• Tunnel and Underground Space
• /
• v.30 no.6
• /
• pp.509-518
• /
• 2020

This study constructed a new simulation platform, including mesh generation process, numerical simulation, and post-processing for results analysis based on exploration data to perform real-scale numerical analysis considering the actual geological structure efficiently. To build the simulation platform, we applied for open-source programs. The source code is open to be available for code modification according to the researcher's needs and compatibility with various numerical simulation programs. First, a three-dimensional model(3D) is acquired based on the exploration data obtained using a drone. Then, the domain's mesh density was adjusted to an interpretable level using Blender, the free and open-source 3D creation suite. The next step is to create a 3D numerical model by creating a tetrahedral volume mesh inside the domain using Gmsh, a finite element mesh generation program. To use the mesh information obtained through Gmsh in a numerical simulation program, a converting process to conform to the program's mesh creation protocol is required. We applied a Python code for the procedure. After we completed the stability analysis, we have created various visualization of the study using ParaView, another open-source visualization and data analysis program. We successfully performed a preliminary stability analysis on the full-scale Dokdo model based on drone-acquired data to confirm the usefulness of the proposed platform. The proposed simulation platform in this study can be of various analysis processes in future research.

• Journal of Convergence Society for SMB
• /
• v.3 no.2
• /
• pp.33-39
• /
• 2013

Recently, lots of anti-phishing schemes have been developed. Several products identify phishing sites and show the results on the address bar of the internet browser. However, they determine only by domain names or IP addresses. Although this kind of method is effective against recent DNS Pharming attacks, there is still a possibility that hidden attacks which modifies HTML codes could incapacitate those anti-phishing programs. In this paper, the cognitive approach which compares images to decide phishing or Pharming is presented by using system tray and balloon tips that are hard to fake with pop-ups or flash in order for users to compare pictures from connecting sites and system tray. It differs from an old method that a program analyzes IP or domains to judge if it is phishing or Pharming. Therefore, proposed method effective cognitive approach against phishing and Pharming attacks.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.19 no.1
• /
• pp.113-124
• /
• 2009

Recently, lots of anti-phishing schemes have been developed. Several products identify phishing sites and show the results on the address bar of the internet browser, but they determine only by domain names or IP addresses. Although this kind of method is effective against recent DNS pharming attacks, there is still a possibility that hidden attacks which modifies HTML codes could incapacitate those anti-phishing programs. In this paper, the cognitive approach which compares images to decide phishing or pharming is presented, using system tray and balloon tips that are hard to fake with pop-ups or flash in order for users to compare pictures from connecting sites and system tray. It differs from an old method that a program analyzes IP or domains to judge if it is phishing or pharming, but observes if there were HTML code changing between plug-ins and a server.

• The Journal of Society for e-Business Studies
• /
• v.27 no.2
• /
• pp.65-77
• /
• 2022

Due to the recent development in electronic financial services, transactions of electronic prepayment are rapidly growing, leading to growing fraud attempts. This paper proposes a methodology that can effectively detect fraud transactions in electronic prepayment by machine learning algorithms, including support vector machines, decision trees, and artificial neural networks. Actual transaction data of electronic prepayment services were collected and preprocessed to extract the most relevant variables from raw data. Two different approaches were explored in the paper. One is a transaction-based approach, and the other is a user ID-based approach. For the transaction-based approach, the first model is primarily based on raw data features, while the second model uses extra features in addition to the first model. The user ID-based approach also used feature engineering to extract and transform the most relevant features. Overall, the user ID-based approach showed a better performance than the transaction-based approach, where the artificial neural networks showed the best performance. The proposed method could be used to reduce the damage caused by financial accidents by detecting and blocking fraud attempts.

• Journal of Internet Computing and Services
• /
• v.24 no.4
• /
• pp.127-135
• /
• 2023

This research paper explores the application of FinBERT, a variational BERT-based model pre-trained on financial domain, for sentiment analysis in the financial domain while focusing on the process of identifying suitable training data and hyperparameters. Our goal is to offer a comprehensive guide on effectively utilizing the FinBERT model for accurate sentiment analysis by employing various datasets and fine-tuning hyperparameters. We outline the architecture and workflow of the proposed approach for fine-tuning the FinBERT model in this study, emphasizing the performance of various datasets and hyperparameters for sentiment analysis tasks. Additionally, we verify the reliability of GPT-3 as a suitable annotator by using it for sentiment labeling tasks. Our results show that the fine-tuned FinBERT model excels across a range of datasets and that the optimal combination is a learning rate of 5e-5 and a batch size of 64, which perform consistently well across all datasets. Furthermore, based on the significant performance improvement of the FinBERT model with our Twitter data in general domain compared to our news data in general domain, we also express uncertainty about the model being further pre-trained only on financial news data. We simplify the complex process of determining the optimal approach to the FinBERT model and provide guidelines for selecting additional training datasets and hyperparameters within the fine-tuning process of financial sentiment analysis models.

