Modular smart bookshelf system having artificial intelligence librarian and control method thereof

Abstract

An embodiment of the present invention may include a smart shelving system comprising: a main kiosk that performs user recognition, information output, and shelving control; and a shelving module that is physically and electrically coupled to at least one side of the main kiosk and stores at least one book; wherein the main kiosk includes an identification unit that recognizes the face of an approaching user to obtain user account information; a display unit that outputs information; and a control unit that executes at least one operation mode among a book guidance mode, a usage guidance mode, a facility guidance mode, and other modes; wherein the shelving module includes an RFID reader unit installed in each storage compartment (Shelf) to detect the status of book entry and exit in real time; and wherein the control unit, when a scan signal of a specific book is detected from the RFID reader unit, analyzes the metadata and original text of the book to generate summary information, converts the summary information into personalized information based on age or interest data included in the acquired user account information, and controls the output of the converted personalized information through the display unit.

Description

Technology Field

[0001] A smart bookshelf system according to an embodiment of the present invention relates to a kiosk and a bookshelf module device equipped with a function for providing customized content based on user recognition and book information. Background Technology

[0002] Libraries have long contributed to the development of human civilization as repositories of knowledge and information. Modern library systems have enhanced the efficiency of book search, lending, and return management by incorporating information technology (IT). Users search for materials in collections through keyword-based search systems (OPAC), and some systems even provide simple book recommendation services based on past borrowing history.

[0003] Recently, attempts have been made to introduce robotic technology or artificial intelligence to automate library operations. For example, Korean Registered Patent No. 10-2748298 (hereinafter referred to as the "Prior Art Document") discloses a system that supports book return and shelving tasks using autonomous robots. The artificial intelligence described in this document focuses primarily on improving the physical operational efficiency of the library, such as by learning obstacles to optimize the robot's driving path or determining the necessity of shelving based on the weight or quantity of returned books.

[0004] However, these conventional technologies and prior art literature fail to understand users' complex and abstract information needs and cannot provide truly personalized and inclusive services. Existing systems rely merely on structured data, such as past borrowing lists, and fail to grasp users' deep interests and preferences contained in unstructured data, such as the content of book reviews, search patterns for specific topics, or notes and highlights left while reading. Furthermore, there has been a lack of customized interfaces that consider the accessibility of diverse user groups, including the elderly, children, foreigners, and people with disabilities.

[0005] Therefore, there is an urgent need to develop a new paradigm of AI-based library systems that go beyond physical automation to intellectually engage with users and satisfy their latent information needs. The problem to be solved

[0006] The present invention aims to solve the problems of the aforementioned prior art. The objective of the present invention is to provide a smart bookshelf system and a method of operation thereof that provide book metadata analysis and user-customized content based on user face recognition and RFID book tracking functions. means of solving the problem

[0007] According to an embodiment of the present invention, a smart shelving system may be included, comprising: a main kiosk that performs user recognition, information output, and shelving control; and a shelving module that is physically and electrically coupled to at least one side of the main kiosk and stores at least one book. The main kiosk includes an identification unit that recognizes the face of an approaching user to obtain user account information; a display unit that outputs information; and a control unit that executes at least one operation mode among a book guidance mode, a usage guidance mode, a facility guidance mode, and other modes. The shelving module includes an RFID reader unit installed in each storage compartment (Shelf) to detect the status of book entry and exit in real time. The control unit, when a scan signal of a specific book is detected from the RFID reader unit, analyzes the metadata and original text of the book to generate summary information, converts the summary information into personalized information based on age or interest data included in the acquired user account information, and controls the output of the converted personalized information through the display unit.

[0008] Additionally, the smart bookshelf system may include a control unit that, when a withdrawal signal indicating that a specific book has been removed from the bookshelf is detected from the RFID sensing unit, analyzes the metadata and original text of the specific book to generate the summary information, converts the vocabulary difficulty and explanatory tone of the summary information based on age or interest data included in the previously acquired user account information, and controls a virtual digital human avatar displayed on the display unit to utter the converted summary information to the user in the explanatory tone.

[0009] Additionally, the smart shelving system may include a main connector for power supply and data communication on the side of the main kiosk, a shelving connector coupled to the main connector on the side of the shelving module, and the control unit automatically updates the book location map of the entire shelving system as the number of connections or arrangement of the shelving modules changes.

[0010] Additionally, the above-mentioned bookshelf module may include a smart bookshelf system characterized by comprising at least one of a general shelf on which books are placed, a monitor for displaying digital information, a display shelf for placing books facing forward, a lower cabinet for storing items, a return monitor for processing returns, and a return stand.

[0011] Additionally, the above-described bookshelf module further includes an LED indicator positioned on the front of each storage compartment, and the control unit may include a smart bookshelf system characterized by illuminating the LED indicator of the storage compartment where the book is located to visually indicate the physical location when a digital human avatar recommends a specific book or guides the user to other books related to a book retrieved, and highlighting and displaying the book information through augmented reality (AR) or visual graphics via the display unit.

[0012] Additionally, the control unit may include a smart bookshelf system characterized by, when a user's query is input, calculating a weight for each operation mode based on a priority bonus point application rule and executing an optimized mode, wherein the priority bonus point application rule applies bonus points to the user guide mode if keywords related to events and announcements are included in the user's query, applies bonus points to the facility guide mode if a location entity is included, applies bonus points to the book guide mode if a cover recognition or ISBN input signal is detected, and applies bonus points to other modes if the current time is within the survey campaign time period, thereby automatically entering the mode that obtains the highest score.

[0013] In addition, the control unit may include a smart bookshelf system characterized by assigning weights to user age information, interest field information, and past query history information based on the user account information, and generating a digital human avatar of a type determined through the following <Mathematical Formula 1> and displaying it on the display unit.

[0014] <Mathematical Formula 1>

[0015]

[0016]

[0017] In addition, the control unit according to an embodiment of the present invention may include a smart bookshelf system characterized by updating user account information so that appearance items of the digital human avatar are added or the growth stage of the avatar changes as the user's cumulative number of loans increases.

