Image generation apparatus, image generation method, and program

Abstract

To provide a new technique for generating an image from a text.SOLUTION: An image generation apparatus acquires sentence data, extracts a plurality of keywords from the sentence data, and inputs the keywords to an image generation model to generate an image related to the sentence data.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present disclosure relates to an image generation device, an image generation method, and a program. [Background technology]

[0002] Technologies for generating images from text have been developed. For example, Patent Document 1 discloses a system that generates a prompt from one or more tags selected by a user and automatically generates a background image for an illustration using the prompt. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 7398723 Summary of the Invention [Problem to be solved by the invention]

[0004] In the system of Patent Document 1, the user needs to select the characteristics of the background image they want to generate from pre-prepared tags. This disclosure has been made in consideration of this problem, and one of its purposes is to provide a new technology for generating images from text. [Means for solving the problem]

[0005] The image generation device according to the present disclosure includes an acquisition means for acquiring text data, an extraction means for extracting a plurality of keywords from the text data, and a generation means for generating an image related to the text data by inputting the plurality of keywords into an image generation model.

[0006] An image generation method according to the present disclosure is executed by a computer and includes an acquisition step of acquiring text data, an extraction step of extracting a plurality of keywords from the text data, and a generation step of generating an image related to the text data by inputting the plurality of keywords into an image generation model.

[0007] The program of the present disclosure causes a computer to execute an acquisition step of acquiring text data, an extraction step of extracting multiple keywords from the text data, and a generation step of generating an image related to the text data by inputting the multiple keywords into an image generation model. [Effects of the Invention]

[0008] According to the present disclosure, new techniques for generating images from text are provided. [Brief explanation of the drawings]

[0009] [Figure 1] FIG. 1 is a diagram illustrating an example of an overview of an image generating device. [Figure 2] FIG. 2 is a block diagram illustrating an example of the functional configuration of the image generating apparatus. [Figure 3] FIG. 2 is a block diagram illustrating an example of the hardware configuration of a computer that realizes the image generating apparatus. [Figure 4] 10 is a flowchart illustrating the flow of a process executed by the image generating device. DETAILED DESCRIPTION OF THE INVENTION

[0010] Hereinafter, embodiments of the present disclosure will be described in detail with reference to the drawings. In each drawing, the same or corresponding elements are designated by the same reference numerals, and duplicate explanations will be omitted as necessary for clarity. Furthermore, unless otherwise specified, predetermined values ​​such as predetermined values ​​and threshold values ​​are stored in advance in a storage device accessible from a device that uses the values. Furthermore, unless otherwise specified, the storage unit is composed of one or any number of storage devices.

[0011] <Summary> Fig. 1 is a diagram illustrating an overview of an image generation device 2000. The operation of the image generation device 2000 illustrated in Fig. 1 is an example intended to facilitate understanding of the image generation device 2000. The operations that can be performed by the image generation device 2000 are not limited to the operations illustrated in Fig. 1.

[0012] The image generating device 2000 generates image data 40, which is an image associated with the content of the text data 10. The text data 10 is text data representing any text related to a specific topic (hereinafter, the target topic). For example, the target topic may be cybersecurity. When the target topic is cybersecurity, for example, the text data 10 may represent an explanation of a malicious attack such as a phishing email, or a security report.

[0013] Here, the image generation device 2000 may be configured to handle only one topic (for example, only cybersecurity) as the target topic, or may be configured to allow a target topic to be selected from multiple topics. In the latter case, the topic to be handled as the target topic is selected in some way in the image generation device 2000. The method for selecting the target topic will be described later.

[0014] To generate image data 40 from text data 10, for example, the image generation device 2000 operates as follows. First, the image generation device 2000 acquires the text data 10. Next, the image generation device 2000 extracts a plurality of keywords 20 from the text data 10. The keywords 20 are words related to the target topic. For example, if the target topic is cybersecurity, words related to cybersecurity are extracted as the keywords 20.

