KR-20260062570-A - SPATIAL INFORMATION SEARCH SYSTEM FOR GENERATIVE AI BASED ON RAG, APPARATUS AND METHOD USING THE SAME

KR20260062570AKR 20260062570 AKR20260062570 AKR 20260062570AKR-20260062570-A

Abstract

The present invention relates to a spatial information search system for RAG-based generative AI, and a spatial information search server and method for the same. In particular, the invention relates to a spatial information search system for RAG-based generative AI and a spatial information search server and method for the same, wherein the spatial information search server can generate and provide location information-based association information so that the RAG-based generative AI server can generate response information by utilizing association information that considers movement paths as a knowledge source.

Inventors

김홍래

Assignees

주식회사 휴플

Dates

Publication Date: 20260507
Application Date: 20241029

Claims (14)

A RAG-based generative AI server that generates response information corresponding to query information input from a user terminal and transmits it to the user terminal; and A spatial information search server that identifies multiple keywords in the above query information, extracts mapped location information using the identified multiple keywords, selects a regional information database using the extracted location information, extracts association information considering movement paths from the extracted location information using the selected regional information database, and transmits to the RAG-based generative AI server to generate response information using the extracted association information; A spatial information retrieval system for RAG-based generative AI characterized by including
In Article 1, The above RAG-based generative AI server is A spatial information retrieval system for RAG-based generative AI characterized by providing a prompt for inputting query information to the above-mentioned user terminal, and when query information is input through the above-mentioned prompt, searching and augmenting the above-mentioned query information based on RAG (Retrieval-Augmented Generation) using a knowledge source including the above-mentioned related information, and then generating response information for the augmented query information through a Large Language Model (LM) model.
In Article 1, The above spatial information search server is A spatial information search system for RAG-based generative AI characterized by extracting a defined core keyword having a location attribute from a plurality of keywords, expanding the plurality of keywords by adding the same keyword or a keyword having the same attribute to each of the plurality of keywords using the extracted core keyword, extracting corresponding plurality of location information using the expanded plurality of keywords, selecting a regional information database using the extracted plurality of location information, and extracting association information using the location information in the selected regional information database.
In Paragraph 3, The above spatial information search server is A spatial information search system for RAG-based generative AI characterized by generating a movement path considering the distance or time of travel between the plurality of location information using the selected regional information database, and extracting the location information of the place as associated information when there is a place matching the attribute item of the plurality of keywords on the generated movement path.
In Paragraph 4, The above spatial information search server is A spatial information search system for RAG-based generative AI characterized by, when there are two or more places matching the above attribute items, calculating scores for similarity and difference items between information about the places and location information used to generate the movement path, and using the calculated scores to select one of the places and extract the location information of the selected place as associated information.
In Article 1, The above spatial information search server is Check the user information of the above user terminal, and A spatial information search system for RAG-based generative AI characterized by extracting the above-mentioned related information using the above-mentioned verified user information.
A keyword verification unit that verifies multiple keywords in query information input from a user terminal to a RAG-based generative AI server; A location information extraction unit that extracts multiple location information mapped using the above multiple keywords and selects a regional information database using the extracted location information; and An association information extraction unit that extracts association information considering movement paths from the extracted location information using the selected regional information database, and transmits it to the RAG-based generative AI server to generate response information corresponding to the query information considering the extracted association information; A spatial information search server characterized by including
In Paragraph 7, The above keyword verification unit A spatial information search server characterized by identifying keywords based on region, time, place, and attribute items in the above query information.
In Paragraph 7, The above keyword verification unit Extracting a defined core keyword having a positional attribute from the above multiple keywords, and expanding the multiple keywords by adding the same keyword or a keyword having the same attribute to each of the above multiple keywords using the extracted core keyword. The above location information extraction unit Using the aforementioned expanded multiple keywords, corresponding multiple location information is extracted, and a local information database is selected using the extracted multiple location information, The above location information A spatial information search server characterized by administrative district information including latitude and longitude coordinate values.
In Paragraph 9, The defined core keyword with the above positional attribute is A spatial information search server characterized by being a region item keyword containing administrative district information or a place item keyword included in the first sentence of the above query information.
In Paragraph 7, The above-mentioned association information extraction unit A spatial information search server characterized by generating a movement path considering the distance or time of travel between the plurality of location information using the selected regional information database, and, if there is a place matching the attribute item of the plurality of keywords on the generated movement path, extracting the location information of that place as associated information.
A spatial information search method in a spatial information search system for RAG-based generative AI, comprising a RAG-based generative AI server that augments query information input from a user terminal based on RAG and generates response information for the augmented query information using an LLM model, and a spatial information search server linked with said RAG-based generative AI server. The above spatial information search server, A step of verifying multiple keywords in the above query information; A step of extracting multiple location information mapped using the above multiple keywords; A step of selecting a regional information database using the extracted plurality of location information; A step of extracting associated information considering movement paths from the extracted location information using the selected area information database; and A step of transmitting to the RAG-based generative AI server to generate response information corresponding to the query information by considering the extracted association information; A spatial information retrieval method characterized by including
In Paragraph 12, The step of extracting the above-mentioned association information is A spatial information search method characterized by generating a movement path considering the distance or time of travel between the plurality of location information using the selected area information database, and, if there is a place matching the attribute item of the plurality of keywords on the generated movement path, extracting the location information of that place as associated information.
A computer-readable recording medium storing a program that executes a spatial information search method described in either paragraph 12 or 13.

