• 지능정보연구
• /
• 제29권4호
• /
• pp.129-164
• /
• 2023

최근 생성형 AI 기술의 발전으로 인해 대형 언어 모델(Large Language Model, LLM)의 활용 및 도입이 확대되고 있는 상황에서 기존 연구들은 기업내부 데이터의 활용에 대한 실제 적용사례나 구현방법을 찾아보기 힘들다. 이에 따라 본 연구에서는 가장 많이 이용되고 있는 LangChain 프레임워크를 이용한 LLM 애플리케이션 아키텍처를 활용하여 생성형 AI 서비스를 구현하는 방법을 제시한다. 이를 위해 LLM의 활용을 중심으로, 정보 부족 문제를 극복하는 다양한 방법을 검토하고 구체적인 해결책을 제시하였다. 이를 위해 파인튜닝이나 직접 문서 정보를 활용하는 방법을 분석하며, 이러한 문제를 해결하기 위한 RAG 모델을 활용한 정보 저장 및 검색 방법에 대해 주요단계에 대해 자세하게 살펴본다. 특히, RAG 모델을 활용하여 정보를 벡터저장소에 저장하고 검색하기 위한 방법으로 유사문맥 추천 및 QA시스템을 활용하였다. 또한 구체적인 작동 방식과 주요한 구현 단계 및 사례를 구현소스 및 사용자 인터페이스까지 제시하여 생성형 AI 기술에 대한 이해를 높였다. 이를 통해 LLM을 활용한 기업내 서비스 구현에 적극적으로 활용할 수 있도록 하는데 의미와 가치가 있다.

• Industrial Engineering and Management Systems
• /
• 제11권2호
• /
• pp.165-169
• /
• 2012

There has been a growing interest in firms' environmental sustainability activities to improve environmental practices in their supply chain. This study aims to deal with supplier evaluation of firm's green supply chain management (GSCM) criteria with incomplete information. Nevertheless, the suitable supplier is a key strategic direction in eliminating environmental impact on supply chain management for manufacturing firms. The firm's GSCM criteria and supplier selection need to be unified as a system to improve the firm's performance.

• 한국신뢰성학회지:신뢰성응용연구
• /
• 제17권3호
• /
• pp.188-196
• /
• 2017

Purpose: The purpose of this research is to determine optimal replacement age using non-informative prior information and Bayesian method. Methods: We propose a novel approach using Bayesian method to determine the optimal replacement age in block replacement policy by defining the prior probability with data on failure time and repair time. The Marcov Chain Monte Carlo simulation is used to investigate the asymptotic distribution of posterior parameters. Results: An optimal replacement age of block replacement policy is determined which minimizes cost and nonoperating time when no information on prior distribution of parameters is given. Conclusion: We find the posterior distribution of parameters when lack of information on prior distribution, so that the optimal replacement age which minimizes the total cost and maximizes the total values is determined.

• Industrial Engineering and Management Systems
• /
• 제13권1호
• /
• pp.15-28
• /
• 2014

Evaluation of the implementation and performance of a firm's green supply chain management (GSCM) is an ongoing process. Balanced scorecard is a multi-criteria evaluation concept that highlights implementation and performance measures. The literature on the framework is abundant literature but scarce on how to build a hierarchical framework under uncertainty with dependence relations. Hence, this study proposes a hybrid approach, which includes applied interpretive structural modeling to build a hierarchical structure and uses the analytic network process to analyze the dependence relations. Additionally, this study applies the fuzzy set theory to determine linguistic preferences. Twenty dependence criteria are evaluated for a GSCM implemented firm in Taiwan. The result shows that the financial aspect and life cycle assessment are the most important performance and weighted criteria.