[0018] Additionally, the control unit may include a smart bookshelf system characterized by recommending books by generating a primary set of book candidates based on the growth level of the user's virtual avatar when the book guidance mode is executed, statistically analyzing the user's borrowing history by category to calculate frequency, recency, and dwell time indicators, refining items within the primary set of book candidates by assigning weights to them, and determining a final ranking that adjusts the difficulty level of the recommended books by applying the user's age information as a final filter to the refined list.

[0019] Additionally, the control unit may include a smart bookshelf system characterized by querying pre-stored recommended book list data from an external organization, verifying whether a book included in the final ranking is included in the recommended book list of the external organization, and guiding the user to the verification result and the basis for the recommendation through the speech of the digital human avatar. Effects of the invention

[0020] According to various embodiments of the present invention, a library system using an artificial intelligence librarian can provide a friendly and intuitive experience to users as if they were conversing with a real librarian through conversational services with a digital human avatar.

[0021] The smart bookshelf system of the present invention recognizes the retrieval status of books in real time and enables the provision of customized information based on user age and interests, thereby improving information search efficiency and user immersion. In addition, it can enhance the user experience through digital human avatars and augmented reality-based guidance functions.

[0022] According to an embodiment of the present invention, by combining book lending / returning (function) and gallery / exhibition (culture) into a single modular system (Nexus), the library can be transformed from a simple materials storage facility into a complex cultural space. Brief explanation of the drawing

[0023] Figure 1 is a diagram showing the overall configuration of a smart shelving system. Figure 2 is a diagram showing the configuration of the main kiosk. FIG. 3 is a diagram showing the operation sequence of a smart shelving system according to an embodiment of the present invention. FIG. 4 is a diagram showing an example of user interaction with a main kiosk according to an embodiment of the present invention. FIG. 5 is a diagram illustrating the determination of the operation mode and detailed execution process of a control unit according to an embodiment of the present invention. Specific details for implementing the invention

[0024] Specific structural or functional descriptions regarding embodiments according to the concept of the present invention disclosed herein are provided merely for the purpose of explaining embodiments according to the concept of the present invention, and embodiments according to the concept of the present invention may be implemented in various forms and are not limited to the embodiments described herein.

[0025] Embodiments according to the concept of the present invention may be subject to various modifications and may take various forms; therefore, embodiments are illustrated in the drawings and described in detail in this specification. However, this is not intended to limit the embodiments according to the concept of the present invention to specific disclosed forms, and includes all modifications, equivalents, or substitutions that fall within the spirit and scope of the present invention.

[0026] Terms such as "first" or "second" may be used to describe various components, but said components should not be limited by said terms. For the sole purpose of distinguishing one component from another, for example, without departing from the scope of rights according to the concept of the present invention, the first component may be named the second component, and similarly, the second component may be named the first component.

[0027] The terms used herein are used merely to describe specific embodiments and are not intended to limit the invention. Singular expressions include plural expressions unless the context clearly indicates otherwise. In this specification, terms such as “comprising” or “having” are intended to indicate the existence of the described features, numbers, steps, actions, components, parts, or combinations thereof, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0028] Additionally, the term “and / or” as used herein should be understood to refer to and include all possible combinations of one or more of the enumerated related items. Furthermore, the term “or” is intended to mean an implied “or” rather than an exclusive “or.” That is, unless otherwise specified or clear from the context, “X uses A or B” may apply to cases where X uses A; X uses B; or X uses both A and B. Additionally, “at least one selected from A and B” may refer to (1) A, (2) at least one of A, (3) B, (4) at least one of B, (5) at least one of A and at least one of B, (6) at least one of A and B, (7) at least one of B and A, and (8) all of A and B.

[0029] Before providing a detailed description of the drawings, it is to clarify that the classification of components in this specification is merely based on the primary function each component is responsible for. That is, two or more components described below may be combined into a single component, or a single component may be divided into two or more components based on more subdivided functions. Furthermore, each component described below may additionally perform some or all of the functions of other components in addition to its own primary function, and it goes without saying that some of the primary functions of each component may be exclusively performed by other components.

[0030] Furthermore, in performing the method or operation method, each process constituting the method may occur differently from the specified order unless a specific order is clearly indicated in the context. That is, each process may occur in the same order as specified, may be performed substantially simultaneously, or may be performed in the reverse order.

[0032] Hereinafter, the present invention will be described in detail by explaining preferred embodiments of the present invention with reference to the attached drawings.

[0033] In the following, the term "user" refers to a user who enters the library and uses all services provided by the library, and the term "user" may refer to a person who directly uses and operates the main kiosk of this specification.

[0035] FIG. 1 is a conceptual diagram showing the overall configuration of a modular smart bookshelf system equipped with an artificial intelligence librarian according to one embodiment of the present invention.

[0036] Referring to FIG. 1, the smart shelving system (1) of the present specification includes a smart shelving device (20), a user terminal (30), and a server (10).

[0037] According to an embodiment of the present invention, a smart bookshelf device (20) may include a main kiosk (200) through which a user can actively search for desired information via a voice, video, and touch-based AI guidance interface, and a bookshelf module (400) coupled to at least one side of the main kiosk (200) for storing books.

[0038] According to an embodiment of the present invention, a server (10) is a computer system that stores and manages data through a network or transmits data to a client, such as a main kiosk (200) and a user terminal (30), to provide various services. The server (10) may include various types of servers, such as a web server, an application server, a database server, a file server, and a mail server.

[0039] In various embodiments, as illustrated in FIG. 1, the server (10) is depicted as communicating via a wireless network outside the main kiosk (200), but is not limited thereto and may be implemented as an integrated server operating as an internal server provided inside the main kiosk (20). When the server (10) operates as an internal server, the internal server may communicate with an external server, such as an external library bibliographic information DB or a public data server, and transmit and receive related data.