Description

Spatial Information Search System for Generative AI Based on RAG, Apparatus and Method for the Same The present invention relates to a spatial information retrieval system for RAG-based generative AI, and a spatial information retrieval server and method for the same. Among generative AI technologies, Large Language Models (LLMs) face a problem in that they are difficult to train for the analysis of queries with positional attributes and the provision of answers because they are supplied with data that interprets the spatial range of information within a set of information containing positional attributes. This can lead to the problem of hallucination, which refers to meaningless or incorrect information that can occur when a generative AI model is not trained with sufficient data, is trained with noisy or contaminated data, or is not provided with sufficient context or constraints. To address these issues, attempts are being made to integrate Retrieval-Augmented Generation (RAG) models into the application of generative AI. However, there is still a problem in that it is difficult to provide more effective information using information sets with location attributes. FIG. 1 is a diagram schematically illustrating a spatial information retrieval system for RAG-based generative AI according to an embodiment of the present invention. FIG. 2 is a block diagram illustrating the main configuration of a spatial information search server according to an embodiment of the present invention. FIG. 3 is an illustrative diagram for explaining the main operations in a spatial information search server according to an embodiment of the present invention. FIG. 4 is an illustrative diagram for explaining the process of generating associated information according to an embodiment of the present invention. FIG. 5 is a flowchart illustrating a spatial information retrieval method in a spatial information retrieval system for RAG-based generative AI according to an embodiment of the present invention. In order to clarify the features and advantages of the means for solving the problem of the present invention, the present invention will be described in more detail with reference to specific embodiments of the present invention illustrated in the attached drawings. However, detailed descriptions of known functions or configurations that may obscure the essence of the invention are omitted in the following description and the attached drawings. Additionally, it should be noted that identical components throughout the drawings are indicated by the same reference numerals whenever possible. Terms and words used in the following description and drawings should not be interpreted as being limited to their ordinary or dictionary meanings, but should be interpreted in a meaning and concept consistent with the technical spirit of the invention, based on the principle that the inventor can appropriately define the concept of terms to best describe his invention. Accordingly, the embodiments described in this specification and the configurations illustrated in the drawings are merely the most preferred embodiments of the invention and do not represent all aspects of the technical spirit of the invention; therefore, it should be understood that various equivalents and modifications capable of replacing them may exist at the time of filing this application. Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a diagram schematically illustrating a spatial information retrieval system for RAG-based generative AI according to an embodiment of the present invention. Referring to FIG. 1, the spatial information search system (500) for RAG-based generative AI of the present invention may be configured to include a user terminal (100), a RAG-based generative AI server (200), and a spatial information search server (300). First, the user terminal (100) is a device that is used by a user and can transmit query information to a RAG-based generative AI server (200) and receive corresponding response information. For example, the user terminal (100) of the present invention may include not only portable terminals such as smartphones, tablet PCs, PDAs, portable multimedia players, and MP3 players, but also fixed terminals such as personal computers and laptop PCs. Of course, the devices listed above are merely examples, and any device capable of processing the functions required by the present invention may be used as the user terminal (100) according to the embodiment of the present invention. In addition, the user terminal (100) of the present invention can be connected to a generative AI server (200) through a communication network (not shown) and can transmit and receive various information. The communication network (not shown) refers to a connection structure or network capable of exchanging information between each node. Various types of communication networks can be used for such a commun