Method, system and non-transitory computer-readable recording medium for recommending font for user context based on artificial intelligence

Abstract

According to one aspect of the present invention, a method for recommending a font suitable for a user situation based on artificial intelligence is provided, comprising the steps of: extracting user environment keywords from text input by a user; and determining a font to be recommended to the user by comparing a first font feature vector output by inputting a vector regarding the user environment keywords into a pre-trained artificial intelligence-based font recommendation model with a second font feature vector regarding a plurality of fonts. The method is provided such that the artificial intelligence-based font recommendation model is pre-trained using a vector regarding keywords for each user environment and a font feature vector labeled for the keywords for each user environment.

Description

Technology Field

[0001] The present invention relates to a method, a system, and a non-transient computer-readable recording medium for recommending a font suitable for a user's situation using artificial intelligence. Background Technology

[0002] As the proportion of text-based communication via digital devices continues to increase, fonts are establishing themselves as important tools for conveying emotions and situations, going beyond mere means of text expression. In particular, the widespread use of social media platforms such as YouTube, Naver Blog, and Instagram, as well as various web-based content creation tools, is further highlighting the importance of font selection. On these platforms, fonts function as a core element that enhances not only the aesthetic aspects of content but also message delivery and user experience.

[0003] For example, using appropriate fonts in YouTube vlog subtitles to reflect specific weather or emotional states can provide viewers with a more immersive experience. As such, the importance of font selection is increasing to enhance the emotional delivery of digital content.

[0004] However, existing font classification methods require a significant amount of time and effort from users to find fonts suitable for specific situations or emotions. For example, when a user wants to express a specific emotion such as "gloomy weather," the conventional method requires them to visually browse through fonts to find the appropriate one. This leads to a lack of intuitiveness and efficiency in the font selection process.

[0005] Accordingly, the inventor(s) intend to propose a novel and advanced technology that utilizes artificial intelligence to recommend fonts suitable for a user's emotions and situation. Prior art literature

[0006] Published Patent Application No. 10-2024-0051898 (April 22, 2024) Registered Patent Application No. 10-2624095 (January 8, 2024) The problem to be solved

[0007] The present invention aims to solve all the problems of the aforementioned prior art.

[0008] In addition, the present invention has another objective of reducing the time and effort consumed in the font selection process by enabling users to intuitively and efficiently select their desired fonts through a font recommendation technology based on emotions and situations.

[0009] In addition, another objective of the present invention is to reflect user preferences in real time by utilizing AI-based reinforcement learning and user feedback, thereby providing a progressively improved personalized font recommendation experience.

[0010] In addition, another objective of the present invention is to innovatively improve the quality of visual communication by learning the systematic association between situations, emotions, and fonts, thereby maximizing the emotional delivery of text messages in various situations. means of solving the problem

[0011] A representative configuration of the present invention for achieving the above objective is as follows.

[0012] According to one aspect of the present invention, a method for recommending a font suitable for a user situation based on artificial intelligence is provided, comprising the steps of: extracting user environment keywords from text input by a user; and determining a font to be recommended to the user by comparing a first font feature vector output by inputting a vector regarding the user environment keywords into a pre-trained artificial intelligence-based font recommendation model with a second font feature vector regarding a plurality of fonts. The method is provided such that the artificial intelligence-based font recommendation model is pre-trained using a vector regarding keywords for each user environment and a font feature vector labeled for the keywords for each user environment.

[0013] According to another aspect of the present invention, a system for recommending a font suitable for a user situation based on artificial intelligence is provided, comprising: a user environment keyword extraction unit that extracts user environment keywords from text input by a user; and a recommended font determination unit that determines a font to be recommended to the user by inputting a vector regarding the user environment keywords into a pre-trained artificial intelligence-based font recommendation model and comparing a first font feature vector output with a second font feature vector regarding a plurality of fonts. The system is provided such that the artificial intelligence-based font recommendation model is pre-trained using a vector regarding keywords for each user environment and a font feature vector labeled for the keywords for each user environment.

[0014] In addition to this, other methods for implementing the present invention, other systems, and non-transient computer-readable recording media on which a computer program for executing said methods is recorded are further provided. Effects of the invention

[0015] According to the present invention, by using a font recommendation technology based on emotion and situation, users can intuitively and efficiently select the font they want, thereby reducing the time and effort consumed in the font selection process.

[0016] In addition, according to the present invention, user preferences can be reflected in real time by utilizing artificial intelligence-based reinforcement learning and user feedback, thereby providing a progressively improved customized font recommendation experience.