[0040] According to an embodiment of the present invention, the main kiosk (200) communicates with the server (10) via a wired / wireless network and can perform intelligent interaction with the user through display output or a digital human avatar. In addition, the main kiosk (200) can monitor and control the status of the physically connected bookshelf module (400) in real time (whether books have been withdrawn, module connection status, etc.).

[0041] Specifically, the main kiosk (200) communicates with the server (10) and visually provides the user with a digital human avatar related to specific content (e.g., summary information of a withdrawn book). By interacting with the digital human avatar during library use, the user can receive book recommendations or be guided to the location of bookshelves, and can grow the digital human avatar or change its appearance according to the user's usage history.

[0042] The bookshelf module (400) can be combined in a modular form on the left or right side of the main kiosk (200) and stores multiple books. The bookshelf module (400) detects the status of book entry and exit through an RFID reader provided on each shelf and transmits this to the main kiosk (200) to help ensure that a related guidance service is executed immediately when a user takes out a book.

[0043] In addition, the specific content provided by the main kiosk (200) may include various modes for the convenience of library users. For example, the modes may include a user guide mode, a facility guide mode, a smart book recommendation mode, and other modes. The main kiosk (200) may change the mode or the speech style and appearance of the digital human avatar based on the user's account information or type (age, interests, etc.). This is explained in detail in FIGS. 3 and 4.

[0044] The user terminal (30) can communicate with the server (10) or the main kiosk (200) and is a terminal used by the library user, and may be a mobile terminal device that moves with the user, such as a mobile phone, smartwatch, or tablet computer. The user terminal (300) can be used to receive AR directions through linkage with the main kiosk (200) or to store a personalized book recommendation list.

[0045] Below, the detailed configurations of the server (10), main kiosk (200), bookshelf module (400), and user terminal (30) will be described with reference to the drawings.

[0046] FIG. 2 is an illustrative diagram for explaining the detailed types and configurations of a bookshelf module (400) according to an embodiment of the present invention.

[0047] Referring to FIG. 2, the bookshelf module (400) of the present invention may include a plurality of units that can be selectively combined according to the installation environment of the library, the size of the space, and the type of collection. Specifically, the bookshelf module (400) may be configured to include at least one of a general shelf unit (41), a new book display monitor unit (42), a display shelf and lower cabinet unit (43), and a return monitor and return stand unit (44).

[0048] The specific characteristics of each unit are as follows.

[0049] The general shelf unit (41) is the most basic type of bookshelf module that includes a multi-layer horizontal shelf and stores a large number of books by spine-out. An RFID reader is built into the bottom surface or side of each layer to detect identification information of the stored books in real time.

[0050] The display monitor unit (42) is a module in the form of digital signage equipped with a large display panel on the front. Normally, the unit provides an aesthetic effect by displaying the cover of a new book or a promotional video, and when a user approaches it, it can function as an information delivery medium that displays a book search screen or recommendation information.

[0051] The display shelf and lower cabinet unit (43) includes a magazine rack-type shelf on the upper side that can be mounted at an angle so that the cover of a book faces forward (face-out), and a cabinet with a door on the lower side that can store items other than books or stock books. This is advantageous for displaying materials where cover exposure is important, such as magazines or picture books.

[0052] The return monitor and return stand unit (44) is a composite module having a small display or tablet mounted on the top and an open space in the middle or bottom for loading returned books. This allows the user to immediately process book returns at the module without visiting a separate desk.

[0053] In addition, the above-mentioned return monitor and return unit (44) can implement a smart return system in which return processing is performed in real time as soon as the user inserts the book.

[0054] Specifically, when a user inserts a book to be returned into the storage space of the unit (44), the built-in RFID reader recognizes the tag and immediately changes the status in the lending system to 'Returned', and displays a return completion message and a book list in real time on the top monitor.

[0055] According to an embodiment of the present invention, the control unit generates a mobile-linked return confirmation QR code and displays it on a monitor instead of printing a paper receipt or a paper receipt. When a user scans the corresponding QR code with their user terminal (300), the return confirmation is transmitted and stored in the form of a digital image or an in-app notification. This provides an eco-friendly service that reduces unnecessary paper waste and allows the user to conveniently manage their return history via mobile.

[0056] One of the key features of the present invention is modularity. Each of the units (41, 42, 43, 44) described above is physically attached to the left or right side of the main kiosk (200) so as to be detachable. At this time, a connector (not shown) for power supply and data communication is provided on the side of each unit, so that when a user simply connects the desired shelving module (400) to the main kiosk (200) as if assembling Lego blocks, the control unit automatically recognizes the new module and updates the collection map of the entire system.

[0057] Furthermore, according to various embodiments of the present invention, the display monitor unit (42) shown in FIG. 2 can itself perform the function of the main kiosk (200).

[0058] That is, in addition to a configuration in which a separate, independent main kiosk (200) hardware is placed in the center, it is also possible to configure the unit (42) as the main control device by embedding a camera (identification unit), a microphone, a speaker, and a processor (control unit) for controlling the entire system inside the display monitor unit (42) for user recognition.

[0059] In this case, the system can be expanded in such a way that general shelf units (41) or display shelf units (43) are connected to the left and right sides of the new book display monitor unit (42), thereby maximizing space utilization and reducing hardware construction costs.

[0060] Meanwhile, the smart bookshelf system of the present invention can operate in an idle state where user access is not detected, or in 'digital gallery mode' according to a separate setting. The control unit sequentially outputs high-definition images of famous paintings, media art, or rare materials held by the library through the large display unit of the main kiosk and the monitor included in the bookshelf module.

[0061] Going beyond a simple information delivery device, this provides an interior effect that transforms the library space itself into an art exhibition space. As implied by the system's name, 'Nexus,' this invention functions as a complex cultural space connecting 'Books,' 'Art,' and 'Users.'

[0062] FIG. 3 is a flowchart illustrating the operation method of a smart shelving system according to an embodiment of the present invention.

[0063] Referring to FIG. 3, the smart bookshelf system of the present invention performs a process of providing user-customized knowledge curation beyond simple book storage through organic interaction between a main kiosk and a bookshelf module coupled thereto. The process may be performed under the control of a control unit, including steps S310 to S350.