[0015] The image generation device 2000 inputs multiple keywords 20 into the image generation model 30. The image generation model 30 is trained in advance to output one or more image data in response to the input of multiple words. The image data output from the image generation model 30 is an image that visualizes information associated with the input multiple words.

[0016] When the image generation model 30 is configured to output multiple pieces of image data, these pieces of image data are, for example, time-series image data. The time-series image data may or may not be video data. In the latter case, for example, the multiple pieces of image data represent changes in a situation or the flow of a procedure in time series, as in a picture-story show. When the image generation model 30 is configured to output time-series image data, the image generation device 2000 can obtain time-series image data 40 in which information associated with multiple keywords 20 is visualized.

[0017] The image generation model 30 may be provided inside the image generation device 2000 or outside the image generation device 2000. In the latter case, the image generation model 30 may be a dedicated image generation model prepared for generating the image data 40, or may be a general-purpose image generation model that can be used for purposes other than generating the image data 40.

[0018] <Examples of effects> According to the image generating device 2000, a plurality of keywords 20 are extracted from the text data 10, and image data 40 is generated using the extracted plurality of keywords 20. In this way, the image generating device 2000 provides a new technique for generating an image from text that is not disclosed in Patent Document 1.

[0019] Furthermore, a user of the system of Patent Document 1 must manually select tags to be assigned to the model. In contrast, the image generation device 2000 automatically extracts keywords 20 from text data 10, eliminating the need for a user of the image generation device 2000 to manually select keywords to be assigned to the image generation model 30. Therefore, the image generation device 2000 can reduce the user's effort required to generate an image.

[0020] Furthermore, in the system of Patent Document 1, tags to be assigned to a model can only be selected from among predetermined tags. In contrast, any text can be assigned to the image generating device 2000. In this way, the image generating device 2000 can handle input of a wider range of content, making it highly convenient.

[0021] The image generating device 2000 of this embodiment will be described in more detail below.

[0022] <Example of functional configuration> 2 is a block diagram illustrating an example of the functional configuration of an image generation device 2000. The image generation device 2000 includes an acquisition unit 2020, an extraction unit 2040, and a generation unit 2060. The acquisition unit 2020 acquires text data 10. The extraction unit 2040 extracts a plurality of keywords 20 from the text data 10. The generation unit 2060 generates image data 40 by inputting the plurality of keywords 20 into an image generation model 30.

[0023] <Example of hardware configuration> Each functional component of image generating device 2000 may be realized by hardware that realizes the functional component (e.g., a hardwired electronic circuit, etc.), or may be realized by a combination of hardware and software (e.g., a combination of an electronic circuit and a program that controls it, etc.). Below, a case where each functional component of image generating device 2000 is realized by a combination of hardware and software will be further described.

[0024] 3 is a block diagram illustrating an example of the hardware configuration of a computer 1000 that realizes the image generating device 2000. The computer 1000 is any computer. For example, the computer 1000 is a stationary computer such as a PC (Personal Computer) or a server machine. Alternatively, the computer 1000 may be a portable computer such as a smartphone or a tablet terminal. The computer 1000 may be a dedicated computer designed to realize the image generating device 2000, or may be a general-purpose computer.

[0025] For example, by installing a predetermined application on the computer 1000, the computer 1000 realizes each function of the image generating device 2000. The application is configured with a program for realizing each functional component of the image generating device 2000. The program can be acquired by any method. For example, the program can be acquired from a storage medium on which the program is stored. The storage medium on which the program is stored can be any storage medium such as a DVD (Digital Versatile Disk) or a USB (Universal Serial Bus) memory. Alternatively, the program can be acquired by downloading the program from a server device that manages the storage device on which the program is stored.

[0026] The computer 1000 has a bus 1020, a processor 1040, a memory 1060, a storage device 1080, an input / output interface 1100, and a network interface 1120. The bus 1020 is a data transmission path for the processor 1040, the memory 1060, the storage device 1080, the input / output interface 1100, and the network interface 1120 to transmit and receive data to and from each other. However, the method for connecting the processor 1040 and the like to each other is not limited to a bus connection.

