• Proceedings of the Korea Information Processing Society Conference
• /
• 2021.11a
• /
• pp.626-628
• /
• 2021

사건 관계 추출 태스크는 구조화되지 않은 텍스트 데이터에서 사건의 구조화된 표현을 얻는 것이다. 하나의 문장에서도 많은 정보를 얻을 수 있는 중요한 태스크임에도 불구하고, 다양한 사전 학습 모델을 적용한 연구는 아직 활발하게 연구되지 않고 있다. 따라서 본 연구에서 사전 학습된 모델의 임베딩 기법 중 BERT, RoBERTa, SpanBERT에 각각 base, large 아키텍처를 적용하여 실험하였다. 사건을 식별하기 위한 trigger와 해당 trigger의 세부 argument를 식별하기 위한 분류기를 상위레이어로 각각 설계하였고, 다양한 배치 크기를 적용하여 실험하였다. 성능평가는 trigger/argument 각각 F1 score를 적용하였고, 결과는 RoBERTa large 모델에서 좋은 성능을 보인 것을 확인하였다.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.8 no.3
• /
• pp.682-696
• /
• 2007

Recently the convergence of Bio technology(BT) and Information technology(IT) based on Nano technology has just emerged into several types of convergence technologies, therefore, this research paper examined and suggested the needs of why BT, IT and NT should be converged into new types technologies through analyzing the existing papers and documents. Among the several types of convergence technologies, this study focused on the types of BIT and BNT only through Delphi research method, and finally laid out the most important 7 specific kinds of technologies of BIT and 10 of BNT, which will give insights of cross-understanding and cross-imagination to institute side, business side, and academic side when they do research and business development of future convergence technologies together as cooperation between.

• Journal of Korea Multimedia Society
• /
• v.18 no.12
• /
• pp.1432-1438
• /
• 2015

We present a methodology to detect moving shadows in video sequences, which is considered as a challenging and critical problem in the most visual surveillance systems since 1980s. While most previous moving shadow detection methods used hand-crafted features such as chromaticity, physical properties, geometry, or combination thereof, our method can automatically learn features to classify whether image segments are shadow or foreground by using a deep learning architecture. Furthermore, applying Markov Random Field enables our system to refine our shadow detection results to improve its performance. Our algorithm is applied to five different challenging datasets of moving shadow detection, and its performance is comparable to that of state-of-the-art approaches.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.42 no.3
• /
• pp.698-701
• /
• 2017

The district heating system supports currently 1.3 million of houses in South Korea, and it is composed of two main parts: provision and distribution. For the provision part, Korea District Heating Corporation supervises the system with proper devices and experts, whereas for the distribution parts, there is no main supervision system and it depends highly on experts although it is very difficult to find the proper experts. In this paper, we consider IoT based remote management system for district heating distribution to properly supervise heating distribution by experts and/or programs remotely.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.13 no.2
• /
• pp.101-107
• /
• 2018

As indoor pollutants such as carbon dioxide and dust mainly affect the respiratory and circulatory systems, there is an increasing need for real-time indoor / outdoor environmental monitoring. In this paper, we have developed a real - time environmental monitoring system using the cloud-based 3G integrated environmental sensor module for environmental monitoring. A highly reliable environmental information monitoring system requires various IT technologies such as infrastructure (server, commercial software, etc.), service application software, security, and authentication. A real-time environment monitoring system based on cloud service that can provide reliable service satisfying these configuration requirements is proposed and implemented. It is expected that this system can be applied to various technologies such as indoor automatic window opening/closing system based on the Internet.