• Journal of the Korean Society of Radiology
• /
• v.15 no.6
• /
• pp.803-811
• /
• 2021

In this study, the exposure amount of IASCC test worker was evaluated by applying the process simulation technology. Using DELMIA Version 5, a commercial process simulation code, IASCC test facility, hot cells, and workers were prepared, and IASCC test activities were implemented, and the cumulative exposure of workers passing through the dose-distributed space could be evaluated through user coding. In order to simulate behavior of workers, human manikins with a degree of freedom of 200 or more imitating the human musculoskeletal system were applied. In order to calculate the worker's exposure, the coordinates, start time, and retention period for each posture were extracted by accessing the sub-information of the human manikin task, and the cumulative exposure was calculated by multiplying the spatial dose value by the posture retention time. The spatial dose for the exposure evaluation was calculated using MCNP6 Version 1.0, and the calculated spatial dose was embedded into the process simulation domain. As a result of comparing and analyzing the results of exposure evaluation by process simulation and typical exposure evaluation, the annual exposure to daily test work in the regular entrance was predicted at similar levels, 0.388 mSv/year and 1.334 mSv/year, respectively. Exposure assessment was also performed on special tasks performed in areas with high spatial doses, and tasks with high exposure could be easily identified, and work improvement plans could be derived intuitively through human manikin posture and spatial dose visualization of the tasks.

• Journal of KIISE:Software and Applications
• /
• v.36 no.9
• /
• pp.732-748
• /
• 2009

The technology of Semantic Web realizes the base technology for context-awareness that creates new services by dynamically and flexibly combining various resources (people, concepts, etc). According to the realization of ubiquitous computing technology, many researchers are currently working for the embodiment of web service. However, most studies of them bring about the only predefined results those are limited to the initial description by service designer. In this paper, we propose a new service supporting model to provide an automatic method for plan related tasks which achieve goal state from initial state. The inputs on an planner are intial and goal descriptions which are mapped to the current situation and to the user request respectively. The idea of the method is to infer context from world model by DL-based ontology reasoning using OWL domain ontology. The context guide services to be loaded into planner. Then, the planner searches and plans at least one service to satisfy the goal state from initial state. This is STRIPS-style backward planner, and combine OWL-S services based on AI planning theory that enabling reduced search scope of huge web-service space. Also, when feasible service do not find using pattern matching, we give user alternative services through DL-based semantic searching. The experimental result demonstrates a new possibility for realizing dynamic service modeler, compared to OWLS-XPlan, which has been known as an effective application for service composition.

• Journal of KIISE:Databases
• /
• v.37 no.6
• /
• pp.333-342
• /
• 2010

Building ontology is the key factor to construct semantic web. However, this is time-consuming process. Hence, there are several approaches which automatically generate the ontologies from relational databases. Current studies on the automatic generation of the ontologies from relational database are focused on generating the ontology by analyzing the database schema and stored data. These studies generate the ontology by analyzing only tables and constraints in the schema and ignore view definitions. However, view definitions are defined by a database designer considering the domain of the database. Hence, by considering view definitions, additional classes and hierarchical relationships can be generated. And these are useful in answering queries and integration of ontologies. In this paper, we formalize the generation of classes and hierarchical relationships by analyzing existing methods, and we propose the method which generates additional classes and hierarchical relationships by analyzing view definitions. Finally, we analyze the generated ontology by applying our method to synthetic data and real-world data. We show that our method generates meaningful classes and hierarchical relationships using view definitions.

• Proceedings of the Korean Society for Agricultural Machinery Conference
• /
• 2017.04a
• /
• pp.162-162
• /
• 2017