• 한국기록관리학회지
• /
• 제24권2호
• /
• pp.89-112
• /
• 2024

기록관리에서 메타데이터는 기록을 구성하는 필수 요소 중 하나로 기록물을 적절하게 관리하고 이해하도록 하는데 매우 중요한 역할을 한다. 기록관리 업무에서 메타데이터 요소들의 자동 부여가 불가능할 경우에는 기록전문가가 메타데이터 값을 직접 입력해야 한다. 이러한 업무의 불편함을 개선하기 위해 본 연구에서는 신기술인 챗GPT를 활용하여 기록관리 메타데이터 요소의 추출 방안을 제시하고자 하였다. 챗GPT 기술을 활용하기 위해 파이썬 프로그램과 랭체인 라이브러리를 이용하여 PDF 문서를 제시하고 질문을 통해 기록물의 메타데이터를 추출해보았고, 챗GPT 온라인 서비스를 통해 여러 건의 PDF 문서를 첨부하여 기록물의 메타데이터 요소를 추출해보았다. 그 결과 챗GPT-3.5 turbo를 사용한 랭체인에서는 보안상으로는 안전한 추출 방법이긴 하나 메타데이터의 정확한 요소를 얻기에는 다소 한계가 있었고, 챗GPT-4 온라인 서비스에서는 보안상 중요 문서를 첨부할 수 없지만 비교적 정확한 결과를 추출하였다. 이를 통해 기록관리에서의 메타데이터 추출을 위한 챗GPT 기술 활용의 가능성을 타진할 수 있었고, 챗GPT 관련 기술의 발달에 따라 좀 더 안전하고 정확한 결과 추출이 가능해질 것이다. 이러한 챗GPT의 장점을 활용함으로써 기록관에서 기록 및 메타데이터의 관리적 측면에서 업무의 효율성 및 생산성을 증대시키는데 도움을 줄 수 있을 것이라 기대한다.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2000년도 ITC-CSCC -1
• /
• pp.92-96
• /
• 2000

In this paper, we propose a call admission control(CAC) scheme fer the mixed voice/data DS-CDMA systems and analyze the Er-lang capacity under the proposed CAC scheme. Voice and data traffics require different system resources based oil their Quality of Service(QoS) requirements. In the proposed CAC scheme, some system resources are reserved exclusively for handoff calls to have high priority Over new calls. Additionally the queueing of both new and handoff data traffics that are not sensitive to delay is allowed. Ar a performance measure for the suggested CAC scheme. Erlang capacity is utilized. For the performance analysis, a four-dimensional Markov chain model is developed. Erlang capacity of a practical IS-95B type system depicts, and optimum values of system parameters such as the number of reservation channels and queue lengths are found with respect to Erlang capacity. Finally, it is observed that Erlang capacity is improved more than two times by properly selecting the system parameters with the proposed CAC scheme.

• Journal of Microbiology and Biotechnology
• /
• 제31권6호
• /
• pp.784-793
• /
• 2021

Previous studies have uncovered the role of circ_0000144 in various tumors. Here, we investigated the function and mechanism of circ_0000144 in gastric cancer (GC) progression. The expression of circ_0000144 in GC tissues and cells was detected through quantitative real-time polymerase chain reaction (qRT-PCR) method. Gain- and loss-of-function experiments including colony formation, wound healing and transwell assays were performed to examine the role of circ_0000144 in GC cells. Furthermore, western blot was conducted to determine the expressions of epithelial mesenchymal transition (EMT)-related proteins. The interaction between circ_0000144 and miR-217 was analyzed by bioinformatic analysis and luciferase reporter assays. The circ_0000144 expression was obviously upregulated in GC tissues and cells. Silencing of circ_0000144 inhibited cell proliferation, migration and invasion of GC cells, but ectopic expression of circ_0000144 showed the opposite results. Moreover, circ_0000144 sponged miR-217, and rescue assays revealed that silencing miR-217 expression reversed the inhibitory effect of circ_0000144 knockdown on the progress of GC. Our findings reveal that circ_0000144 inhibition suppresses GC cell proliferation, migration and invasion via absorbing miR-217, providing a new biomarker and potential therapeutic target for treatment of GC.