• Composites Research
• /
• v.32 no.2
• /
• pp.96-101
• /
• 2019

Interfacial adhesion between fiber and resin are related to composites performance, so it is very important to evaluate them accurately. In this study, the interfacial properties of microdroplets under fatigue loading conditions were evaluated. The mechanical properties and interfacial adhesion of epoxy resin with dopamine were studied. Tensile specimens were prepared to evaluate mechanical properties and epoxy microdroplets specimens were used for the evaluation of interfacial adhesion. In addition, in the microdroplet fatigue test, the same diameter of the microdroplet was used and the experiment was performed under the same conditions. As a result, it was confirmed that mechanical and interfacial properties were improved when dopamine was applied to epoxy resin through tensile and microdroplet experiments. It is considered that dopamine improves the degree of curing of the epoxy resin and imparts hydroxyl groups to the epoxy resin to increase the mechanical properties and the interfacial adhesion between the glass fibers.

• Journal of the Ergonomics Society of Korea
• /
• v.30 no.5
• /
• pp.651-657
• /
• 2011

Objective: The aim of this study is to investigate the system to integrate and manage the in-vehicle interactions between the drivers and the in-vehicle mobile IT devices. Background: As the mobile IT technology is being used anywhere, the drivers are interacting with the mobile IT device on driving situations. The distraction of the driver's attention causes the car accidents. It is necessary to develop the HVI(Human Vehicle Interface System) to integrate and manage the in-vehicle interactions with IT devices. Method: The HVI System is designed not as the interfacing subject but as the supervising system to monitor the driver's status and support the driver to concentrate on the primary tasks. The HVI system collects the status information of the car and driver and estimate the driving workload. Results: The HVI system controls how to provide the output information based on the driving workload. We implemented the HVI system prototype and applied in the real vehicle with the HVI cell phone and the HVI car navigation system. Conclusion: Depending on the driving situations, the HVI system prevented the information output in dangerous situation and diversified the modality and the intensity of the output information. Application: We will extend the HVI system to be connected the other various IT devices and verity the effectiveness of the system through various experiments.

• Composites Research
• /
• v.31 no.5
• /
• pp.186-191
• /
• 2018

Condition and properties of composites with different chemical structure of epoxy matrix were observed after saline solution treatment. Epoxy was used as matrix and the flexibility was controlled by using 2 typed-epoxies and 3 types hardeners (amine, acid anhydride and amide). Saline water treatment was conducted with 6 wt% NaCl solution at $60^{\circ}C$ for 0, 15, and 30 days. Cross section was observed and interfacial and mechanical and properties was evaluated. Amine type exhibited the highest crosslinking density and mechanical and interfacial properties whereas water absorbance was lowest. It is because that the water molecules can be hardly penetrate into the epoxy matrix or the interface between epoxy and glass fiber and it leads to saline water resistance of composites.

• Journal of information and communication convergence engineering
• /
• v.17 no.1
• /
• pp.84-90
• /
• 2019

Applications of unmanned aerial vehicles (UAVs) in the military environment have become popular because they require minimum human contribution and can avoid accidents during missions. UAVs are employed in various missions such as reconnaissance, observation, aggression, and protection. Consequently, counter-measures, known as anti-drone technologies, have been developed as well. In order to protect against threats from anti-drone technologies and enhance the survivability of UAVs, this study proposes an evasive measure. The proposed bio-inspired evasive maneuver of a UAV mimics a dragonfly's irregular flight. The unpredictable UAV movement is able to confuse enemies and avoid threats, thereby enhancing the UAV's survivability. The proposed system has been implemented on a commercial UAV platform (AR Drone 2.0) and tested in a real environment. The experiment results demonstrate that the proposed flight pattern has larger displacement values compared to a regular flight maneuver, thus making the UAV's position is difficult to predict.

• Journal of Information Technology Services
• /
• v.13 no.1
• /
• pp.57-69
• /
• 2014

IT convergence technology can boost the advancement of weapon systems and efficient operation for the military reformation in Korea. This study investigated IT firms' concern about the defense acquisition process for IT-convergent systems and the role of IT firms in the process with Focus Group Interview and in-depth interview on executives of IT firms, and classified the concern to three major issues. They are the clarification of IT firms' role in the defense IT-convergent efforts, the establishment of the control tower for defense IT convergence, and reasonable estimation of software development cost in the defense acquisition process.