[0064] First, a step is performed in which the main kiosk recognizes the user approaching and obtains user account information (S310).

[0065] The identification unit of the main kiosk recognizes the face of a user approaching through a camera or performs a login through the user's mobile terminal tag, etc. The control unit can obtain and load user account information including age, occupation, interest data (e.g., a 10-year-old child who likes dinosaurs, an expert in the field of artificial intelligence, etc.) and past loan history from the database of the identified user.

[0066] According to an embodiment of the present invention, the RFID reader of the bookshelf module detects a scan signal of a specific book (S320).

[0067] The above shelving module is physically and electrically coupled to at least one side of the main kiosk and may include an RFID reader installed in each storage compartment (Shelf).

[0068] When a user picks up a specific book among the books stored in the bookshelf module, the RFID reader detects changes as the book's tag moves away from the antenna range or moves, generates a withdrawal signal (scan signal) indicating that the book has been removed from the bookshelf, and transmits it to the control unit.

[0069] This enables the implementation of a 'Pick & Play' method, allowing users to begin interacting with the system through the natural act of picking up a book without going through a separate search process.

[0070] In addition, according to another embodiment of the present invention, the step S320 may further include a 'tagging' operation in which the user brings a book from another location that is not stored in the corresponding bookshelf module and actively recognizes it.

[0071] For example, after a user discovers an interesting book in another section or on another shelf of the library and retrieves it, they can bring the book close to or place it near an empty space in the shelving module where the user is currently located or near a shelf equipped with an RFID reader.

[0072] At this time, the RFID reader detects that a new book tag, which was not previously detected, has entered the antenna range, interprets this as an active information request signal from the user, generates a scan signal, and transmits it to the control unit. Even if the book in question is not originally stored in this bookshelf module, the control unit determines it to be a book of interest currently held by the user and performs the summary and customized guidance process following step S330 in the same manner.

[0073] Through this, users can instantly receive book description services from the smart bookshelf simply by intuitively placing the book against any smart bookshelf module in front of them, without the need to carry heavy books to a distant search station or take out a smartphone to search.

[0074] Furthermore, the present invention provides in-depth analysis information regarding not only the action of a user taking a book out of a bookshelf but also the action of placing a book in a specific zone. A 'Book Reading Zone' for book recognition may be provided on the front of the bookshelf module or the main kiosk. When a user places a book in the corresponding zone while contemplating whether to take a book out of the bookshelf and read it, an RFID reader detects this and sends a signal to the control unit.

[0075] In this case, the control unit goes beyond simple book introductions and displays information such as the 'book difficulty index,' 'estimated time to finish reading,' and 'key summary information' calculated through AI analysis on the display unit. For example, by providing analysis information such as, "This book is difficult due to the many technical terms, but the content of Chapter 3 matches your interests by 90%," it facilitates decision-making that provides practical assistance to the user in deciding whether to borrow the book. It performs decision support functions.

[0076] According to an embodiment of the present invention, the control unit can generate summary information by analyzing the metadata and original text of the specific book (S330).

[0077] The control unit described above can comprehensively analyze not only the bibliographic information (title, author, publisher) of a book, similar to existing RFID systems, but also the original text content, metadata, and external data of the book through an AI analysis module. Through this, the control unit can generate summary information in real time that includes the book's core theme, overall plot, major keywords, and difficulty level information.

[0078] Specifically, the control unit according to an embodiment of the present invention can be performed through a 'Vector Alignment-based Weighted Summary' process, which, when a scan signal of a specific book is detected from the RFID reader, analyzes the metadata and original content of the book to generate summary information, and, based on age or interest data included in the acquired user account information, mathematically calculates the correlation between the user's potential intention and the detailed content of the book to extract the most suitable core information when converting the summary information into personalized information.

[0079] Specifically, the control unit may perform the following detailed data processing steps to generate summary information.

[0080] According to an embodiment of the present invention, vector embedding and segmentation steps may be performed.

[0081] According to an embodiment of the present invention, the control unit generates a user interest vector by converting the user account information obtained in step S310 into a high-dimensional real vector through the embedding layer of a natural language processing (NLP) model. At this time, the user account information includes the user's age and occupation, as well as keywords of past borrowed books, search history, etc.

[0082] At the same time, the control unit divides the original text data of the book identified through the RFID reader into multiple segments in the form of chapters or paragraphs, and vectorizes each segment to generate multiple 'book segment vectors'.

[0083] According to an embodiment of the present invention, a similarity calculation and an attention map calculation step may be performed.

[0084] The control unit calculates the cosine similarity between the generated single user interest vector and the plurality of book segment vectors, respectively.

[0085] Through this, the control unit can generate a 'Personalized Attention Map' that assigns a high weight to segments within the entire content of the book that are close to the user's interests and mathematically close, and assigns a low weight to segments that are far away.

[0086] Finally, the step of generating weighted summary information can be performed.

[0087] The control unit selectively extracts only the key segments that have obtained a score above a preset threshold from the attention map and inputs them as the context of the generative AI model (LLM). Based on the input key segments, the AI ​​model reconstructs the sentence to finally generate personalized summary information that highlights the parts the user is most interested in (Highlight).

[0088] For example, assume that a user picked up the same book, 'Introduction to Artificial Intelligence', from the bookshelf module.

[0089] If the user's past history is focused on 'software development' and 'coding,' the control unit detects that the user's interest vector shows high similarity to segment vectors related to 'algorithm implementation' or 'Python practice' within the book. Accordingly, the generated summary information focuses on technical implementation methods.

[0090] On the other hand, if the user's history is focused on 'humanities' or 'social sciences,' the control unit assigns a high weight to segments related to 'AI ethics' or 'social impact' within the book to generate summary information described from a humanities perspective.

[0091] Unlike existing static book summarization technologies, this configuration is characterized by its technical feature of dynamically extracting and providing different 'valid information' for each user through technical means that match the user's interests with the book's details in a vector space.