[0017] In addition, according to the present invention, the quality of visual communication can be innovatively improved by learning the systematic association between situations, emotions, and fonts to maximize the emotional delivery of text messages in various situations. Brief explanation of the drawing

[0018] FIG. 1 is a diagram showing the schematic configuration of an overall system for recommending a font suitable for a user's situation using artificial intelligence according to one embodiment of the present invention. FIG. 2 is a drawing illustrating in detail the internal configuration of a font recommendation system according to one embodiment of the present invention. FIG. 3 is a diagram exemplarily illustrating the process of recommending a font according to one embodiment of the present invention. Specific details for implementing the invention

[0019] The following detailed description of the invention refers to the accompanying drawings, which illustrate specific embodiments in which the invention may be practiced. These embodiments are described in sufficient detail to enable those skilled in the art to practice the invention. It should be understood that various embodiments of the invention are different but need not be mutually exclusive. For example, specific shapes, structures, and characteristics described herein may be modified from one embodiment to another without departing from the spirit and scope of the invention. It should also be understood that the location or arrangement of individual components within each embodiment may be modified without departing from the spirit and scope of the invention. Accordingly, the following detailed description is not meant to be limiting, and the scope of the invention should be understood to encompass the scope claimed by the claims and all equivalents thereof. Similar reference numerals in the drawings indicate identical or similar components across various aspects.

[0020] Hereinafter, in order to enable a person skilled in the art to easily practice the present invention, various preferred embodiments of the present invention will be described in detail with reference to the attached drawings.

[0021] Configuration of the entire system

[0022] FIG. 1 is a diagram showing the schematic configuration of an overall system for recommending a font suitable for a user's situation using artificial intelligence according to one embodiment of the present invention.

[0023] As illustrated in FIG. 1, the entire system according to one embodiment of the present invention may include a communication network (100), a font recommendation system (200), and a user device (300).

[0024] First, a communication network (100) according to one embodiment of the present invention can be configured regardless of the mode of communication, such as wired communication or wireless communication, and can be configured as various communication networks such as a Local Area Network (LAN), a Metropolitan Area Network (MAN), or a Wide Area Network (WAN). Preferably, the communication network (100) referred to in this specification may be the known Internet or the World Wide Web (WWW). However, the communication network (100) may include at least a known wired / wireless data communication network, a known telephone network, or a known wired / wireless television communication network, without being limited thereto.

[0025] For example, the communication network (100) may be a wireless data communication network and may implement conventional communication methods such as WiFi communication, WiFi-Direct communication, Long Term Evolution (LTE) communication, 5G communication, Bluetooth communication (including Bluetooth Low Energy (BLE) communication), infrared communication, ultrasonic communication, etc., in at least a part thereof.

[0026] Next, a font recommendation system (200) according to one embodiment of the present invention can communicate with a user device (300) through a communication network (100), extract user environment keywords from text input by the user, and input a vector regarding the user environment keywords into a pre-trained artificial intelligence-based font recommendation model to output a first font feature vector, which is then compared with a second font feature vector regarding a plurality of fonts to determine a font to recommend to the user. In addition, this artificial intelligence-based font recommendation model can be pre-trained using a vector regarding keywords for each user environment and a font feature vector labeled for keywords for each user environment. Meanwhile, this font recommendation system (200) may be a system that runs on a server equipped with memory means and equipped with a microprocessor to have computational capabilities.

[0027] The configuration and functions of the font recommendation system (200) according to the present invention will be examined in detail through the following detailed description.

[0028] Next, a user device (300) according to one embodiment of the present invention is a digital device that includes a function to communicate after connecting to a font recommendation system (200). Any digital device equipped with memory means and equipped with a microprocessor to have computational capabilities, such as a smartphone, a video recording device, a tablet, a smart watch, smart glasses, a desktop computer, a laptop computer, a workstation, etc., can be adopted as the user device (300) according to the present invention. Here, the user device (300) according to one embodiment of the present invention may include means for receiving text input from a user (e.g., a keyboard), a display means for providing recommended fonts (e.g., an OLED, an LCD), etc.

[0029] Meanwhile, the user device (300) may further include an application program for performing functions according to the present invention. Such an application may exist in the form of a program module within the user device (300). The nature of such a program module may be generally similar to the components of the font recommendation system (200) described below (i.e., user environment keyword extraction unit (210), recommended font determination unit (220), communication unit (230), and control unit (240)). Here, at least a part of the application may be replaced with a hardware device or firmware device capable of performing substantially the same or equivalent functions as needed.