• 유통과학연구
• /
• 제20권8호
• /
• pp.9-20
• /
• 2022

Purpose: Distribution competence plays a key role in supply chain management. A successful distribution system will bring high values and profit for company. Omni-channel is a modern type of distribution which is used by many companies to access customers more effectively. This research is about the influence of distribution competence on customer value creation, competitive advantage and business performance of retailers in omni-channel. Research design, data and methodology: Questionnaires were issued to 200 managers who have experiences in using an Omni-channel to provide consumer goods. PLS-SEM is used to check the relationship among these above factors in omni-channel in Vietnam context. Results: Distribution competence has positive direct effect on business performance. At the same time, it has positive indirect effect on business performance through the mediating role of customer value creation and competitive advantage. Moreover, the role of mediating variables make contribution to strengthen the relationship of distribution competence and business performance in omni-channel. Conclusions: Through distribution competence, retailers can get competitive advantage, create more values for customers and increase business performance in omni channel in Vietnam.

• Journal of Microbiology and Biotechnology
• /
• 제20권4호
• /
• pp.670-677
• /
• 2010

An alkaline pectate lyase, Bsp165PelA, was purified to homogeneity from the culture broth of alkaliphilic Bacillus sp. N16-5. The enzyme showed a specific activity as high as 1,000 U/mg and had optimum activity at pH 11.5 and $50^{\circ}C$. It was composed of a single polypeptide chain with a molecular mass of 42 kDa deduced from SDS-PAGE, and its isoelectric point was around pH 6.0. It could efficiently depolymerize polygalacturonate and pectin. Characterization of product formation revealed unsaturated digalacturonate and trigalacturonate as the main products. The pectate lyase gene (pelA) contained an open reading frame (ORF) of 1,089 bp, encoding a 36-amino acids signal peptide and a mature protein of 326 amino acids with a calculated molecular mass of 35.943 Da. The deduced amino acid sequence from the pelA ORF exhibited significant homology to those of known pectate lyases in polysaccharide lyase family 1. Some conserved active-site amino acids were found in the deduced amino acid sequence of Bsp165PelA. $Ca^{2+}$ was not required for activity on pectic substrates.

• Journal of Forest and Environmental Science
• /
• 제33권2호
• /
• pp.113-121
• /
• 2017

In flight particulate matter particularly emissions generated by incomplete combustion processes has become a subject of global concern due to the health problems and environmental impacts associated with them. This has compelled most countries to set standards for coarse and fine particles due to their conspicuous impacts on environment and public health. This contribution therefore explores forest fire emissions and how its particulates affects air quality, damage to vegetation, water bodies and biological functions as architects for lung diseases and other degenerative illnesses such as oxidative stress and aging. Soot was collected from simulated forest fire using a clean glass surface and carefully transferred into amber vials for analysis. Volatile components of soot were collected over 10 mL dichloromethane and analyzed using a QTOF Premier-Water Corp Liquid Chromatography hyphenated to a mass selective detector (MSD), and Gas Chromatograph coupled to a mass spectrometer (GC-MS). To characterize the size and surface morphology of soot, a scanning electron microscope (SEM) was used. The characterization of molecular volatiles from simulated forest fire emissions revealed long chain compounds including octadec-9-enoic acid, octadec-6-enoic acid, cyclotetracosane, cyclotetradecane, and a few aromatic hydrocarbons (benzene and naphthalene). Special classes of organics (dibenzo-p-dioxin and 2H-benzopyran) were also detected as minor products. Dibenzo-p-dioxin for instance in chlorinated form is one of the deadliest environmental organic toxins. The average particulate size of emissions using SEM was found to be $11.51{\pm}4.91{\mu}m$. This study has shown that most of the emissions from simulated forest fire fall within $PM_{10}$ particulate size. The molecular by-products of forest fire and particulate emissions may be toxic to both human and natural ecosystems, and are possible precursors for various respiratory ailments and cancers. The burning of a forest by natural disasters or man-made fires results in the destruction of natural habitats and serious air pollution.