• KIISE Transactions on Computing Practices
• /
• v.21 no.4
• /
• pp.327-332
• /
• 2015

Constructing a k-nearest neighbor (k-NN) graph is a primitive operation in the field of recommender systems, information retrieval, data mining and machine learning. Although there have been many algorithms proposed for constructing a k-NN graph, either the existing approaches cannot be used for various types of similarity measures, or the performance of the approaches is decreased as the number of nodes or dimensions increases. In this paper, we present a novel algorithm for k-NN graph construction based on "balanced" canopy clustering. The experimental results show that irrespective of the number of nodes or dimensions, our algorithm is at least five times faster than the brute-force approach while retaining an accuracy of approximately 92%.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2011.11a
• /
• pp.1270-1271
• /
• 2011

소셜 네트워크 서비스는 가트너에서 2011년에 이어 2012년에도 각광받을 기술의 하나로 선정된 만큼 미래 인터넷의 핵심 키워드 중 하나로도 뽑히며, 엔터테인먼트, 검색, 방송, 커머스 등의 여러 가지 서비스와 직접 연결된다. 이러한 소셜 네트워크 서비스 가운데 하이브리드형 서비스는 사용자의 정보를 관리 및 파악하여 사용자가 원하는 제품을 예측하고 추천해주고 있으며, 이를 위해 그래프 마이닝 기술을 적용하고 있다. 하지만 그래프 마이닝 기술은 아직 복잡한 그래프 구조의 데이터에서 정보를 추출하기에 제약사항들이 발생하므로 이에 대하여 많은 연구가 활발히 이루어지고 있다. 이러한 그래프 마이닝 기술을 나아가 더 발전시켜 활용하면 기존의 하이브리드형 서비스에서 사용자의 정보를 파악하여 충성도를 높여줄 뿐 아니라 기업에서의 타켓 마케팅과 원투원 마케팅을 가능하게 해주고 기존 사용자에 대한 교차 판매와 격상판매의 전략들을 도출할 수 있을 것이다.

• Annual Conference on Human and Language Technology
• /
• 2023.10a
• /
• pp.185-189
• /
• 2023

Text to SPARQL은 지식 그래프 기반 질의응답의 한 형태로 자연어 질문을 지식 그래프 검색 쿼리로 변환하는 태스크이다. SPARQL 쿼리는 지식 그래프의 정보를 기반으로 작성되어야 하기 때문에 기존 언어 모델을 통한 코드 생성방법으로는 잘 동작하지 않는다. 이에 우리는 거대 언어 모델을 활용하여 Text to SPARQL를 해결하기 위해 프롬프트에 지식 그래프의 정보를 증강시켜주는 방법론을 제안한다. 이에 더하여 다국어 정보 활용에 대한 영향을 검증하기 위해 한국어, 영어 각각의 레이블을 교차적으로 실험하였다. 추가로 한국어 Text to SPARQL 실험을 위하여 대표적인 Text to SPARQL 벤치마크 데이터셋 QALD-10을 한국어로 번역하여 공개하였다. 위 데이터를 이용해 지식 증강 프롬프팅의 효과를 실험적으로 입증하였다.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2015.05a
• /
• pp.418-420
• /
• 2015

The SNOMED CT browser is a system for searching and browsing of a huge volume of medical terminologies included in SNOMED CT. Previous browsers provide a simple list-up of search results while they are similar from each other. It leads to a serious confusion in selecting an appropriate term among them. This paper presents a novel browser system which provides a comparative search of search results. To do this, the proposed system includes a terminology search module, a topology search module, and a sub-graph generation module for the results.

• Journal of KIISE:Software and Applications
• /
• v.36 no.11
• /
• pp.928-935
• /
• 2009

Ontology schema reasoning is used to maintain consistency of concepts and build concept hierarchy automatically. For the purpose, the search of concepts must be inevitably performed. Ontology schema reasoning performs the test of subsumption relationships of all the concepts delivered in the test set. The result of subsumption tests is determined based on the creation of complete graphs, which seriously weighs with the performance of reasoning. In general, the process of creating complete graph has been known as expressive procedure. This process is essential in improving the leading performance. In this paper, we propose a method enhancing the classification performance by identifying unnecessary subsumption test supported by optimized searching method on subsumption relationship test among concepts. It is achieved by propagating subsumption tests results into other concept.

