• Journal of Intelligence and Information Systems
• /
• v.29 no.1
• /
• pp.249-263
• /
• 2023

Collaborative Filtering, a representative recommendation system methodology, consists of two approaches: neighbor methods and latent factor models. Among these, the latent factor model using matrix factorization decomposes the user-item interaction matrix into two lower-dimensional rectangular matrices, predicting the item's rating through the product of these matrices. Due to the factor vectors inferred from rating patterns capturing user and item characteristics, this method is superior in scalability, accuracy, and flexibility compared to neighbor-based methods. However, it has a fundamental drawback: the need to reflect the diversity of preferences of different individuals for items with no ratings. This limitation leads to repetitive and inaccurate recommendations. The Adaptive Deep Latent Factor Model (ADLFM) was developed to address this issue. This model adaptively learns the preferences for each item by using the item description, which provides a detailed summary and explanation of the item. ADLFM takes in item description as input, calculates latent vectors of the user and item, and presents a method that can reflect personal diversity using an attention score. However, due to the requirement of a dataset that includes item descriptions, the domain that can apply ADLFM is limited, resulting in generalization limitations. This study proposes a Generalized Adaptive Deep Latent Factor Recommendation Model, G-ADLFRM, to improve the limitations of ADLFM. Firstly, we use item ID, commonly used in recommendation systems, as input instead of the item description. Additionally, we apply improved deep learning model structures such as Self-Attention, Multi-head Attention, and Multi-Conv1D. We conducted experiments on various datasets with input and model structure changes. The results showed that when only the input was changed, MAE increased slightly compared to ADLFM due to accompanying information loss, resulting in decreased recommendation performance. However, the average learning speed per epoch significantly improved as the amount of information to be processed decreased. When both the input and the model structure were changed, the best-performing Multi-Conv1d structure showed similar performance to ADLFM, sufficiently counteracting the information loss caused by the input change. We conclude that G-ADLFRM is a new, lightweight, and generalizable model that maintains the performance of the existing ADLFM while enabling fast learning and inference.

• Journal of the Korea Institute of Building Construction
• /
• v.19 no.1
• /
• pp.9-18
• /
• 2019

Concrete, which is the most widely used building material in modern times, has been improved not only in strength but also in structural performance such as increase in toughness and ductility, weight reduction, and improvement in quality of human life. Due to the surge in demand for the building, there is a tendency to be used variously from architectural panel and architecture to interior accessories. In Korea, a light-transmitting concrete, LEFC(Light Emotion Friendly Concrete), that insert plastic rods to stimulate emotional sensation through the combination of light and concrete has developed. In previous research, it was confirmed that the use of a synthetic foam agent rather than an animal foam agent did not cause a fogging phenomenon. In this study, lightweight by applying foaming agent to LEFC and two types of fiber (Nylon Fiber, Polyvinyl Alcohol) were compared to achieve to investigate the fiber to be applied in future. An equation that can predict the loss and adhesion reduction of the concrete section according to the diameter of the rod (5mm, 10mm) and the interval (10mm, 15mm, 20mm) was proposed.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.24 no.1
• /
• pp.89-95
• /
• 2024

Nowadays, the competitiveness of a company depends on the ability of all organizational members to share and utilize the organizational knowledge accumulated by the organization. As if to prove this, the world is now focusing on ChetGPT service using generative AI technology based on LLM (Large Language Model). However, it is still difficult to apply the ChetGPT service to work because there are many hallucinogenic problems. To solve this problem, sLLM (Lightweight Large Language Model) technology is being proposed as an alternative. In order to construct sLLM, corporate data is essential. Corporate data is the organization's ERP data and the company's office document knowledge data preserved by the organization. ERP Data can be used by directly connecting to sLLM, but office documents are stored in file format and must be converted to data format to be used by connecting to sLLM. In addition, there are too many technical limitations to utilize office documents stored in file format as organizational knowledge information. This study proposes a method of storing office documents in DB format rather than file format, allowing companies to utilize already accumulated office documents as an organizational knowledge system, and providing office documents in data form to the company's SLLM. We aim to contribute to improving corporate competitiveness by combining AI technology.