[0092] According to an embodiment of the present invention, the control unit can convert summary information into personalized information based on user account information (S340).

[0093] This step is the core process of the present invention, which provides different information depending on the user profile, even for the same book.

[0094] According to an embodiment of the present invention, the control unit can reprocess the summary information generated in step S330 based on the user account information (age or interest data) obtained in step S310.

[0095] Specifically, if the user is a child, the control unit lowers the vocabulary difficulty of the summary information and converts it into an easy explanation, and if the user is an expert, it converts it to include academic expressions and in-depth summary information. In addition, the control unit sets the explanation tone according to the user's preferences; for example, it can convert the speech style to a friendly, informal tone for children and a polite, honorific tone for adults.

[0096] According to an embodiment of the present invention, a step of controlling to output personalized information converted through a display unit is performed (S350).

[0097] According to an embodiment of the present invention, the control unit visually and audibly outputs content generated through the display unit of the main kiosk. In particular, according to one embodiment of the present invention, the control unit displays a virtual digital human avatar on the display unit and controls the avatar to directly utter summary information converted into an explanatory tone and speech style set in step S340 to the user.

[0098] For example, a digital human avatar can provide a higher sense of immersion than text by explaining the contents of a book in a conversational manner using natural language, such as, "This book is a story about OO. There is a similar book on that shelf over there."

[0099] In addition, the digital human avatar can perform gestures that reflect emotions and facial expressions appropriate to the content, so as to be differentiated from expressionless text guidance.

[0100] Additionally, the control unit can provide intuitive visual guidance by simultaneously providing avatar guidance through the display unit and highlighting (lighting up, etc.) the location of other books associated with the book within the bookshelf module or displaying them as AR (augmented reality) graphics.

[0102] FIG. 4 is a diagram showing an example of interaction between a main kiosk and a user according to an embodiment of the present invention.

[0103] Referring to FIG. 4, the control unit of the present invention performs the step of visually and audibly outputting content generated through the display unit of the main kiosk (S350). In particular, according to one embodiment of the present invention, the control unit displays a virtual digital human avatar on the display unit and controls the avatar to directly utter summary information converted into a predefined descriptive tone and speech style to the user.

[0104] For example, as illustrated in FIG. 4(a), when a user first approaches, the digital human avatar can prepare for interaction by uttering a waiting phrase such as "How can I help you?". Subsequently, when the user retrieves a specific book or requests a recommendation, the control unit can generate a customized recommendation phrase by analyzing the user's account information (growth stage, reading history, age group, etc.).

[0105] Figure 4(b) is a specific example of such a customized interaction. The control unit can control the recommendation of books that fuse the two interests when it is confirmed that the user's avatar level is 'scientist' and that the user has recently read books in the field of history.

[0106] At this time, the digital human avatar can be controlled to speak in colloquial language that includes specific evidence, such as, "Your avatar has grown to the 'scientist' stage, and we have checked your recent reading history history. We recommend 'Bill Bryson's A Short History of Nearly Everything' tailored to your age group. For your reference, this book is a verified book that has also been included in the National Library of Korea's '100 Recommended Books for Young Adults'." rather than a simple list of text.

[0107] In this way, as the digital human avatar explains and recommends book content in a conversational manner using natural language, users can experience a much higher level of immersion compared to existing text-based searches. Furthermore, to differentiate itself from expressionless text guidance, the digital human avatar maximizes the effectiveness of information delivery by performing gestures (e.g., hand movements recommending a book) that reflect emotions and facial expressions appropriate to the recommendations.

[0108] Meanwhile, the smart bookshelf system of the present invention also has a differentiated feature in the hardware configuration of the bookshelf module that is physically coupled to the main kiosk.

[0109] According to an embodiment of the present invention, the bookshelf module may be composed of a combination of various units depending on the type of collection and operational purpose of the library. Specifically, the bookshelf module comprises at least one of a general shelf for storing a large number of books sideways, a monitor for displaying digital information such as recommended book information or advertisements, a display shelf mounted at an angle so that the covers of magazines or new books are visible, a lower cabinet for storing items other than books or inventory, and a return monitor and return table capable of immediately performing book return processing.

[0110] Furthermore, the smart bookshelf system of the present invention provides intuitive curation by linking the guidance of a digital human avatar with the visual display of a physical bookshelf.

[0111] Specifically, the shelving module further includes an LED indicator placed on the front of each shelf. The control unit can blink or change the color of the LED indicator of the shelf where the related book is actually placed, in accordance with the timing when a digital human avatar speaks, "A similar book is 'Sun' on that shelf over there," and guides the user to a related book. Through this, the user can immediately identify the physical location of the book while simultaneously hearing the avatar's explanation.

[0112] At the same time, the control unit can provide a three-dimensional book guidance experience that combines online and offline by highlighting the cover or location information of the book in the form of augmented reality (AR) graphics or a 3D map pop-up on the display screen of the main kiosk.

[0113] FIG. 5 is a diagram illustrating the determination of the operation mode and detailed execution process of a control unit according to an embodiment of the present invention.

[0114] Referring to FIG. 5, the control unit of the present invention executes at least one operation mode among a book guidance mode, a usage guidance mode, a facility guidance mode, and other modes, and performs an intelligent judgment logic that automatically enters an optimized mode according to the user's input situation.

[0115] First, when a user's query is input, the control unit calculates weights for each operation mode based on pre-set priority bonus point application rules, and executes the optimized mode accordingly.

[0116] The above priority bonus point application rule is a core algorithm for identifying user intent and operates as follows.

[0117] The control unit applies bonus points to the user guide mode if the user's query includes keywords related to events and announcements, such as 'event', 'announcement', etc. Additionally, if the query includes location entities such as 'reading room', 'restroom', or '3rd floor', it applies bonus points to the facility guide mode to prepare for the way guidance service.

[0118] In addition, if a book cover recognition signal is received via the camera or an ISBN input signal is detected via the barcode reader, it is determined that the user desires information about a specific book, and bonus points are applied to the book guidance mode.