[0027] The processor 1040 is one of various processors such as a central processing unit (CPU), a graphics processing unit (GPU), or a field-programmable gate array (FPGA). The memory 1060 is a main storage device realized using a random access memory (RAM) or the like. The storage device 1080 is an auxiliary storage device realized using a hard disk, a solid state drive (SSD), a memory card, or a read only memory (ROM) or the like.

[0028] The input / output interface 1100 is an interface for connecting the computer 1000 to an input / output device. For example, the input / output interface 1100 is connected to an input device such as a keyboard and an output device such as a display device.

[0029] The network interface 1120 is an interface for connecting the computer 1000 to a network. This network may be a LAN (Local Area Network) or a WAN (Wide Area Network).

[0030] The storage device 1080 stores a program (a program that realizes the above-mentioned application) that realizes each functional component of the image generating device 2000. The processor 1040 reads this program into the memory 1060 and executes it, thereby realizing each functional component of the image generating device 2000.

[0031] The image generating device 2000 may be realized by one computer 1000, or may be realized by multiple computers 1000. In the latter case, the configurations of the computers 1000 do not need to be the same, and can be different from each other.

[0032] <Processing flow> 4 is a flowchart illustrating the flow of processing executed by the image generation device 2000. The acquisition unit 2020 acquires text data 10 (S102). The extraction unit 2040 extracts multiple keywords 20 from the text data 10 (S104). The generation unit 2060 generates image data 40 by inputting the multiple keywords 20 into the image generation model 30 (S106).

[0033] <Getting text data 10: S102> The acquisition unit 2020 acquires the text data 10 (S102). There are various methods for the acquisition unit 2020 to acquire the text data 10. For example, the acquisition unit 2020 provides an input screen on which text can be input to the user of the image generating device 2000. In this case, the acquisition unit 2020 acquires text data representing the text input on the input screen as the text data 10. Note that the input screen may be configured to allow a file (e.g., a document file in any format) containing text data representing the text to be specified instead of inputting text data. In this case, the acquisition unit 2020 acquires the text data contained in the file specified on the input screen as the text data 10.

[0034] Alternatively, for example, the text data 10 may be stored in advance in an arbitrary storage unit in a manner accessible from the image generating device 2000. The acquiring unit 2020 acquires the text data 10 by reading the text data 10 from the storage unit.

[0035] Alternatively, for example, the image generating device 2000 may be configured to operate in cooperation with another application. In this case, the text data 10 can be input to the image generating device 2000 from the other application.

[0036] Note that when text in various languages ​​can be input to image generating device 2000, image generating device 2000 may translate the input text into a specific language and treat the text obtained by the translation as text data 10. In this way, the targets of subsequent processing such as keyword extraction can be narrowed down to text in a specific language.

[0037] For example, suppose that the image generating device 2000 handles English sentences as sentence data 10, while the input screen allows input of sentences in any language, such as Japanese or French. In this case, the acquiring unit 2020 translates the input sentences into English and handles the English sentences obtained by the translation as sentence data 10.

[0038] <Keyword 20 extraction: S104> The extraction unit 2040 extracts a plurality of keywords 20 from the text data 10 (S104). Specifically, the extraction unit 2040 extracts words related to the target topic from the text data 10.

[0039] As described above, the image generation device 2000 may be configured to handle only one topic as the target topic, or may be configured to allow the target topic to be selected from multiple topics. The former case will be described below.

[0040] For example, the extraction unit 2040 uses information in which a plurality of words related to a target topic are defined as keywords (hereinafter referred to as keyword information). Specifically, the extraction unit 2040 extracts words indicated in the keyword information from the text data 10 and treats the extracted words as keywords 20. The keyword information is stored in advance in an arbitrary storage unit in a format that can be acquired from the image generating device 2000.