• Journal of KIISE:Databases
• /
• v.27 no.2
• /
• pp.185-198
• /
• 2000

Recently the technological advance in the hardware dealing with multimedia data as well as the explosive increase of the volume of multimedia data bring about new interest in the use of multimedia presentations in many application domains. To use multimedia presentations efficiently, the integration of multimedia presentations into DBMS is necessary. This paper presents a multimedia presentatation query language based on contents and query processing techniques. Presently, multimedia presentation authoring tools denote a multimedia presentation using a presentation graph which is a DAG. A Node in the graph is a same type of media stream and edges denote a play-out order and a synchronization way among nodes. The contents of presentations graphs are the information of each stream, the sequential order of the information inside each stream and the play-out order among the streams. GCalculus/S is a calculus-based query language and can deal with the contents of a presentation graph and physical characteristics of multimedia data. It expresses the sequential order of information inside each stream and the play-out order of streams of a presentation graph using temporal operators Next, Connected and Until. O-Algebra, which is object algebra, is extended to process GCalculus/S queries.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.38 no.1
• /
• pp.33-42
• /
• 2001

Recently, the interests in the 3D image, generated from the range data and CAD, have exceedingly increased, accordingly a various 3D image database is being constructed. The efficient and fast scheme to access the desired image data is the important issue in the application area of the Internet and digital library. However, it is difficult to manage the 3D image database because of its huge size. Therefore, a proper descriptor is necessary to manage the data efficiently, including the content-based search. In this paper, the proposed shape descriptor is based on the voxelization of the 3D image. The medial axis transform, stemming from the mathematical morphology, is performed on the voxelized 3D image and the graph, which is composed of node and edge, is generated from skeletons. The generated graph is adequate to the novel shape descriptor due to no loss of geometric information and the similarity of the insight of the human. Therefore the proposed shape descriptor would be useful for the recognition of 3D object, compression, and content-based search.

• The KIPS Transactions:PartD
• /
• v.13D no.7 s.110
• /
• pp.1017-1026
• /
• 2006

Recently, distributed collaborative development environment has been recognized an alternative environment for product development in which multidisciplinary participants are naturally involving. Reuse of Product design information has long been recognized as one of core requirements for efficient product development. This paper addresses an image-based retrieval system to support product design information reuse. In the system, product images obtained from multi-modal devices are utilized to reuse design information. The proposed system conducts the segmentation of a product image by using a labeling method and generates an attributed relational graph (ARG) that represents properties of segmented regions and their relationships. The generated ARG is extended by integrating corresponding part/assembly information. In this manner, the reuse of assembly design information using a product image has been realized. The main advantages of the presented system are following. First, the system is not dependent to specific design tools, because it utilizes multimedia images that can be obtained easily from peripheral devices. Second ratio-based features extracted from images enable image retrievals that contain various sizes of parts. Third, the system has shown outstanding search performance, because we applied various information of segmented part regions and their relationships between parts.

• Journal of Korean Society of Archives and Records Management
• /
• v.21 no.3
• /
• pp.61-78
• /
• 2021

Records on Japanese "Comfort Women" have been individually managed by private sectors or institutions, and some are provided as digital archives on the Internet. However, records of digital archives differ in the composition and representation of metadata by individual institutions. Meanwhile, there is a lack of a consistent structure to describe the relationships between and among these records, leading to their fragmentation and disconnectedness. This paper proposes a knowledge model for interlinking the digital archival resources and builds a knowledge graph by integrating the records from distributed digital archives. It derives common elements by analyzing metadata from the diverse digital archives and expresses them in standard vocabularies to semantically describe multiple entities and relationships of the digital archival resources. In particular, the study includes the refinement of collected data to search and thread dispersed records and the enrichment of external data to provide significant contextual information of records. An evaluation of the knowledge graph is performed via a query measuring the (dis)connectivity between the distributed records. As a result, the knowledge graph is capable of interlinking and retrieving fragmented records, providing substantial contextual information on the records with external data enrichment, and searching accurately to match the user's intentions through semantic-based queries.

• Journal of KIISE:Computer Systems and Theory
• /
• v.37 no.1
• /
• pp.27-34
• /
• 2010

Web crawlers are fundamental programs which iteratively download web pages by following links of web pages starting from a small set of initial URLs. Previously several web crawling orderings have been proposed to crawl popular web pages in preference to other pages, but some graph search techniques whose characteristics and efficient implementations had been studied in graph theory community have not been applied yet for web crawling orderings. In this paper we consider various graph search techniques including lexicographic breadth-first search, lexicographic depth-first search and maximum cardinality search as well as well-known breadth-first search and depth-first search, and then choose effective web crawling orderings which have linear time complexity and crawl popular pages early. Especially, for maximum cardinality search and lexicographic breadth-first search whose implementations are non-trivial, we propose linear-time web crawling orderings by applying the partition refinement method. Experimental results show that maximum cardinality search has desirable properties in both time complexity and the quality of crawled pages.