[0119] If the current time falls within the survey campaign timeframe set by the library, the control unit applies bonus points to other modes to encourage user participation. By controlling the system to automatically enter the mode that ultimately obtains the highest score, the control unit enables the user to immediately access the desired service without having to manually select menus.

[0120] Meanwhile, when the above-mentioned operation mode is executed or a user logs in, the control unit generates a digital human avatar to provide a user-friendly interface and displays it on the display unit. At this time, the control unit assigns weights to user age information, interest information, and past query history information based on user account information, and generates a digital human avatar of a type determined by [Equation 1] below. Through this, an avatar with an appearance and persona that best matches the characteristics of the user, such as a child or an elderly person, can be provided.

[0121] For example, when a user approaches a main kiosk (200) placed within a library, an interaction begins. According to one embodiment of the present invention, the main kiosk (200) is normally in a user guidance mode, but can detect a user's approach through a motion detection sensor or an image sensor included in an identification unit. When a user's approach is detected, a digital human avatar is immediately displayed on the main kiosk's display to speak to the user and perform an interaction.

[0123] How to create a digital human avatar

[0124] According to an embodiment of the present invention, when user recognition is completed, the main kiosk can obtain user account information (age, interests, etc.) stored in a database and past interactive information search history.

[0125] The main kiosk can determine the most suitable initial digital human avatar form for the user based on the above-mentioned acquired data.

[0126] At this time, the form of the initial digital human avatar may be generated differently based on the user's account information.

[0127] For example, initial digital human avatars can be generated differently based on the user's age group. For instance, avatars in the form of friendly animals or cartoon characters can be provided to elementary school users, avatars representing specific professions (e.g., scientists, writers) to middle and high school students, and professional digital human avatars specialized for information delivery to adults. Additionally, initial digital human avatars can be generated differently based on the user's categories of interests. For instance, initial digital human avatars can also be generated differently based on the user's past query history.

[0128] Meanwhile, the above initial avatar form ( ) can be determined through a decision function such as the mathematical formula 1 below.

[0129]

[0130]

[0131] According to an embodiment of the present invention, the initial avatar form can be determined by integrating user age information, interest field information, and the results of a past query history analysis, and the integration process may include a multi-stage vector operation and a probability-based selection function as shown in [Equation 1].

[0132] The above It standardizes user age information to prevent distortion of the effect caused by differences in input scale.

[0133] The above It facilitates the calculation of associations by converting categorical areas of interest into fixed-length real vectors.

[0134] also, It quantifies past query history based on meaning to reflect the user's information search tendencies.

[0135] Each input characteristic is assigned a weighting coefficient such as α1, α2, and α3, allowing the importance of specific attributes to be dynamically adjusted. By applying these weighting coefficients, it is possible to create customized avatars, such as by strengthening the influence of the user's age group on avatar selection or increasing the influence of areas of interest.

[0136] The above-mentioned learned weight matrix and bias are parameters learned through a multilayer neural network structure, which model the non-linear combination relationships between different attributes to enable more precise avatar type prediction than simple rule-based methods.

[0137] The non-linear activation function enables learning relationships between complex attributes, and the Softmax function can numerically express the confidence of the final selection by calculating the probability distribution for each avatar type.

[0138] As such, the operation process according to [Equation 1] has the advantage of being able to clearly present the reliability of the final avatar type determination result while reflecting user-specific characteristics in various ways by performing input information preprocessing, weighted combination, non-linear transformation, and probability-based selection in stages.

[0139] The main kiosk can combine these input variables to output one of the predefined avatar types (e.g., character, specific job class, digital human).

[0141] According to another embodiment of the present invention, after the initial creation of a digital human avatar, the digital human avatar can be updated to grow based on the loan history of the user account.

[0142] For example, the control unit updates user account information as the user's cumulative number of loans increases, thereby controlling the addition of appearance items (e.g., glasses, graduation cap, etc.) of the digital human avatar or the change in the avatar's growth stage (e.g., student -> researcher -> doctorate).

[0143] More specifically, if a user repeatedly borrows books from a specific field (category), accessories or clothing related to that field may be added to the avatar.

[0144] For example, the 'scientist gown' item is given to the avatar of a user who has borrowed 10 or more books in the science field, and the 'palette and brush' item is added to the avatar of a user who mainly uses art-related books.

[0145] In addition, whenever the cumulative number of loans reaches a specific milestone (e.g., 50 books, 100 books), the appearance of the avatar matures or new interaction features (e.g., new drawing skills) are unlocked, which can have the effect of encouraging users to continue using the library and generating interest.

[0147] Provides customized function modes (S431, S432, S433, S434)

[0148] According to an embodiment of the present invention, user recognition is completed and a digital human avatar corresponding to the user can be output.

[0149] Afterwards, when a user's query is entered into the main kiosk (10), the main kiosk (10) identifies the intent of the user's query and activates the optimal function mode corresponding to it, thereby providing a customized service through the main kiosk (200).

[0150] Specifically, the main kiosk (10) can activate an optimal function mode based on keywords and application rules included in the user's query.

[0151] The above keywords can be defined as [Table 1] as follows.

[0152] Mode number Mode name Keyword Examples Explanation of applicable rules S431 User Guide Mode Notice, Announcements, Information, Operating Hours, Opening Hours, Closing Hours, Rental, Return, Extension, Overdue, Events, Reservation Guide, Membership Registration If the query includes keywords related to "User Guide Mode," points are added to the S431 score. If there are keywords related to events or announcements, priority is given to entering S431. S432 Facility Guide Mode Seat, seat, reading room, crowded, space, location, floor, map, way, directions, route, restroom, copier, printer, scanner Adds to the S432 score if keywords related to facility location, route, or congestion are included. Increases priority if a location entity exists. S433 Smart Book Recommendation Mode Recommendation, Similar, Similar, Author, Genre, Comfort, Warm, Touching, Funny, Reading Material, Novel, Comic, Cover Recognition, ISBN, Related Books S433 points are added for book recommendations, similar books, and sentiment-based requests. S433 priority entry is granted if cover recognition or ISBN input is provided. S434 Other modes Survey, satisfaction, opinions, feedback, event participation, promotion, informative content, quiz S434 points are added for requests for surveys, events, or promotional content. Additional points are awarded during survey campaign time slots.