[0041] Alternatively, for example, the extraction unit 2040 may use a trained machine learning model (hereinafter referred to as a keyword extraction model). The keyword extraction model is trained in advance so that, when a sentence is input, it extracts keywords related to a target topic from the sentence. The extraction unit 2040 inputs sentence data 10 into the keyword extraction model. Then, the extraction unit 2040 treats each keyword extracted by the keyword extraction model as a keyword 20.

[0042] When the image generating device 2000 is configured to allow selection of a target topic, the above-described keyword information or keyword extraction model is prepared for each topic. For example, the extraction unit 2040 extracts keywords 20 from the text data 10 using keyword information corresponding to the selected target topic. Alternatively, for example, the extraction unit 2040 extracts keywords 20 from the text data 10 by inputting the text data 10 to a keyword extraction model corresponding to the selected target topic.

[0043] There are various methods for selecting the target topic, for example, the target topic may be set in advance by an administrator of the image generation device 2000.

[0044] Alternatively, for example, the target topic may be specified by the user of the image generating device 2000. In this case, for example, the acquiring unit 2020 acquires information indicating the target topic (hereinafter, topic information) along with the text data 10.

[0045] There are various methods for the acquisition unit 2020 to acquire topic information. For example, the acquisition unit 2020 provides an input screen on which the target topic and the text data 10 can be input to the user of the image generating device 2000. The acquisition unit 2020 acquires the text data 10 and topic information from the information input to the input screen.

[0046] For example, the input screen includes an input interface that allows a user to select one of a plurality of topics prepared in advance, and the acquisition unit 2020 treats the topic selected using the input interface as the target topic.

[0047] The text data 10 and the topic information may be stored in advance in a storage unit, in which case the acquisition unit 2020 acquires the text data 10 and the topic information from the storage unit.

[0048] Alternatively, for example, the text data 10 and topic information may be input to the image generating device 2000 from another application.

[0049] The target topic may be estimated from the content of the text data 10. In this case, for example, the extraction unit 2040 has a machine learning model (hereinafter, a topic model) that is trained in advance to estimate the topic of a text in response to the input of the text. The extraction unit 2040 inputs the text data 10 into the topic model to identify the target topic.

[0050] Here, in order to have the image generation model 30 generate image data 40 using the keywords 20, it is preferable that the keywords 20 are words that can be understood by the image generation model 30 (in other words, words that can be correctly interpreted by the image generation model 30). However, if the keywords 20 are technical terms rather than everyday words, there is a possibility that the image generation model 30 will not be able to correctly interpret the keywords 20.

[0051] Therefore, the extraction unit 2040 may replace the word extracted from the text data 10 with another word (hereinafter, a replacement word) that is assumed to be easy for the image generation model 30 to interpret, and treat the replacement word as a keyword 20.

[0052] The word replacement is performed, for example, using the above-mentioned keyword information. In this case, the keyword information indicates a replacement word for a word that needs to be replaced among the words to be extracted from the text data 10.

[0053] The replacement words are words that are relatively easy for the image generation model 30 to interpret. For example, the keyword information indicates, for each proper noun in the target topic, a word that is a superordinate concept of the proper noun as a replacement word.

[0054] The keyword information may indicate a plurality of replacement words associated with one word. In this case, by extracting one word from the text data 10, a plurality of replacement words are extracted from the text data 10. For example, a plurality of words that play an important role in explaining the meaning of a certain technical term can be used as a plurality of replacement words corresponding to the technical term.

[0055] Keyword information including replacement words is used, for example, as follows: The extraction unit 2040 detects words indicated in the keyword information from the text data 10. The following process is performed for each word detected from the text data 10. If a replacement word corresponding to a detected word is not indicated in the keyword information, the extraction unit 2040 treats the detected word as is as the keyword 20. On the other hand, if a replacement word corresponding to a detected word is indicated in the keyword information, the extraction unit 2040 treats the replacement word corresponding to the detected word as the keyword 20.

