• 지능정보연구
• /
• 제24권2호
• /
• pp.21-35
• /
• 2018

본 논문에서는 스테레오 음악에 오디오 워터마크를 삽입하기 위한 알고리즘을 제안하였다. 스테레오 음악은 2개의 채널을 갖고 있기 때문에 기존 워터마킹 기술은 일반적으로 각 채널을 독립적으로 생각하고 처리하는 경우가 많다. 그러나 스테레오를 모노로 변환하는 과정에서 워터마크의 손실이 발생하는 경우가 많이 발생할 수 있다. 제안한 알고리즘은 스테레오를 모노로 변환하더라도 워터마크의 손실이 발생하지 않도록 워터마크를 삽입할 때 스테레오와 모노변환의 특성을 이용하였다. 제안된 알고리즘에 사용된 오디오 워터마크는 "Copyright"와 "Copy_free"라는 두 가지 정보를 터보코드를 이용하여 생성하였다. 두 워터마크는 9바이트(72비트)로 이루어져 있으며, 오류정정을 위하여 터보코드를 적용하면 222비트로 삽입해야 하는 정보량이 늘어난다. 222비트의 워터마크는 추가적인 오류에 강인하도록 1024비트로 확장하여 최종적으로 스테레오 음악에 삽입할 워터마크로 사용하였다. 평균적으로 SNR은 40dB를 넘어서서 전통적인 양자화 방식보다 10dB 이상의 음질 개선을 가져왔다. 이는 상대적으로 10배의 음질 개선도를 의미하는 것으로 매우 유의미한 결과이다. 또한 워터마크의 추출에 필요한 샘플길이는 1초 이내의 길이면 충분히 추출이 가능하고, 128Kbps의 비트레이트를 갖는 MP3 압축에 대해서도 모두 1초 이내 길이의 음악 샘플로부터 워터마크의 완전한 추출이 가능하였다. 전통적인 양자화 방식이 10초 길이의 샘플을 이용해도 대부분 워터마크의 추출에 실패한 것에 비하면 1/10에 불과한 길이로 워터마크의 추출이 가능하다.

• 환경정책연구
• /
• 제6권1호
• /
• pp.41-60
• /
• 2007

생물다양성협약의 핵심의제인 '유전자원에 대한 접근 및 이로부터 발생하는 이익의 공유(ABS)'에 관한 주요 논의 요소를 도출하고, 우리나라에 미치는 영향과 이를 대응하기 위한 정책방향을 제안하고자 하였다. ABS에 관한 주요 논의 내용은 유전자원의 접근과 유전자원의 이용으로부터 도출되는 이익의 공정하고 균등한 배분에 대한 Bonn 지침'의 법적인 구속력을 발휘할 수 있는 국제레짐(International Regime)의 마련, 유전자원의 기원 및 출처 등에 대한 국제인증제도, ABS의 사전통보승인(PIC)과 상호합의조건(MAT)의 의무준수 등이다. 풍부한 유전자원의 보유국이대 개도국들은 유전자원에 대한 이익공유의 최대화, 유전자원에 대한 전통지식의 권리인정, 자국의 유전자원 보전과 이용을 위한 선진외국으로부터의 기술이전 및 재정지원 요구 등의 주장을 통하여 생물다양성협약을 자국의 이익을 추구할 수 있는 주요한 수단으로 이용하고 있다. 반면 선진국들은 자국의 이익을 최대화하기 위하여 외국의 유전자원에 용이하게 접근하고 ABS에 유전자원의 이용기술의 권리인 지적재산권을 반영하도록 노력하는 등 개도국의 주장과는 대립되어 있다. Bonn지침과 ABS의 의무준수 등에 관한 동 협약의 결정내용 등은 우리나라에 긍정적이고 또한 부정적인 영향을 나타내고 있다. 특히 유전자원이 풍부하지 않은 우리나라의 경우, 국민의 의식주에 필요로 하는 유전자원의 수입이 더욱 어려워질 것이고, 유전자원을 훼손하는 국내 각종 개발 사업은 더욱 큰 제한을 받게 될 것이다. 따라서 해외의 유전자원을 확보하고, 유전자원의 보전 및 이용에 관한 고유기술 개발 등에 관한 경제적 부담이 증가 할 것이다. 이러한 부정적인 영향을 최소화하기 위해서는 ABS에 관한 우리의 입장을 구체적으로 정립하고, 국제사회의 흐름을 우리에게 유리하게 이끌어 나가야 한다. 이를 위해서는 i) 국제적인 ABS 논의동향과 향후 논의의 결정사항이 우리나라에 미치는 영향에 대한 과학적이고 체계적인 연구의 추진, ii) ABS에 관련된 영향을 대응할 수 있는 국가의 이행계획을 수립하고 이행, iii) 상기 이행계획의 지속적인 모니터링을 통해 그 효율성을 검정하고 이를 통해 이행계획을 개선, 그리고 iv) ABS에 관한 여타의 국제적인 논의에도 적극적으로 임해야 할 것이다.