[0153] According to one embodiment of the present invention, when a user's query (Q) is entered, the control unit of the main kiosk calculates the final score (Sm) of each function mode (m) according to the rules defined in [Table 1] and activates the mode with the highest score. The final activated mode (Mact) can be determined through the following [Equation 2].

[0154]

[0155]

[0156] The priority bonus point Pm has a non-zero positive value (e.g., 100 points) only when the priority entry conditions specified in the 'Description of Application Rules' in [Table 1] are satisfied, and has a value of 0 otherwise. The specific conditions for Pm for each mode are as follows.

[0157] According to an embodiment of the present invention, in the case of Ps431 (user guide mode), if the processor includes specific keywords in the query Q such as 'event', 'notice', etc., where the immediacy of information is judged to be important, Ps431 has a preset bonus point (e.g., 100 points).

[0158] Ps432 (Facility Guide Mode): As a result of Natural Language Processing (NLP) analysis, if clear location entities such as '3rd Floor Reading Room' or 'Restroom' are recognized in the query Q, Ps432 has a pre-set bonus point.

[0159] Ps433 (Smart Book Recommendation Mode): If the type of the user's input Q is 'cover recognition' via camera or 'ISBN scan', Ps433 has a preset bonus point.

[0160] Ps434 (Other Modes): If the current system time falls within the 'survey campaign' period set by the library, Ps434 has a pre-set bonus point.

[0161] According to an embodiment of the present invention, when the user guidance mode (S431) is activated, the control unit can retrieve notification information (e.g., announcements, event information) stored in the database (130) and process and provide information customized according to the user type (e.g., first-time visitors, parents accompanied by children).

[0162] For example, events in the 'Family' category can be filtered and presented first to parents accompanied by children.

[0163] According to an embodiment of the present invention, when the facility guidance mode (S432) is activated, the control unit analyzes status information, such as the real-time congestion level of the reading room, in conjunction with various sensors within the library and recommends an optimal space to the user. Additionally, when the user requests directions to a specific facility, the server (10) transmits AR (augmented reality) route data to the user terminal (30), so that the user can receive route guidance through the personal terminal screen.

[0164] In addition, according to an embodiment of the present invention, when a user is located on a path requesting directions to a specific facility, and the user is identified at a main kiosk, the main kiosk can update and output directions from the current location to the specific facility.

[0165] According to an embodiment of the present invention, when the smart book recommendation mode (S433) is activated, the main kiosk can recommend books to the user based on user account information, avatar growth level, and loan history category.

[0166] According to an embodiment of the present invention, when the other mode (S434) is activated, the system provides additional functions such as surveys or informational content. For example, if a user provides a survey opinion by voice, the data converted into text through speech-to-text (STT) technology can be stored in a database.

[0167] The main kiosk can recommend books to users based on user account information, avatar growth level, and borrowing history categories.

[0168] Specifically, when the smart book recommendation mode (S433) of the present invention is activated, the control unit of the main kiosk (10) can recognize the user's access or respond to a general request such as "book recommendation" and perform personalized book recommendations through multi-stage analysis and filtering.

[0169] First, the control unit can generate a primary candidate group based on the growth level of the user's digital human avatar (S710).

[0170] The avatar may be designed to acquire items or evolve its appearance based on the user's accumulated reading volume or reading activities in a specific field, and the control unit may consider the current growth stage of the avatar as the user's 'High-level Interest'.

[0171] For example, if the user's avatar has acquired the 'scientist gown' item through science reading activities, the processor can first extract books belonging to categories such as 'science' and 'technology' as primary candidates.

[0172] According to an embodiment of the present invention, the control unit can statistically analyze the loan history stored in the database (130) by category to calculate indicators such as frequency, recency, and dwell time, and refine items within the primary candidate group by assigning weights.

[0173] For example, the control unit can analyze the categories of books most frequently borrowed by the user from the database (130) (e.g., modern Korean novels, Western philosophy) and adjust the recommendation ranking by giving extra points to books within the first group of candidates that match or are highly relevant to the category.

[0174] For example, even if a candidate group of the 'Science' category is generated through primary filtering, if the proportion of the 'History' category in the user's borrowing history is high, the processor may adjust the recommendation ranking of books corresponding to the intersection of the two categories, such as 'History of Science', upward within the candidate group.

[0175] This has the effect of broadening the scope of recommendations by cross-analyzing the user's primary interests as well as potential and secondary interests.

[0176] According to an embodiment of the present invention, the control unit can determine a final ranking optimized for an individual by applying the user's account information as a final filter to a recommendation list refined up to the second stage.

[0177] In particular, the control unit can adjust the level of recommended books by utilizing the user's age information as an important variable. For example, regarding books on the same subject of 'history of science,' if the user account age is 15, books for teenagers can be ranked higher in the final recommendation list, and if the user account age is 40, books with academic depth can be ranked higher.

[0178] FIG. 4(b) is an example showing a screen in which a digital human avatar according to an embodiment of the present invention outputs smart book recommendation results through a kiosk.

[0179] According to an embodiment of the present invention, the main kiosk recommends the finally selected book and can output the basis for recommendation generated by synthesizing the analysis process through a display.

[0180] Finally, the digital human avatar can deliver the result to the user by explaining the specific reason for the recommendation in a form such as, "Your avatar has grown to the 'scientist' stage, and we have checked your recent reading history history. Based on your age group, we recommend 'Bill Bryson's Nearly Everything: A Short History' which satisfies both of these interests."

[0181] Subsequently, the control unit can perform a verification process to increase the objective reliability of the recommended books.

[0182] The control unit described above can retrieve recommended book list data from major domestic institutions, such as the National Library of Korea and the Seoul Metropolitan Library, which is pre-stored in a database. Subsequently, it compares and analyzes which of the recommended book lists of the aforementioned institutions the finally recommended book is included in, or what rank it corresponds to.