[0056] The keyword information may indicate text data (hereinafter referred to as semantic text) that indicates the meaning of a word, in association with the word to be extracted from the text data 10. The semantic text can also be expressed as an explanation of the keyword.

[0057] The semantic text is a sentence made up of words that are easy for the image generation model 30 to interpret. For example, for each proper noun in the target topic, the keyword information indicates an explanation of the proper noun using general words as the semantic text of the proper noun.

[0058] Keyword information including semantic text is used, for example, as follows: The extraction unit 2040 detects words indicated in the keyword information from the sentence data 10. The following process is performed for each word detected from the sentence data 10. If the keyword information does not indicate semantic text corresponding to the detected word, the extraction unit 2040 treats the detected word as a keyword 20. On the other hand, if the keyword information indicates semantic text corresponding to the detected word, the extraction unit 2040 performs keyword detection on the semantic text corresponding to the detected word. Then, the extraction unit 2040 treats the word extracted from the semantic text by this keyword detection as a keyword 20. The method for extracting keywords from semantic text is the same as the method for extracting keywords from the sentence data 10.

[0059] Here, the extraction unit 2040 may be configured to store a keyword once it has been extracted from a semantic text corresponding to a certain word, thereby preventing multiple keyword extractions from being performed on the same semantic text.

[0060] For example, the keyword information is configured to indicate a meaning text and a replacement word in association with a word to be extracted as a keyword. However, the keyword information in the initial state indicates the meaning text but does not indicate a replacement word.

[0061] When the keyword information indicates a replacement word corresponding to a word detected from the sentence data 10, the extraction unit 2040 treats the replacement word corresponding to that word as a keyword 20. On the other hand, when the keyword information does not indicate a replacement word corresponding to a word detected from the sentence data 10, and the keyword information indicates a meaning text corresponding to that word, the extraction unit 2040 extracts a keyword from the meaning text and treats that keyword as a keyword 20. Furthermore, the extraction unit 2040 stores the keyword extracted from the meaning text in the keyword information as a replacement word corresponding to that word.

[0062] According to the above-described process, there is no need to prepare replacement words to be included in the keyword information in advance. This saves the effort of preparing replacement words. Furthermore, it is possible to avoid extracting keywords multiple times for one semantic text.

[0063] <Generation of image data 40: S106> The generation unit 2060 generates the image data 40 by inputting a plurality of keywords 20 into the image generation model 30 and obtaining the image data 40 from the image generation model 30 (S106). The image generation model 30 may be any model that has been trained in advance to output one or more image data related to a plurality of words in response to the input of the plurality of words.

[0064] The generation unit 2060 may assign weights to each keyword 20 extracted by the extraction unit 2040. In this case, the generation unit 2060 inputs a plurality of keywords 20 and the weight of each keyword 20 to the image generation model 30. The image generation model 30 is trained in advance to generate image data in response to the input of a plurality of words and the weight of each word.

[0065] The weight of each keyword 20 is expressed, for example, as a rank. For example, the generating unit 2060 sets a default weight (for example, 0) for each keyword 20. The generating unit 2060 then weights the keywords 20 by increasing or decreasing the weight of each keyword 20 based on one or more criteria.

[0066] The default weight may be the same for all keywords 20, or may be different for each keyword. For example, the default weight is indicated in the keyword information. The more important a word is in the target topic, the larger the default weight is set.

[0067] The criteria used to weight the keywords 20 are explained below.

[0068] <<Criterion 1>> The more frequently a keyword 20 appears in the text data 10, the more likely it is to be a more important word in the text data 10. Therefore, for example, the weight of a keyword 20 is determined based on the number of times that keyword 20 appears in the text data 10. Specifically, the generation unit 2060 assigns a larger weight to a keyword 20 the more frequently it appears in the text data 10.

