• Database Research
• /
• v.34 no.3
• /
• pp.3-13
• /
• 2018

Text classification is one of the text mining technologies that classifies a given textual document into its appropriate categories and is used in various fields such as spam email detection, news classification, question answering, emotional analysis, and chat bot. In general, the text classification system utilizes machine learning algorithms, and among a number of algorithms, naïve Bayes and support vector machine, which are suitable for text data, are known to have reasonable performance. Recently, with the development of deep learning technology, several researches on applying deep neural networks such as recurrent neural networks (RNN) and convolutional neural networks (CNN) have been introduced to improve the performance of text classification system. However, the current text classification techniques have not yet reached the perfect level of text classification. This paper focuses on the fact that the text data is expressed as a vector only with the word dimensions, which impairs the semantic information inherent in the text, and proposes a neural network architecture based upon the semantic tensor space model.

• Journal of the Korea Society of Computer and Information
• /
• v.26 no.3
• /
• pp.29-34
• /
• 2021

With the advent of the 4th industrial revolution era, IT technology is creating new services that have not existed by converging with various existing industries and fields. In particular, in the field of artificial intelligence, chatbots and the latest technologies have developed dramatically with the development of natural language processing technology, and various business processes are processed through chatbots. This study is a study on a system that provides a close answer to the question the user wants to find by creating a structural form for legal inquiries through Slot Filling-based chatbot technology, and inputting a predetermined type of question. Using the proposal system, it is possible to construct question-and-answer data in a more structured form of legal information, which is unstructured data in text form. In addition, by managing the accumulated Q&A data through a big data storage system such as Apache Hive and recycling the data for learning, the reliability of the response can be expected to continuously improve.

• Journal of ILASS-Korea
• /
• v.25 no.4
• /
• pp.162-169
• /
• 2020

The worldwide focus on reducing the emissions, fuel and lubricant consumption in T-GDI engines is leading engineers to consider the crankcase ventilation and oil mist separation system as an important means of control. In today's passenger cars, the oil mist separation systems mainly use the inertia effect (e.g. labyrinth, cyclone etc.). Therefore, this study has investigated high efficiency cylinder head-integrated oil-mist separator by using a compact multi-impactor type oil mist separator system to ensure adequate oil mist separation performance. For this purpose, engine dynamometer testing with oil particle efficiency measurement equipment and 3D two-phase flow simulation have been performed for various engine operating conditions. Tests with an actual engine on a dynamometer showed oil aerosol particle size distributions varied depending on operating conditions. For instance, high rpm and load increases bot only blow-by gases but the amount of small size oil droplets. Submicron-sized particles (less than 0.5 ㎛) were also observed. It is also found that the impactor type separator is able to separate nearly no droplets of diameter lower than 3 ㎛. CFD results showed that the complex aerodynamics processes that lead to strong impingement and break-up can strip out large droplets and generate more small size droplets.

• Journal of the Korean Society of Manufacturing Process Engineers
• /
• v.20 no.12
• /
• pp.30-35
• /
• 2021

In this study, we developed an automatic module changer for agricultural implements for using in unmanned agricultural robots. An automatic module changer is attached by lowering from the top to bottom of the implements and fixing the four fastener bars attached to the implements in combination. The lift function was implemented using seesaw-type structures to keep the engagement point constant when the automatic module changer climbs and descends, and the switching function of the automatic module changer was implemented using the link device in the cam structure. We developed an algorithm to check the presence of attachment and opening/closing of the workpiece using limit switches and verified the performance through combination assessment and weight lift test to assess whether the combination was within the error range.

• Journal of Advanced Navigation Technology
• /
• v.26 no.5
• /
• pp.312-318
• /
• 2022

Currently, the Republic of Korea is facing the problem of a decrease in military service resources due to the demographic cliff, and is pursuing military restructuring and changes in the military force structure in order to respond to this. In this situation, the Army is pushing forward the deployment of a drone-bot combat system that will lead the future battlefield. The battlefield of the future will be changed into an integrated battlefield concept that combines command and control, surveillance and reconnaissance, and precision strike. According to these changes, unmanned combat system, including dronebots, will be widely applied to combat situations that are high risk and difficult for humans to perform in actual combat. In this paper, as one of the countermeasures to these changes, autonomous behavior software with a BDI architecture-based decision-making system was developed. The autonomous behavior software applied a framework structure to improve applicability to multiple models. Its function was verified in a PC-based environment by assuming that the target UAV is a battalion-level surveillance and reconnaissance UAV.

