the cutoff conditions, so evaluation values ​​are derived for them by the evaluation unit 36.

[0036] The following is an example of an evaluation rule for manga panels in the romance genre, with a score assigned to each evaluation rule. The evaluation rule is determined, for example, based on the relationship between the characteristic parts of the panels used in banners that have been distributed as advertisements and the click-through rate of the banners.

[0037] Rule 1: Two characters' faces are close together (10 points) Rule 2: Anything else (5 points) Rule 3: Panel size is 50% or more of a page (+2 points)

[0038] Piece 40C is evaluated at 5 points because it conforms to Rule 2. Piece 40D is evaluated at 7 points, which is the sum of 5 points and 2 points, because it conforms to both Rule 2 and Rule 3. These individual scores become the evaluation values ​​for each piece.

[0039] Furthermore, the criteria for deriving evaluation values ​​(evaluation rules in this embodiment) differ for each manga genre. That is, the evaluation rules differ for each genre, such as romance, fantasy, battle, and sports. This allows for the use of appropriate panels in advertising data for each manga genre. Also, the cutoff conditions may differ for each manga genre.

[0040] Figure 4 is a flowchart showing the flow of the manga advertisement generation process in this embodiment.

[0041] First, in step 100, the operator uses PC 12 to select the manga to be used for banner generation. The selected manga is stored in the memory unit 20 of server 14.

[0042] In the next step 102, the cutting unit 30 uses image recognition to cut out panels from the manga selected in step 100. Note that the panels to be cut out are not all the panels from all volumes and pages of the manga, but rather panels included in a predetermined number of pages, chapters, or volumes, for example.

[0043] In the next step 104, the extraction unit 32 extracts the characteristic parts of the frame.

[0044] In the next step 106, the cutoff unit 34 selects frames that meet the cutoff criteria and excludes them from evaluation.

[0045] In the next step 108, for panels that do not meet the cutoff criteria, the evaluation unit 36 ​​derives an evaluation value for each panel using evaluation rules appropriate to the genre of the manga.

[0046] In the next step 110, the banner generation unit 38 displays the frames with the highest evaluation values ​​on the PC 12's display 13. The frames displayed on the display 13 are, for example, 20 frames in descending order of evaluation value. Alternatively, for example, a group of consecutive frames (hereinafter referred to as a "group of consecutive frames") with a high average evaluation value may be displayed.

[0047] In the next step 112, the worker selects at least one frame from the multiple frames displayed on the display 13 to be used for the banner.

[0048] In the next step 114, the banner generation unit 38 generates a banner using the frame selected in step 110, and the manga advertisement generation process ends.

[0049] As described above, the manga advertisement generation process of this embodiment derives an evaluation value for each panel based on predetermined evaluation rules for each characteristic part of the panel, and generates a banner using the panel selected based on the evaluation value.

[0050] (Second Embodiment) The manga advertisement generation process in this embodiment uses a machine learning model to evaluate the panels. Note that the configuration of the manga advertisement generation system 10 in the second embodiment is the same as that of the manga advertisement generation system 10 in the first embodiment, so its description is omitted.

[0051] Figure 5 is a functional block diagram relating to the manga advertisement generation process executed by the server 14 of this embodiment. Note that parts of the functional block diagram according to the second embodiment that are the same as those in the first embodiment are denoted by the same reference numerals, and their descriptions are omitted.

[0052] The calculation unit 24 includes a model generation unit 50, a cutting unit 30, an extraction unit 32, a leg trimming unit 34, an evaluation unit 52, and a banner generation unit 38.

[0053] The model generation unit 50 generates an evaluation model using machine learning with the frames used in the banner and the click rate for the banner as training data. This evaluation model derives an evaluation value indicating how likely each input frame is to attract attention.

[0054] For example, the model generation unit 50 generates an evaluation model using training data based on the placement history of delivered banners, assigning high scores to banners with high click rates and low scores to banners with low click rates. This allows the model to learn the characteristics of banners that have high click rates.

