• Proceedings of the Korea Information Processing Society Conference
• /
• 2022.05a
• /
• pp.356-358
• /
• 2022

추천 시스템에서 사용되는 피드백은 단일 클래스와 다중 클래스 피드백으로 구분할 수 있다. 추천 시스템을 위한 지식 증류 기법들은 단일 클래스 환경에서 주로 연구되어 왔다. 우리는 다중 클래스 환경에서 또한 추천 시스템을 위한 최신 지식 증류 기법들이 효과적인지에 대해 알아보고자 하며, 해당 방법들 간의 추천 정확도를 비교해보고자 한다. 추천 시스템에서 보편적으로 사용되는 데이터 셋들을 기반으로 한 실험들을 통해 추천 시스템을 위한 지식 증류 기법들은 같은 조건의 기본적인 추천 시스템에 비해 정확도가 최대 193%까지 개선되는 것을 확인했다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2007.11a
• /
• pp.448-451
• /
• 2007

매일 방대한 양의 다양하고 복잡한 데이터가 발생하는 오늘날 정보의 홍수 속에서 유용한 정보를 추출하기 위해서는 상당한 시간과 노력이 소요된다. 본 논문에서는 주어진 데이터로부터 공통속성을 갖는 객체단위의 정보를 추출, 분석하여 동치관계를 기반으로 클래스계층구조로 표현하는 클래스계층화 분석기법을 소개한다. 또한, 다양하고 복잡한 실세계 데이터들 속에 숨겨져 있는 개념들과 관계들을 파악하여 수월하게 클래스계층구조를 구축하기위하여, 본 연구에서 개발 중인 3가원소 컨텍스트를 토대로하는 클래스계층화분석 자동화지원도구를 제안한다.

• The KIPS Transactions:PartA
• /
• v.12A no.3 s.93
• /
• pp.229-234
• /
• 2005

The virtual machine is a programming environment that supports device and platform independence. So far, virtual machines such as JVM and KVM have been used in a variety of environments for the Java language. Some virtual machines similar to them are also being developed and used. This paper Presents the experiences of extracting elements essential for small sized devices such as PDA from Java Class files(*.class) and designing a converted class file(*.rclass) for runtime efficiency by modifying its class file format and developing its translator. In addition, a virtual machine is developed to receive the translated class file entered and output the runtime results.

• Journal of KIISE:Computing Practices and Letters
• /
• v.16 no.5
• /
• pp.571-575
• /
• 2010

Mutation testing is known as a very useful technique for measuring the effectiveness of a test data set and finding weak points of the test set. An equivalent mutant degrades the effectiveness of mutation testing. Elimination of equivalent mutants is a very important problem in mutation testing.In this paper, we proposed kinds of methods for detecting class-level equivalent mutants. These methods judge the equivalency of mutants through structural informations and behavioral information of the original program and mutants using static analysis. We found that our approach can detect not a few of equivalent mutants and expected that the cost of mutation testing can be saved considerably.

• Journal of KIISE:Software and Applications
• /
• v.36 no.12
• /
• pp.1024-1027
• /
• 2009

In this letter, we evaluate the classification performance of mixed numeric and categorical data for comparing the efficiency of feature filtering and feature wrapping. Because the mixed data is composed of numeric and categorical features, the feature selection method was applied to data set after discretizing the numeric features in the given data set. In this study, we choose the feature subset for improving the classification performance of the data set after preprocessing. The experimental result of comparing the classification performance show that the feature wrapping method is more reliable than feature filtering method in the aspect of classification accuracy.

• Annual Conference on Human and Language Technology
• /
• 2018.10a
• /
• pp.367-370
• /
• 2018

웹 또는 모바일 사용자는 고객 센터에 구축된 자주 묻는 질문을 이용하여 원하는 서비스를 제공받는다. 그러나 자주 묻는 질문은 사용자가 직접 핵심어를 입력하여 검색된 결과 중 필요한 정보를 찾아야 하는 어려움이 있다. 이러한 문제를 해결하기 위해 본 논문에서는 사용자 질의를 입력 받아 질의에 해당하는 클래스를 분류해주는 문장 분류 모델을 제안한다. 제안모델은 웹이나 모바일 환경의 오타나 맞춤법 오류에 대한 강건함을 위해 자소 단위 합성곱 신경망을 사용한다. 그리고 기계 번역 이외에도 자연어 처리 부분에서 큰 성능 향상을 보여주는 주의 집중 방법과 클래스 임베딩을 이용한 문장 분류 시스템을 사용한다. 457개의 클래스 분류와 769개의 클래스 분류에 대한 실험 결과 Micro F1 점수 기준 81.32%, 61.11%의 성능을 보였다.

• Journal of Korea Multimedia Society
• /
• v.13 no.10
• /
• pp.1514-1524
• /
• 2010

The standard Java class library does not support the platform-dependent features needed by the application. Therefore, the Java application including the platform -dependent features must supplement the required features by invoking native functions using JNI. The native language programmer has to explicitly specify how to connect to various Java objects and later to disconnect from them. In this paper, I suggest a way to avoid these annoying works. The native method in the pure java class can not contain a native code block. By providing a native code block for the native method, it is possible for programmer to write a native code without being aware of JNI. To achieve this, I introduced the native class that is a java class on the native environment, and made it possible to interchange data by placing an arbitrator between the java class and the native class.

• The KIPS Transactions:PartD
• /
• v.13D no.4 s.107
• /
• pp.619-624
• /
• 2006

Currently domestic game market rapidity is developmental but the game production process does not become systematization. Also it is bringing the failure of game that problem point of conversation between game planner and game developer.. The research which to conversation between game planner and game developer it sees does extracting a game element to expressed game characteristics from the product for game planning and it change extracted element into class and In order to express the relationship of element for it presents a relationship of class. Instance research it leads. It grasps a relationship among extracted classes and it supports systematic game planning.

• Journal of the Korean Society for information Management
• /
• v.31 no.2
• /
• pp.41-56
• /
• 2014

The necessity of the raw data management and reuse is issued by diffusion of the recognition that research data is a national asset. In this paper, a metadata design model by schema class inheritance and a metadata integrated search model by schema objects are suggested for a structural management of the data. A data architecture in which an schema object has an 1 : 1 relation to the data collection was designed. A suggested model was testified by creation of a virtual schema class and objects which inherit the schema class. It showed the possibility of implement systematically. A suggested model can be used to manage the data which are produced by government agencies because schema inheritance and integrated search model present way to overcome the weak points of the 'Top-dow Hierarchy model' which is being used to design the metadata schema.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2002.11a
• /
• pp.373-377
• /
• 2002

Unlike desktop systems, embedded systems usually meet a strict restriction on using memory. It is required to allocate several class files on memory to run a Java program. A Java class file contains such data including a constant pool, class overview, fields information, and methods information. Some of them are used merely for a debugging purpose, others for a program execution. This paper analyzes the internal structure, or format of the class files used for embedded Java systems. We also investigate how much memory will be necessary for each part of the class files when the files are allocated in memory. The experiment was performed on the simpleRTJ, an open-source commercial embedded Java system.