• Korean Journal of Remote Sensing
• /
• v.39 no.5_3
• /
• pp.997-1008
• /
• 2023

Ship detection at sea can be performed in various ways. In particular, satellites can provide wide-area surveillance, and Synthetic Aperture Radar (SAR) imagery can be utilized day and night and in all weather conditions. To propose an efficient ship detection method from SAR images, this study aimed to apply the You Only Look Once Version 5 (YOLOv5) model to Sentinel-1 images and to analyze the difference between individual vs. integrated models and the accuracy characteristics by polarization. YOLOv5s, which has fewer and lighter parameters, and YOLOv5x, which has more parameters but higher accuracy, were used for the performance tests (1) by dividing each polarization into HH, HV, VH, and VV, and (2) by using images from all polarizations. All four experiments showed very similar and high accuracy of 0.977 ≤ AP@0.5 ≤ 0.998. This result suggests that the polarization integration model using lightweight YOLO models can be the most effective in terms of real-time system deployment. 19,582 images were used in this experiment. However, if other SAR images,such as Capella and ICEYE, are included in addition to Sentinel-1 images, a more flexible and accurate model for ship detection can be built.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.19 no.10
• /
• pp.136-142
• /
• 2018

Additive manufacturing process technology, known as the 3D printing process, is expanding its utilization from simple model realization to commercialized part production based on continuous material development. Recently, research and development have been actively carried out to fabricate lightweight and high-strength parts using polymers, such as polyamide (polyamide), which is a high-strength engineering plastic material. In this study, the Izod impact characteristics were analyzed for polyamide 12 (PA12) materials. For the specimen production, selective laser sintering process technology, which has excellent mechanical properties of finished products, was applied. In addition, PA12 and glass bead reinforced PA12 materials were produced. The specimens were classified according to the production direction on the production platform, and each specimen was subjected to an Izod test at test temperatures of $-25^{\circ}C$, $25^{\circ}C$, and $60^{\circ}C$. As a result, the impact strength of PA12 and glass bead-reinforced PA12 of vertical direction specimens were 48.8% and 16.3% lower than those of the parallel specimens at a $25^{\circ}C$ test temperature and the impact strength of parallel specimens was improved by 46.5% and 20.4% at a test temperature of $60^{\circ}C$ compared to that at $-25^{\circ}C$.

• Journal of Korean Society of Steel Construction
• /
• v.18 no.6
• /
• pp.793-804
• /
• 2006

Various PSC and steel-concrete composite railway bridges are being developed for short-medium spans with structural and economic efficiency. According to the design concept, the prestressed composite girder bridge has the advantages of being lightweight and having low girder depth, with the capacity for long spans. However, the dynamic behavior under a passing train is one of the critical issues concerning these railway bridges designed with more flexibility. Therefore, it is very important to evaluate the modal parameters before performing dynamic analyses. In this paper, real-scale prestressed composite girders were fabricated as a test model and modal testing was carried out to evaluate modal parameters including natural frequency and modal damping ratio. During the modal testing, a digitally controlled vibration exciter as well as an impact hammer was applied to obtain frequency-response functions, and the modal parameters were also evaluated after the fracture of test models. With application of reliable properties from modal tests, the estimation of dynamic performances of prestressed composite girder railway bridges can be obtained from various parametric studies on dynamic behavior under the passage of a moving train.

• Journal of the Society of Naval Architects of Korea
• /
• v.54 no.2
• /
• pp.151-160
• /
• 2017

Growing demand and rapid development of the synthetic fiber rope in mooring system have taken place since it has been used in deep water platform lately. Unlike a chain mooring, synthetic fiber rope composed of lightweight materials such as Polyester(polyethylene terephthalate), HMPE(high modulus polyethylene) and Aramid(aromatic polyamide). Non-linear stiffness and another failure mode are distinct characteristics of synthetic fiber rope when compared to mooring chain. When these ropes are exposed to environmental load for a long time, the length of rope will be increased permanently. This is called 'the creep phenomenon'. Due to the phenomenon, The initial characteristics of mooring systems would be changed because the length and stiffness of the rope have been changed as time goes on. The changed characteristics of fiber rope cause different mooring tension and vessel offset compared to the initial design condition. Commercial mooring analysis software that widely used in industries is unable to take into account this phenomenon automatically. Even though the American Petroleum Institute (API) or other classification rules present some standard or criteria with respect to length and stiffness of a mooring line, simulation guide considers the mechanical properties that is not mentioned in such rules. In this paper, the effect of creep phenomenon in the fiber rope mooring system under specific environment condition is investigated. Desiged mooring system for a Mobile Offshore Drilling Unit(MODU) with HMPE rope which has the highest creep is analyzed in a time domain in order to investigate the effects creep phenomenon to vessel offset and mooring tension. We have developed a new procedure to an analysis of mooring system reflecting the creep phenomenon and it is validated through a time domain simulation using non-linear mooring analysis software, OrcaFlex. The result shows that the creep phenomenon should be considered in analysis procedure because it affects the length and stiffness of synthetic fiber rope in case of high water temperature and permanent mooring system.