[0069] For this purpose, the generation unit 2060 counts the number of times each keyword 20 appears in the text data 10. Then, the generation unit 2060 sets a weight for each keyword 20 based on the number of times that keyword 20 appears.

[0070] For example, a threshold is set in advance to determine whether or not to increase the weight of the keyword 20. In this case, the generation unit 2060 adds a predetermined value (for example, 1) to the weight of the keyword 20 when the number of appearances of the keyword 20 is equal to or greater than the threshold.

[0071] Here, instead of the number of times the keyword 20 appears, the ratio of the number of times the keyword 20 appears to the total number of keywords 20 detected from the text data 10 may be used. For example, assume that the number of times the keyword "e-mail" appears is n and the total number of keywords 20 detected from the text data 10 is N. In this case, the generation unit 2060 increases the weight of the keyword "e-mail" when n / N is equal to or greater than a threshold value.

[0072] The threshold may be set in multiple stages. For example, assume that two thresholds Th1 and Th2 (Th2>Th1) are set. In this case, if the number of occurrences of keyword 20 is equal to or greater than Th2, the generation unit 2060 increases the weight of keyword 20 by A2. Also, if the number of occurrences of keyword 20 is smaller than Th2 but equal to or greater than Th1, the generation unit 2060 increases the weight of keyword 20 by A1 (A2>A1).

[0073] The generating unit 2060 may decrease the weight of the keyword 20 if the number of times the keyword 20 appears is small.

[0074] <<Criterion 2>> If the keyword 20 is not a polysemous word and is a word related to the target topic, it can be predicted that the importance of the keyword 20 is relatively high. Therefore, if the keyword 20 is not a polysemous word and is a word related to the target topic, the generation unit 2060 increases the weight of the keyword 20 by a predetermined value (for example, 1).

[0075] On the other hand, if the keyword 20 is a polysemous word and is not a word related to the target topic, it is considered that the importance of the keyword 20 is relatively low. Therefore, if the keyword 20 is a polysemous word and is not a word related to the target topic, the generation unit 2060 decreases the weight of the keyword 20 by a predetermined value (for example, 1).

[0076] Here, whether or not the keyword 20 is a polysemous word can be determined by using, for example, dictionary data. As the dictionary data, it is preferable to use data from a dictionary (such as a Japanese dictionary or an encyclopedia) that comprehensively indicates various meanings for each word. If the dictionary data indicates multiple meanings of the keyword 20, the generation unit 2060 determines that the keyword 20 is a polysemous word. On the other hand, if the dictionary data indicates only one meaning of the keyword 20, the generation unit 2060 determines that the keyword 20 is not a polysemous word.

[0077] Whether or not the keyword 20 is a word related to the target topic can be determined, for example, by using dictionary data prepared for each topic. For example, assume that the target topic is cybersecurity. In this case, dictionary data indicating words related to cybersecurity is used. If the keyword 20 is indicated in the dictionary data corresponding to the target topic, the generation unit 2060 determines that the keyword 20 is a word related to the target topic. On the other hand, if the keyword 20 is not indicated in the dictionary data corresponding to the target topic, the generation unit 2060 determines that the keyword 20 is not a word related to the target topic.

[0078] The topic used to determine whether to increase or decrease the weight, or both, may be a topic that is a higher-level concept of the target topic (hereinafter referred to as a superordinate topic). For example, if the target topic is cybersecurity, IT (Information Technology) may be used as a superordinate topic.

[0079] When the higher-level topics are used to determine whether to increase the weight, the generation unit 2060 increases the weight of the keyword 20 by a predetermined value if the keyword 20 is not a synonymous word and is a word related to the higher-level topic. When the higher-level topics are used to determine whether to decrease the weight, the generation unit 2060 decreases the weight of the keyword 20 by a predetermined value if the keyword 20 is a synonymous word and is not a word related to the higher-level topic.