[0030] Composition of the font recommendation system

[0031] Below, we will examine the internal configuration of the font recommendation system (200) that performs important functions for the implementation of the present invention and the functions of each component.

[0032] FIG. 2 is a drawing illustrating in detail the internal configuration of a font recommendation system (200) according to one embodiment of the present invention.

[0033] As illustrated in FIG. 2, according to one embodiment of the present invention, a font recommendation system (200) may include a user environment keyword extraction unit (210), a recommended font determination unit (220), a communication unit (230), and a control unit (240). According to one embodiment of the present invention, the user environment keyword extraction unit (210), the recommended font determination unit (220), the communication unit (230), and the control unit (240) of the font recommendation system (200) may be program modules, at least some of which communicate with an external system (not shown). Such program modules may be included in the font recommendation system (200) in the form of an operating system, an application program module, and other program modules, and may be physically stored in various known storage devices. Additionally, such program modules may be stored in a remote storage device capable of communicating with the font recommendation system (200). Meanwhile, such program modules encompass, but are not limited to, routines, subroutines, programs, objects, components, data structures, etc., that perform specific tasks or execute specific abstract data types as described below according to the present invention.

[0034] Meanwhile, although the font recommendation system (200) has been described as above, this description is exemplary, and it is obvious to those skilled in the art that at least some of the components or functions of the font recommendation system (200) may be implemented within a user device (300) or a server (not shown) or included within an external system (not shown) as needed.

[0035] First, a user environment keyword extraction unit (210) according to one embodiment of the present invention can perform the function of extracting user environment keywords from text input by a user. The user environment keywords according to one embodiment of the present invention may include keywords related to external factors (i.e., situations) of the user or keywords related to internal factors (i.e., emotions).

[0036] Specifically, the user environment keyword extraction unit (210) can tokenize text input from a user using a BertTokenizer and perform preprocessing such as padding and truncation in a format suitable for a KoBERT model. Then, the user environment keyword extraction unit (210) can input the above preprocessed text into a KoBERT model (the model may be fine-tuned using response data for user environments (e.g., emotions, situations, etc.) for multiple fonts obtained through a crowdsourcing-based survey so that classification of user environments (e.g., emotions or situations) from the text can be provided as a result, and in the output layer, core keywords regarding the user environment can be extracted using probability values ​​based on a SoftMax function) and extract user environment keywords corresponding to the text by referring to the output result.

[0037] Meanwhile, during the process of analyzing the above text, important (or core) keywords (or user environment keywords) in the text may be identified by utilizing importance analysis techniques or algorithms such as KeyBERT, TextRank, and TF-IDF.

[0038] Next, a recommended font determination unit (220) according to one embodiment of the present invention can perform the function of determining a font to be recommended to a user by inputting a vector regarding user environment keywords (e.g., each keyword can be converted into a unique vector and an embedding technique can be applied) into a pre-trained artificial intelligence-based font recommendation model and comparing the output first font feature vector with a second font feature vector regarding a plurality of fonts.

[0039] For example, the recommended font determination unit (220) may input a first vector corresponding to a user environment keyword extracted by the user environment keyword extraction unit (210) into a pre-trained XGBoost model, and cause a first font feature vector corresponding to the first vector to be output from the XGBoost model. Then, among the second font feature vectors for a plurality of pre-prepared fonts, the font corresponding to the font feature vector that is judged to have the highest similarity to the above first font feature vector (for example, such similarity can be calculated based on cosine similarity) may be determined as the font to be recommended to the user.

[0040] Additionally, the recommended font determination unit (220) can train the AI-based font recommendation model (e.g., the XGBoost model above) so that when a vector corresponding to a user environment keyword is input, a font feature vector is output.

[0041] Specifically, the recommended font determination unit (220) can train an AI-based font recommendation model using keywords for each user environment and font feature vectors labeled for keywords for each user environment. An XGBoost model, an Elastic Net model, etc., can be used as such an AI-based font recommendation model.

[0042] For example, the recommended font determination unit (220) may obtain a plurality of font images from at least one font providing website (or external database), convert the plurality of font images into pixels of the same size (e.g., 244 x 244 pixels), and normalize the scale of the pixel values ​​to 0 to 1 (this may be to increase the learning speed and promote the convergence of the model). Then, the recommended font determination unit (220) may input the above normalized font images into a VGG16 variant model from which three fully connected layers, which are the output layers, have been removed (e.g., a deep neural network composed of 16 weight learning layers (13 convolutional layers and 3 fully connected layers), which may be a model pre-trained with a large dataset such as ImageNet, which can learn meaningful features even in small regions within an image using a 3x3 filter, and can maximize performance even with a small dataset by utilizing transfer learning based on this) to extract font image feature vectors. Next, the recommended font determination unit (220) can train an AI-based font recommendation model using a font feature vector (i.e., dependent variable) extracted through the VGG16 model and a vector regarding keywords for each user environment (i.e., independent variable) (for example, the AI-based font recommendation model may be a regression model and relationship mapping may be performed).