• Journal of Korean Society of Steel Construction
• /
• v.17 no.4 s.77
• /
• pp.499-509
• /
• 2005

The excessive wind-induced motion of tall buildings most frequently result from vortex-shedding-induced across-wind oscillations. This form of excitation is most pronounced for relatively flexible, lightweight, and lightly damped high-rise buildings with constant cross-sections. This paper discusses the aerodynamic means ofmitigating the across-wind vortex shedding induced in such situations. Openings are added in both the drag and lift directions in the buildings to provide pressure equalization. Theytend to reduce the effectiveness of across-wind forces by reducing their magnitudes and disrupting their spatial correlation. The effects of buildings with several geometries of openings on aerodynamic excitations and displacement responses have been studied for high-rise buildings with square cross-sections and an aspect ratio of 8:1 in a wind tunnel. High-frequency force balance testshave been carried out at the Kumoh National University of Technology using rigid models with 24 kinds of opening shapes. The measured model's aerodynamic excitations and displacement were compared withthose of a square cylinder with no openings to estimate the effectiveness of openings for wind-induced oscillations. From these results, theopening shape, size, and location of buildings to reduce wind-induced vortex shedding and responses were pointed out.

• Composites Research
• /
• v.32 no.5
• /
• pp.284-289
• /
• 2019

In the present paper, the dynamic force-moment equilibrium equations, driving power and energy equations are analyzed to formulate the equation for fuel economy(km/liter) equivalent to the driving distance (km) divided by the fuel volume (liter) of the vehicle, a selected model of gasoline powered KIA K3 (1.6v). In addition, the effects of the dynamic parameters such as speed of vehicle (V), vehicle total weight(M), rolling resistance ($C_r$) between tires and road surface, inclined angle of road (${\theta}$), as well as the aerodynamic parameters such as drag coefficient ($C_d$) of vehicle, air density(${\rho}$), cross-sectional area (A) of vehicle, wind speed ($V_w$) have been analyzed. And the possibility of alternative materials such as lightweight metal alloys, fiber reinforced plastic composite materials to replace the conventional steel and casting iron materials and to reduce the weight of the vehicle has been investigated by Ashby's material index method. Through studies, the following results were obtained. The most influencing parameters on the fuel economy at high speed zone (100 km/h) were V, the aerodynamic parameters such as $C_d$, A, ${\rho}$, and $C_r$ and M. While at low speed zone (60 km/h), they are, in magnitude order, dynamic parameters such as V, M, $C_r$ and aerodynamic ones such as $C_d$, A, and ${\rho}$, respectively.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.17 no.3
• /
• pp.121-129
• /
• 2024

The rapid advancement of digital technology and the COVID-19 pandemic have significantly accelerated the growth of online commerce, highlighting the need for support mechanisms that enable small business owners to effectively respond to these market changes. In response, this paper presents a foundational technology leveraging the Online to Offline (O2O) strategy to automatically capture products displayed on retail shelves and utilize these images to create virtual stores. The essence of this research lies in precisely identifying and recognizing the location and names of displayed products, for which a single-class-targeted, lightweight model based on YOLOv8, named ESD-YOLOv8, is proposed. The detected products are identified by their names through feature-point-based technology, equipped with the capability to swiftly update the system by simply adding photos of new products. Through experiments, product name recognition demonstrated an accuracy of 74.0%, and position detection achieved a performance with an F2-Score of 92.8% using only 0.3M parameters. These results confirm that the proposed method possesses high performance and optimized efficiency.