[0080] <<Correction based on appearance order>> The generation unit 2060 may correct the weight of each keyword 20 determined by one or more of the above-mentioned criteria based on the order of appearance in the text data 10. For example, the correction based on the order of appearance is performed among multiple keywords 20 that have the same weight assigned to them. In this case, for example, the generation unit 2060 adds a value that is larger the earlier the keyword 20 appears in the order of appearance, and that is at most less than 1, to multiple keywords 20 that have the same weight assigned to them. In this way, it is possible to impart differences in importance among multiple keywords 20 that have the same rank assigned to them as weights.

[0081] <Result output> The image generating device 2000 outputs information representing the processing results (hereinafter, output information). The content of the output information varies. For example, the output information includes image data 40. Alternatively, the output information may be a document or an image generated by arranging text data 10 and image data 40 in predetermined positions. For example, if the text data 10 is a security report, the output information may be a document file to which an image associated with the security report is attached.

[0082] The output information may be output in various ways. For example, the image generating device 2000 stores the output information in an arbitrary storage unit. Alternatively, the image generating device 2000 may output the output information to an arbitrary display device, thereby displaying the contents of the output information on the display device. Alternatively, for example, when text data 10 is input to the image generating device 2000 from another application, the image generating device 2000 may output the output information to that application.

[0083] Although the present disclosure has been described above with reference to the embodiments, the present disclosure is not limited to the above-described embodiments. Various modifications that can be understood by those skilled in the art can be made to the configuration and details of the present disclosure within the scope of the present disclosure. Furthermore, each embodiment can be combined with other embodiments as appropriate.

[0084] Each drawing is merely an example for describing one or more embodiments. Each drawing may relate not only to one particular embodiment, but also to one or more other embodiments. As will be understood by those skilled in the art, various features or steps described with reference to any one drawing can be combined with features or steps shown in one or more other drawings to create, for example, an embodiment not explicitly shown or described. Not all features or steps shown in any one drawing are necessary to describe an exemplary embodiment, and some features or steps may be omitted. The order of steps described in any drawing may be changed as appropriate.

[0085] In the present disclosure, a program includes a set of instructions (or software code) that, when loaded into a computer, causes the computer to perform one or more functions described in the embodiments. The program may be stored in a non-transitory computer-readable medium or a tangible storage medium. By way of example and not limitation, computer-readable medium or tangible storage medium includes random-access memory (RAM), read-only memory (ROM), flash memory, solid-state drive (SSD) or other memory technology, CD-ROM, digital versatile disc (DVD), Blu-ray disc or other optical disk storage, magnetic cassette, magnetic tape, magnetic disk storage or other magnetic storage device. The program may also be transmitted on a transitory computer-readable medium or communication medium. By way of example and not limitation, transitory computer-readable medium or communication medium includes electrical, optical, acoustic, or other forms of propagated signals.

[0086] A part or all of the above-described embodiments can be described as, but not limited to, the following supplementary notes. (Appendix 1) An acquisition means for acquiring text data; extraction means for extracting a plurality of keywords from the text data; and generating means for generating an image related to the text data by inputting the plurality of keywords into an image generation model. (Appendix 2) The extraction means Acquire keyword information indicating a semantic text representing the meaning of one or more of the words to be extracted as keywords; An image generating device as described in Appendix 1, wherein if a word associated with the meaning text in the keyword information is included in the sentence data, one or more keywords are extracted from the meaning text corresponding to that word. (Appendix 3) The extraction means Acquire keyword information indicating a replacement word corresponding to one or more of the words to be extracted as keywords; 2. The image generating device according to claim 1, wherein, when a word associated with the replacement word in the keyword information is included in the text data, the replacement word corresponding to that word is extracted as the keyword. (Appendix 4) The generating means determining a weight for each of the keywords based on the characteristics of each of the keywords; 2. The image generation device of claim 1, wherein each of the keywords and a weight of each of the keywords is input into the image generation model. (Appendix 5) 5. The image generating device according to claim 4, wherein the generating means determines a weight for each of the keywords based on the number of times each of the keywords appears in the text data. (Appendix 6) 5. The image generating device according to claim 4, wherein the generating means increases the weight of the keyword if the keyword is not a polysemous word and is a word related to a specific topic. (Appendix 7) 5. The image generating device according to claim 4, wherein the generating means reduces the weight of the keyword if the keyword is a polysemous word and is not a word related to a specific topic. (Appendix 8) an acquisition step of acquiring sentence data; an extraction step of extracting a plurality of keywords from the text data; and generating an image related to the text data by inputting the plurality of keywords into an image generation model. (Appendix 9) In the extraction step, Acquire keyword information indicating a semantic text representing the meaning of one or more of the words to be extracted as keywords; An image generation method as described in Appendix 8, wherein, if a word associated with the meaning text in the keyword information is included in the sentence data, one or more keywords are extracted from the meaning text corresponding to that word. (Appendix 10) an acquisition step of acquiring sentence data; an extraction step of extracting a plurality of keywords from the text data; a generation step of generating an image related to the text data by inputting the plurality of keywords into an image generation model.