[0043] Meanwhile, looking at the detailed configuration of the VGG16 model mentioned above, the model's initial convolutional layers serve to extract basic visual elements of font images. In this stage, unique morphological features such as strokes, curvature, and slant are learned hierarchically, and high-level abstracted features can be generated by progressively extracting complex features. This enables the effective recognition of the unique visual patterns inherent to each font. Additionally, the model's max pooling layers are applied in a 2x2 size to reduce the size of feature maps while emphasizing key font characteristics. This process improves computational efficiency by removing unnecessary information, while helping to preserve only the essential features of the font images. In the final stage of the model, while image classification is typically optimized through fully connected layers, the three fully connected layers in the output layer can be removed to focus on extracting font feature vectors instead of typeface classification. Specifically, the model can be modified to consist of only 13 convolutional layers to learn the visual patterns of fonts more intensively. The final output of the model is generated in the form of a 4-dimensional (4D) tensor, which can be converted into a 1-dimensional (1D) vector by flattening. This vector represents the key features of the font and can then be mapped to user environment keyword labels and utilized in the font recommendation process. Through this transformation structure, the VGG16 model can be effectively utilized to extract and analyze the visual patterns of fonts and has the advantage of being able to learn efficiently through transfer learning even with a small amount of data.

[0044] Meanwhile, performance evaluations were performed on the various models that can be used as the AI-based font recommendation models mentioned above, and the XGBoost model demonstrated the best performance. This performance evaluation was conducted based on the Root Mean Square Error (RMSE) (see Table 1).

[0045] <Performance Evaluation Results> model Performance evaluation figures XGBoost 38.57 LightGBM 48.41 DNN (Deep Neural Network) 44.66 Random Forest 45.06 Elastic Net 41.23

[0046] In other words, it was confirmed that the XGBoost model effectively performs mapping between user environment keywords and font features through a structure capable of precisely learning non-linear relationships. This result demonstrates that the XGBoost model exhibits the lowest prediction error in linking text keywords and the structural features of fonts, proving the superior accuracy of font recommendation according to the present invention.

[0047] In addition, the recommended font determination unit (220) can perform deep reinforcement learning based on feedback provided by the user regarding the recommended font.

[0048] Specifically, the recommended font determination unit (220) performs reinforcement learning by reflecting user feedback in a reward function, and may assign a high reward value when the user provides positive feedback on the recommended font and a low reward value when the user provides negative feedback. That is, the reinforcement learning agent can be trained to recommend the optimal font by reflecting user preferences.

[0049] Meanwhile, the recommended font determination unit (220) can calculate a recommendation suitability index for each font (or recommended font) based on feedback provided by the user regarding the recommended font, and can determine the font to recommend to the user by further referring to this suitability index.

[0050] Next, the communication unit (230) according to one embodiment of the present invention can perform the function of enabling data transmission and reception from / to the user environment keyword extraction unit (210) and the recommended font determination unit (220).

[0051] Finally, a control unit (240) according to one embodiment of the present invention can perform the function of controlling the flow of data between the user environment keyword extraction unit (210), the recommended font determination unit (220), and the communication unit (230). That is, by controlling the flow of data from / to the outside of the font recommendation system (200) or the flow of data between each component of the font recommendation system (200), the control unit (240) according to the present invention can control the user environment keyword extraction unit (210), the recommended font determination unit (220), and the communication unit (230) to perform their respective unique functions.

[0052] FIG. 3 is a diagram exemplarily illustrating the process of recommending a font according to one embodiment of the present invention.

[0053] Referring to FIG. 3, first, according to one embodiment of the present invention, response data regarding user environments (e.g., emotions, situations, etc.) for a plurality of fonts obtained through a crowdsourcing-based survey is obtained, and based on this, keyword vectors for user environments regarding fonts and user environments (e.g., Nanum Gothic -> [calmness 0.8, official announcement 0.9]) can be generated.

[0054] Next, according to one embodiment of the present invention, a font feature vector can be extracted by utilizing a VGG16 model in which three fully connected layers, which are the output layer, have been removed.