[0055] Furthermore, the model generation unit 50 generates different evaluation models for each manga genre. That is, the evaluation model differs for each genre, such as romance, fantasy, battle, and sports. For this reason, the distributed banners used as training data are banners from manga of the same genre. As a result, the evaluation model has different criteria for each manga genre, and appropriate panels according to the manga genre can be used in the advertising data.

[0056] The evaluation model may learn the relationship between the time-series position of the banner panels in the manga used as training data and the click-rate, and assign a high evaluation value to panels with appropriate time-series positions as banners.

[0057] The evaluation unit 52 inputs the frames cut out by the cutting unit 30 into the evaluation model generated by the model generation unit 50, thereby deriving an evaluation value for each frame. As described above, the evaluation model differs for each genre of manga, so the evaluation unit 36A uses the evaluation model corresponding to the genre of manga for which the banner is to be generated to derive an evaluation value for each frame. Frames that the trimming unit 34 determines to meet the trimming criteria are not evaluated by the evaluation model.

[0058] Figure 6 is a flowchart showing the flow of the manga advertisement generation process in this embodiment. Parts of the manga advertisement generation process in the second embodiment that are the same as those in the first embodiment are denoted by the same reference numerals, and their descriptions are omitted.

[0059] In step 106, the cutoff unit 34 selects a frame that meets the cutoff criteria, and then proceeds to step 200.

[0060] In step 200, the evaluation unit 52 uses the evaluation model to derive an evaluation value for each frame, and then proceeds to step 108.

[0061] As described above, the manga advertisement generation process of this embodiment uses the characteristic parts of the panels used in the banner and the click rate for the banner as training data to input the panels into an evaluation model, derives an evaluation value for each panel, and generates a banner using the panels selected based on the evaluation value.

[0062] Although the present invention has been described above using the embodiments described above, the technical scope of the present invention is not limited to the scope described in the embodiments above. Various modifications or improvements can be made to the embodiments above without departing from the spirit of the invention, and such modified or improved forms are also included in the technical scope of the present invention.

[0063] In the above embodiment, a form was described in which the evaluation value derived by the evaluation unit 36 ​​is expressed as a score, but the present invention is not limited thereto. For example, the evaluation unit 36 ​​may express the evaluation value as a rank such as A, B, C, instead of a score. In the first embodiment, for example, a score is derived as the evaluation value, but the range of scores may be expressed as a rank such as A, B, C as the output presented to the worker. In the second embodiment, an evaluation model may be generated by ranking the delivered banners as A, B, C according to the click rate and using this as training data, and ranks A, B, C may be derived for the frames input into the evaluation model.

[0064] In the above embodiment, the banner generation unit 38 was described as generating a banner using frames with high evaluation values ​​selected by the operator, but the present invention is not limited to this. The banner generation unit 38 may automatically generate a banner using frames with high evaluation values ​​without the operator selecting any frames.

[0065] The following additional notes are provided regarding the embodiments described above. [Note 1] The information processing device described in Appendix 1 comprises: a cutting unit that cuts out panels from a manga; an evaluation unit that derives an evaluation value indicating the likelihood of each panel attracting attention based on the characteristic parts of each panel cut out by the cutting unit; and a generation unit that generates advertising data using the panels selected based on the evaluation value.

[0066] According to this configuration, an evaluation value indicating the likelihood of each panel attracting attention is derived based on the characteristic features of each panel extracted from the manga, and advertising data is generated using panels selected based on these evaluation values. Therefore, this configuration makes it possible to easily create highly appealing manga advertisements.

[0067] [Note 2] In the information processing device described in Appendix 1, the evaluation unit may derive the evaluation value for each frame based on a predetermined evaluation rule for each characteristic part of the frame.

[0068] This configuration allows for easy evaluation of each frame as an advertisement, as evaluation values ​​are derived for each frame based on predetermined evaluation rules.