[0183] Finally, by synthesizing all the analysis results of the digital human avatar above, it is possible to effectively assist the user in making a choice by providing objective reliability information along with specific reasons for the recommendation in the form of: "Your avatar has grown to the 'scientist' stage, and we have confirmed your recent reading history history. We recommend 'A Short History of Nearly Everything' by Bill Bryson, tailored to your age group. For reference, this book is a verified book that is also included in the National Library of Korea's '100 Recommended Books for Young Adults'."

[0184] Those skilled in the art will understand that the various exemplary logic blocks, modules, processors, means, circuits, and algorithm steps described in connection with the embodiments disclosed herein may be implemented by electronic hardware, various forms of programs or design code (referred to herein as software for convenience), or a combination of all of these.

[0185] The present invention described above can be implemented as computer-readable code on a medium on which a program is recorded. A machine-readable recording medium includes all types of recording devices in which data that can be read by a computer system is stored. Examples of machine-readable media include HDD (Hard Disk Drive), SSD (Solid State Disk), SSD (Silicon Disk Drive), ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical data storage device, etc.

[0186] In one embodiment, the recording medium may be a memory. In one embodiment, the recording medium may be implemented in a distributed form in a networked computer system, etc. Software may be stored and executed in a distributed manner in a computer system, etc. The recording medium may be a non-transitory recording medium. A non-transitory recording medium may refer to a tangible medium that exists regardless of whether data is stored semi-permanently or temporarily.

Claims

Claim 1 A main kiosk that performs user recognition, information output, and shelving control; and a shelving module that can be physically and electrically coupled to at least one side of the main kiosk and stores at least one book; wherein the main kiosk comprises: an identification unit that recognizes the face of an approaching user to obtain user account information; and a display unit that outputs information. The system includes a control unit that executes at least one operation mode among a book guide mode, a usage guide mode, a facility guide mode, and other modes; wherein the shelving module includes an RFID reader unit installed in each storage compartment (Shelf) to detect the status of book entry and exit in real time; and the control unit (a) when a withdrawal signal of a specific book is detected from the RFID reader unit, (b) converts the user account information into a high-dimensional real vector through the embedding layer of a natural language processing model to generate a user interest vector, (c) divides the original text of the specific book into a plurality of segments and vectorizes each segment to generate a plurality of book segment vectors, (d) calculates a personalized attention map by calculating the cosine similarity between the user interest vector and the plurality of book segment vectors respectively and assigning weights to each segment, (e) extracts key segments that have obtained a score above a preset threshold from the attention map and inputs them into a generative AI model to generate summary information that emphasizes parts that are mathematically close to the user's interests, and (f) the vocabulary difficulty of the summary information based on age or interest data included in the user account information. A smart bookshelf system characterized by converting the and descriptive tone to reprocess it into personalized information, (g) controlling the converted personalized information to be output through the display unit. Claim 2 A smart bookshelf system according to claim 1, wherein the control unit, when a withdrawal signal indicating that the specific book has been removed from the bookshelf is detected from the RFID reader, analyzes the metadata and original text of the specific book to generate the summary information, converts the vocabulary difficulty and explanatory tone of the summary information based on age or interest data included in previously acquired user account information, and controls a virtual digital human avatar displayed on the display unit to utter the converted summary information to the user in the explanatory tone. Claim 3 A smart shelving system according to claim 2, wherein a main connector for power supply and data communication is provided on the side of the main kiosk, and a shelving connector coupled to the main connector is provided on the side of the shelving module, and the control unit automatically updates the book location map of the entire shelving system as the number of connections or arrangement type of the shelving module changes. Claim 4 A smart shelving system according to claim 3, wherein the shelving module comprises at least one of a general shelf on which books are placed, a monitor for displaying digital information, a display shelf for placing books facing forward, a lower cabinet for storing items, a return monitor for return processing, and a return stand. Claim 5 A smart bookshelf system according to claim 4, wherein the bookshelf module further includes an LED indicator placed on the front of each storage compartment, and the control unit illuminates the LED indicator of the storage compartment where the book is located to visually guide the physical location when a digital human avatar recommends a specific book or guides other books related to a book retrieved by the user, and highlights and displays the book information using augmented reality (AR) or visual graphics through the display unit. Claim 6 A smart bookshelf system according to claim 1, wherein the control unit, upon inputting a user query, calculates a weight for each operation mode based on a priority bonus point application rule and executes the mode that obtains the highest score, and the priority bonus point application rule applies bonus points to the user guide mode if keywords related to events and announcements are included in the user query, applies bonus points to the facility guide mode if a location entity is included, applies bonus points to the book guide mode if cover recognition or ISBN input signal is detected, and applies bonus points to other modes if the current time is a survey campaign time period, thereby automatically entering the mode that obtains the highest score. Claim 7 A smart bookshelf system according to claim 6, wherein the control unit assigns weights to user age information, interest field information, and past query history information based on the user account information, generates a digital human avatar of a type determined through the following <Mathematical Formula 1>, and displays it on the display unit. <Mathematical Formula 1> Claim 8 A smart bookshelf system according to claim 7, wherein the control unit updates user account information such that appearance items of the digital human avatar are added or the growth stage of the avatar changes as the user's cumulative number of loans increases. Claim 9 A smart bookshelf system according to claim 8, wherein the control unit, when the book guidance mode is executed, generates a primary book candidate group based on the growth level of the user's virtual avatar, statistically analyzes the user's borrowing history by category to calculate frequency, recency, and dwell time indicators, refines items within the primary book candidate group by assigning weights to them, and determines a final ranking by applying the user's age information as a final filter to the refined list to adjust the difficulty level of the recommended books. Claim 10 A smart bookshelf system according to claim 9, wherein the control unit retrieves pre-stored recommended book list data from an external institution, verifies whether a book included in the final ranking is included in the recommended book list of the external institution, and guides the user to the verification result and the basis for recommendation through the speech of the digital human avatar.

Images