[0087] Some or all of the elements (e.g., configurations and functions) described in Appendix 2 that are dependent on Appendix 1 may also be dependent on Appendix 10. Also, some or all of the elements (e.g., configurations and functions) described in Appendix 3 to Appendix 7 that are dependent on Appendix 1 may also be dependent on Appendix 9 and Appendix 10 in the same dependency relationship as Appendix 3 to Appendix 7. Some or all of the elements described in any appendix may be applied to various hardware, software, recording means for recording software, systems, and methods. [Explanation of symbols]

[0088] 10 Text data 20 Keywords 30 Image generation model 40 Image data 1000 computers 1000 each computer 1020 Bus 1040 processor 1060 memory 1080 storage device 1100 Input / Output Interface 1120 Network Interface 2000 Image Generation Device 2020 Acquisition Department 2040 Extraction part 2060 Generation part

Claims

1. An acquisition means for acquiring text data; extraction means for extracting a plurality of keywords from the text data; and generating means for generating an image related to the text data by inputting the plurality of keywords into an image generation model.

2. The extraction means Acquire keyword information indicating a semantic text representing the meaning of one or more of the words to be extracted as keywords; The image generating device according to claim 1 , wherein when a word associated with the meaning text in the keyword information is included in the sentence data, one or more keywords are extracted from the meaning text corresponding to the word.

3. The extraction means Acquire keyword information indicating a replacement word corresponding to one or more of the words to be extracted as keywords; The image generating device according to claim 1 , wherein when a word associated with the replacement word in the keyword information is included in the text data, the replacement word corresponding to that word is extracted as the keyword.

4. The generating means determining a weight for each of the keywords based on the characteristics of each of the keywords; The image generation device according to claim 1 , wherein each of the keywords and a weight of each of the keywords are input to the image generation model.

5. The image generating device according to claim 4 , wherein said generating means determines a weight for each of said keywords based on the number of times each of said keywords appears in said text data.

6. The image generating device according to claim 4 , wherein the generating means increases the weight of the keyword when the keyword is not a polysemous word and is a word related to a specific topic.

7. The image generating device according to claim 4 , wherein said generating means reduces the weight of said keyword when said keyword is a polysemous word and is not a word related to a specific topic.

8. an acquisition step of acquiring sentence data; an extraction step of extracting a plurality of keywords from the text data; and generating an image related to the text data by inputting the plurality of keywords into an image generation model.

9. In the extraction step, Acquire keyword information indicating a semantic text representing the meaning of one or more of the words to be extracted as keywords; The image generating method according to claim 8 , wherein, when a word associated with the meaning text in the keyword information is included in the sentence data, one or more keywords are extracted from the meaning text corresponding to the word.

10. an acquisition step of acquiring sentence data; an extraction step of extracting a plurality of keywords from the text data; a generation step of generating an image related to the text data by inputting the plurality of keywords into an image generation model.

Images