[0069] [Note 3] In the information processing device described in Appendix 1, the evaluation unit may derive the evaluation value for the frame by inputting the frame into an evaluation model that has been machine-trained using the frame used in the advertising data and the click rate for the advertising data as training data.

[0070] With this configuration, evaluation values ​​are derived for each frame based on a machine learning-based evaluation model, making it easy to evaluate each frame as an advertisement.

[0071] [Note 4] In any of the information processing devices described in Appendix 1 to Appendix 3, the evaluation value does not need to be derived for the frame that satisfies the predetermined conditions.

[0072] With this configuration, since frames that are not suitable as advertisements are not evaluated, the overall processing load for evaluating frames as advertisements is reduced.

[0073] [Note 5] In any of the information processing devices described in Appendix 1 to Appendix 4, the cutting unit may cut out a panel from the manga that is included in at least one of a predetermined page, chapter, and volume, as the target for deriving the evaluation value.

[0074] This configuration allows for evaluation of only specific frames that are appropriate as advertisements. These specific frames are, for example, frames containing exciting scenes in the manga, which can increase viewers' anticipation for the manga. Furthermore, by evaluating only specific frames, the overall processing load is reduced compared to evaluating all frames.

[0075] [Note 6] The information processing device described in Appendix 1 to Appendix 5 may have different criteria for deriving the evaluation value for each genre of manga.

[0076] According to this configuration, appropriate panels corresponding to the genre of manga can be used in advertising data.

[0077] [Note 7] In the information processing device described in Appendix 1, the frames may be displayed on a display based on the evaluation value, and the generation unit may generate the advertising data using selected frames from among those displayed on the display.

[0078] This configuration allows advertising staff to identify the appropriate frames for use as advertising data.

[0079] [Note 8] The advertising data generation method described in Appendix 8 comprises: a first step in which a cutting unit cuts out a panel from a manga; a second step in which an evaluation unit derives an evaluation value indicating the likelihood of the panel attracting attention based on the characteristic portion of the panel cut out by the cutting unit; and a third step in which a generation unit generates advertising data using the panel selected based on the evaluation value. [Explanation of Symbols]

[0080] 10 Ad Generation System 14. Server (Information Processing Device) 30 Cutout 36 Evaluation Department 38. Generation Unit (Banner Generation Unit)

Claims

1. The part for cutting out panels from the manga, An evaluation unit that derives an evaluation value indicating the likelihood of each frame being noticed based on the characteristic portion of each frame cut out by the aforementioned cutting unit, A generation unit that generates advertising data using the frame selected based on the evaluation value, An information processing device equipped with the following features.

2. The information processing apparatus according to claim 1, wherein the evaluation unit derives the evaluation value for each frame based on a predetermined evaluation rule for each characteristic part of the frame.

3. The information processing apparatus according to claim 1, wherein the evaluation unit derives the evaluation value for the frame by inputting the frame into an evaluation model that has been machine-trained using the frame used in the advertising data and the click rate for the advertising data as training data.

4. The information processing apparatus according to claim 1, wherein the evaluation value is not derived for the frame that satisfies the predetermined conditions.

5. The information processing apparatus according to claim 1, wherein the cutting section cuts out a panel from the manga that is included in at least one of a predetermined page, chapter, and volume, as the target for deriving the evaluation value.

6. The information processing apparatus according to claim 1, wherein the criteria for deriving the evaluation value differ for each genre of manga.

7. Based on the evaluation value, the frame is displayed on the display. The information processing apparatus according to claim 1, wherein the generation unit generates the advertising data using selected frames from among the frames displayed on the display.

8. The cutting section is the first step of cutting out panels from the manga, A second step in which the evaluation unit derives an evaluation value indicating the likelihood of the frame attracting attention based on the characteristic portion of the frame cut out by the cutting unit, A third step in which the advertising data generation unit generates advertising data using the frame selected based on the evaluation value, A method for generating advertising data that includes the following:

Images