구글에서 공개한 Tensorflow를 이용한 여러 학문 분야의 연구가 활발하다. 농업 시설환경을 대상으로 한 빅데이터의 축적이 증가함과 아울러 실효적인 정보 획득을 위한 각종 데이터 분석 및 마이닝 기법에 대한 연구 또한 활발한 상황이다. 한편, 타 분야의 성공적인 심층학습기법 응용사례에 비하여 농업 분야에서의 응용은 초기 성장 단계라 할 수 있다. 이는 농업 현장에서 취득한 정보의 난해성 및 완성도 높은 생육/환경 모델링 정보의 부재로 실효적인 전과정 처리 기술 도출에 소요되는 시간, 비용, 연구 환경이 상대적으로 부족하기 때문일 것이다. 특히, 센서 기반 데이터 취득 기술 증가에 따라 비약적으로 방대해진 수집 데이터를 시간 복잡도가 높은 심층 학습 모델링 연산에 기계적으로 단순 적용할 경우 시간 효율적인 측면에서 성공적인 결과 도출에 애로가 있을 것이다. 매우 높은 시간 복잡도를 해결하기 위하여 제시된 하드웨어 가속 기능의 경우 일부 개발환경에 국한이 되어 있다. 일례로, 구글의 Tensorflow는 오픈소스 기반 병렬 클러스터링 기술인 MPICH를 지원하는 알고리즘을 공개하지 않고 있다. 따라서, 본 연구에서는 심층학습 기법 연구에 있어서, 예상 가능한 다양한 자원을 활용하여 최대한 연산의 결과를 빨리 도출할 수 있는 하드웨어적인 접근 방법을 모색하였다. 호스트에서 수행하는 일방적인 학습 알고리즘과 달리 이기종간 심층 학습이 가능하기 위해선 우선, NFS(Network File System)를 이용하여 데이터 계층이 상호 연결이 되어야 한다. 이를 위해서 고속 네트워크를 기반으로 한 NFS의 이용이 필수적이다. 둘째로 제한된 자원의 한계를 극복하기 위한 메모 공유 라이브러리가 필요하다. 셋째로 이기종간 프로세서에 최적화된 병렬 처리용 컴파일러를 이용해야 한다. 가장 중요한 부분은 이기종간의 처리 능력에 따른 작업을 고르게 분배할 수 있는 작업 스케쥴링이 수행되어야 하며, 이는 처리하고자 하는 데이터의 형태에 따라 매우 가변적이므로 해당 데이터 도메인에 대한 엄밀한 사전 벤치마킹이 수행되어야 한다. 이러한 요구조건을 대부분 충족하는 Open-CL ver1.2(https://www.khronos.org/opencl/)를 이용하였다. 최신의 Open-CL 버전은 2.2이나 본 연구를 위하여 준비한 4가지 이기종 시스템에서 모두 공통적으로 지원하는 버전은 1.2이다. 실험적으로 선정된 4가지 이기종 시스템은 1) Windows 10 Pro, 2) Linux-Ubuntu 16.04.4 LTS-x86_64, 3) MAC OS X 10.11 4) Linux-Ubuntu 16.04.4 LTS-ARM Cortext-A15 이다. 비교 분석을 위하여 NVIDIA 사에서 제공하는 Pascal Titan X 2식을 SLI로 구성한 시스템을 준비하였다. 개별 시스템에서 별도로 컴파일 된 바이너리의 이름을 통일하고, 개별 시스템의 코어수를 동일하게 균등 배분하여 100 Hz의 데이터로 입력이 되는 온도 정보와 조도 정보를 입력으로 하고 이를 습도정보에 Linear Gradient Descent Optimizer를 이용하여 Epoch 10,000회의 학습을 수행하였다. 4종의 이기종에서 총 32개의 코어를 이용한 학습에서 17초 내외로 연산 수행을 마쳤으나, 비교 시스템에서는 11초 내외로 연산을 마치는 결과가 나왔다. 기보유 하드웨어의 적절한 활용이 가능한 심층학습 기법에 대한 연구를 지속할 것이다

• Journal of Internet Computing and Services
• /
• v.16 no.6
• /
• pp.89-95
• /
• 2015

Traditional frequent pattern mining discovers valid patterns with no smaller frequency than a user-defined minimum threshold from databases. In this framework, an enormous number of patterns may be extracted by a too low threshold, which makes result analysis difficult, and a too high one may generate no valid pattern. Setting an appropriate threshold is not an easy task since it requires the prior knowledge for its domain. Therefore, a pattern mining approach that is not based on the domain knowledge became needed due to inability of the framework to predict and control mining results precisely according to the given threshold. Top-k frequent pattern mining was proposed to solve the problem, and it mines top-k important patterns without any threshold setting. Through this method, users can find patterns from ones with the highest frequency to ones with the k-th highest frequency regardless of databases. In this paper, we provide knowledge both on frequent and top-k pattern mining. Although top-k frequent pattern mining extracts top-k significant patterns without the setting, it cannot consider both item quantities in transactions and relative importance of items in databases, and this is why the method cannot meet requirements of many real-world applications. That is, patterns with low frequency can be meaningful, and vice versa, in the applications. High utility pattern mining was proposed to reflect the characteristics of non-binary databases and requires a minimum threshold. Recently, top-k high utility pattern mining has been developed, through which users can mine the desired number of high utility patterns without the prior knowledge. In this paper, we analyze two algorithms related to top-k high utility pattern mining in detail. We also conduct various experiments for the algorithms on real datasets and study improvement point and development direction of top-k high utility pattern mining through performance analysis with respect to the experimental results.