• Steel and Composite Structures
• /
• v.43 no.6
• /
• pp.821-837
• /
• 2022

This research is devoted to study the effects of humidity and temperature on the bending behavior of functionally graded (FG) ceramic-metal porous plates resting on Pasternak elastic foundation using a quasi-3D hyperbolic shear deformation theory developed recently. The present plate theory with only four unknowns, takes into account both transverse shear and normal deformations and satisfies the zero traction boundary conditions on the surfaces of the functionally graded plate without using shear correction factors. Material properties of porous FG plate are defined by rule of the mixture with an additional term of porosity in the through-thickness direction. The governing differential equations are obtained using the "principle of virtual work". Analytically, the Navier method is used to solve the equations that govern a simply supported FG porous plate. The obtained results are checked by comparing the results determined for the perfect and imperfect FG plates with those available in the scientific literature. Effects due to material index, porosity factors, moisture and thermal loads, foundation rigidities, geometric ratios on the FG porous plate are all examined. Finally, this research will help us to design advanced functionally graded materials to ensure better durability and efficiency for hygro-thermal environments.

• Journal of the Korean Geotechnical Society
• /
• v.22 no.8
• /
• pp.63-72
• /
• 2006

In the ASTM and ISRM, an uniaxial compressive strength(${\sigma}_{c}$) has been estimated to be 23(ASTM) or $20{\sim}25$(ISRM) times of point load strength index using a diametral test regardless of the rock rating or geological conditions. This paper presents a relationship between $I_{s}$ and ${\sigma}_{c}$ of a weak sedimentary rocks on Ulsan of the Kyung-Sang Basin in Korea. In the results of 291 for ${\sigma}_{c}$ test and 2310 for $I_{s}$ test from 77 spots, the predicted errors of ${\sigma}_{c}$ determined by strength ratio of ${\sigma}_{c}/I_{s}$ have been relatively less than those determined by linear regression analysis. And in case of weak sedimentary rocks such as mudstones, shales and sandstones, ${\sigma}_{c}$ should be lower than those suggested by ISRM and ASTM.

• Convergence Security Journal
• /
• v.22 no.1
• /
• pp.19-27
• /
• 2022

With the innovative development of ICT technology, hacking techniques of hackers are also evolving into sophisticated and intelligent hacking techniques. Threat detection research to counter these cyber threats was mainly conducted in a passive way through hacking damage investigation and analysis, but recently, the importance of cyber threat information collection and analysis is increasing. A bot-type automation program is a rather active method of extracting malicious code by visiting a website to collect threat information or detect threats. However, this method also has a limitation in that it cannot prevent hacking damage because it is a method to identify hacking damage because malicious code has already been distributed or after being hacked. Therefore, to overcome these limitations, we propose a model that detects actual threats by acquiring and analyzing threat information while identifying and managing cyber bases. This model is an active and proactive method of collecting threat information or detecting threats outside the boundary such as a firewall. We designed a model for detecting threats using cyber strongholds and validated them in the defense environment.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2022.05a
• /
• pp.211-211
• /
• 2022

The marketization reform from the open-door policy in 1978 was not only booming export-oriented industries with foreign investment but also expanding the role of private actors in the Chinese water sector. Private Sector Participation (PSP) has become an important element in developing urban infrastructure by providing better services with advanced facilities. The rapid development of PSP-driven urban water infrastructure in China has a positive impacted on Chinese economic development, particularly in coastal areas. PPPs in some coastal areas have successfully spread out over China since China applied the first Build-Operate-Transfer (BOT) mode in the water sector in the early 1990s. The market-oriented water and wastewater, Public-Private Partnership (PPP) mechanism in the initial period of China has been transformed into a state-dominated PPP mechanism. The development pattern of the water and wastewater PPPs in China has been divided in four stages: the first period from 1984 to 2002, the second period from 2003 to 2008, the third period from 2009 to 2014, and the last period after 2015. The study aims to investigate the successful process of water and wastewater PPPs in local areas through five socioeconomic elements: export-oriented economic strategy, urbanization, cheap land policy, infrastructure investment, and water issues and climate change. In addition, the study focuses on analyzing the extent to which the Chinese government re-asserted its control over the PPP mechanism by classifying five elements in three different development Phases from early 2000 to 2020. The Jiangsu Province in the estern coastal area has actively invited PPP projects in the water and wastewater sectors. The successful introduction and rapid growth of PPPs in the urban water infrastructure has made the province an attractive area for a foreign investor.

• Journal of Internet of Things and Convergence
• /
• v.9 no.6
• /
• pp.11-16
• /
• 2023

The technology of 3D reconstruction, primarily relying on point cloud data, is essential for digitizing objects or spaces. This paper aims to utilize reinforcement learning to achieve the acquisition of point clouds in a given environment. To accomplish this, a simulation environment is constructed using Unity, and reinforcement learning is implemented using the Unity package known as ML-Agents. The process of point cloud acquisition involves initially setting a goal and calculating a traversable path around the goal. The traversal path is segmented at regular intervals, with rewards assigned at each step. To prevent the agent from deviating from the path, rewards are increased. Additionally, rewards are granted each time the agent fixates on the goal during traversal, facilitating the learning of optimal points for point cloud acquisition at each traversal step. Experimental results demonstrate that despite the variability in traversal paths, the approach enables the acquisition of relatively accurate point clouds.