[0055] Next, according to one embodiment of the present invention, a vector regarding user environment keywords can be labeled to a font feature vector, and an XGBoost model (i.e., an artificial intelligence-based font recommendation model) can be trained based on the data.

[0056] Then, according to one embodiment of the present invention, when user text (e.g., "report") is input through a user device (300), a user environment keyword (i.e., "report") is extracted from the text, and a vector regarding the keyword is input into the above-mentioned pre-trained XGBoost (i.e., an AI-based font recommendation model), and the output font feature vector is compared with font feature vectors regarding a plurality of fonts to determine a font to recommend to the user. At this time, similarity can be measured based on cosine similarity, for example, if the result of calculating cosine similarity is that 'Nanum Gothic' is 0.999998, 'Jeju Myeongjo' is 0.364295, and 'KCC Kim Whanki' is -0.153990, 'Hakgyo Ansim Dunggeun Miso B' is -0.239746, 'ACC Eorini Maeum Goun' is -0.315151, and 'Jeongseon Arirang' is -0.343174, then 'Nanum Gothic' can be determined as the font to be recommended to the user.

[0057] Next, according to one embodiment of the present invention, the name of the above recommended font may be provided to a user device (300).

[0058] The embodiments according to the present invention described above may be implemented in the form of program instructions that can be executed through various computer components and recorded on a computer-readable recording medium. The computer-readable recording medium may include program instructions, data files, data structures, etc., either individually or in combination. The program instructions recorded on the computer-readable recording medium may be those specifically designed and configured for the present invention or those known and available to those skilled in the art of computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes; optical recording media such as CD-ROMs and DVDs; magneto-optical media such as floptical disks; and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, and flash memory. Examples of program instructions include machine code, such as that generated by a compiler, as well as high-level language code that can be executed by a computer using an interpreter, etc. Hardware devices may be modified into one or more software modules to perform processing according to the present invention, and vice versa.

[0059] Although the present invention has been described above with reference to specific details such as specific components, limited embodiments, and drawings, this is provided only to aid in a more comprehensive understanding of the invention, and the invention is not limited to the above embodiments, and a person skilled in the art to which the invention belongs can make various modifications and changes from this description.

[0060] Accordingly, the scope of the present invention should not be limited to the embodiments described above, and all scopes equivalent to or equivalently modified from the claims set forth below, as well as the claims set forth below, shall be considered to fall within the scope of the concept of the present invention. Explanation of the symbols

[0061] 100: Communication network 200: Font Recommendation System 210: User Environment Keyword Extraction Unit 220: Recommended Font Decision Section 230: Communications Department 240: Control unit 300: User device

Claims

Claim 1 A method for recommending a font suitable for a user's situation based on artificial intelligence, comprising the steps of: extracting user environment keywords from text input by a user; and determining a font to recommend to the user by comparing a first font feature vector, which is output by inputting a vector regarding the user environment keywords into a pre-trained artificial intelligence-based font recommendation model, with a second font feature vector regarding a plurality of fonts. The artificial intelligence-based font recommendation model is pre-trained using a vector regarding keywords for each user environment and a font feature vector labeled for the keywords for each user environment. Claim 2 A method for tokenizing the input text and extracting the user environment keywords using a KoBERT model in claim 1. Claim 3 In paragraph 2, the above artificial intelligence-based font recommendation model is configured based on the XGBoost model. Claim 4 In paragraph 3, the first font feature vector and the second font feature vector are determined by referring to the result of inputting a plurality of font images, composed of pixels of the same size and having normalized pixel values, into a VGG16 model from which the output layer has been removed. Claim 5 In paragraph 4, a method for determining the font of the font feature vector with the highest similarity to the first font feature vector as the font to be recommended to the user. Claim 6 A method according to claim 1, wherein reinforcement learning is performed by reflecting the user's feedback on the recommended font in a reward function, wherein a high reward value is assigned when the user provides positive feedback on the recommended font, and a low reward value is assigned when the user provides negative feedback. Claim 7 A non-transient computer-readable recording medium for recording a computer program for executing the method according to paragraph 1. Claim 8 A system for recommending fonts suitable for a user's situation based on artificial intelligence, comprising: a user environment keyword extraction unit that extracts user environment keywords from text input by a user; and a recommended font determination unit that determines a font to be recommended to the user by inputting a vector regarding the user environment keywords into a pre-trained artificial intelligence-based font recommendation model and comparing a first font feature vector output with a second font feature vector regarding a plurality of fonts. The system wherein the artificial intelligence-based font recommendation model is pre-trained using a vector regarding keywords for each user environment and a font feature vector labeled for the keywords for each user environment.

Images