• Asia pacific journal of information systems
• /
• 제18권2호
• /
• pp.41-72
• /
• 2008

The purpose of this study is the investigation of the effects of user participation, network effect, social influence, and usefulness on stickiness and continued use on Internet communities. In this research, stickiness refers to repeat visit and visit duration to an Internet community. Continued use means the willingness to continue to use an Internet community in the future. Internet community-based companies can earn money through selling the digital contents such as game, music, and avatar, advertizing on internet site, or offering an affiliate marketing. For such money making, stickiness and continued use of Internet users is much more important than the number of Internet users. We tried to answer following three questions. Fist, what is the effects of user participation on stickiness and continued use on Internet communities? Second, by what is user participation formed? Third, are network effect, social influence, and usefulness that was significant at prior research about technology acceptance model(TAM) still significant on internet communities? In this study, user participation, network effect, social influence, and usefulness are independent variables, stickiness is mediating variable, and continued use is dependent variable. Among independent variables, we are focused on user participation. User participation means that Internet user participates in the development of Internet community site (called mini-hompy or blog in Korea). User participation was studied from 1970 to 1997 at the research area of information system. But since 1997 when Internet started to spread to the public, user participation has hardly been studied. Given the importance of user participation at the success of Internet-based companies, it is very meaningful to study the research topic of user participation. To test the proposed model, we used a data set generated from the survey. The survey instrument was designed on the basis of a comprehensive literature review and interviews of experts, and was refined through several rounds of pretests, revisions, and pilot tests. The respondents of survey were the undergraduates and the graduate students who mainly used Internet communities. Data analysis was conducted using 217 respondents(response rate, 97.7 percent). We used structural equation modeling(SEM) implemented in partial least square(PLS). We chose PLS for two reason. First, our model has formative constructs. PLS uses components-based algorithm and can estimated formative constructs. Second, PLS is more appropriate when the research model is in an early stage of development. A review of the literature suggests that empirical tests of user participation is still sparse. The test of model was executed in the order of three research questions. First user participation had the direct effects on stickiness(${\beta}$=0.150, p<0.01) and continued use (${\beta}$=0.119, p<0.05). And user participation, as a partial mediation model, had a indirect effect on continued use mediated through stickiness (${\beta}$=0.007, p<0.05). Second, optional participation and prosuming participation significantly formed user participation. Optional participation, with a path magnitude as high as 0.986 (p<0.001), is a key determinant for the strength of user participation. Third, Network effect (${\beta}$=0.236, p<0.001). social influence (${\beta}$=0.135, p<0.05), and usefulness (${\beta}$=0.343, p<0.001) had directly significant impacts on stickiness. But network effect and social influence, as a full mediation model, had both indirectly significant impacts on continued use mediated through stickiness (${\beta}$=0.11, p<0.001, and ${\beta}$=0.063, p<0.05, respectively). Compared with this result, usefulness, as a partial mediation model, had a direct impact on continued use and a indirect impact on continued use mediated through stickiness. This study has three contributions. First this is the first empirical study showing that user participation is the significant driver of continued use. The researchers of information system have hardly studies user participation since late 1990s. And the researchers of marketing have studied a few lately. Second, this study enhanced the understanding of user participation. Up to recently, user participation has been studied from the bipolar viewpoint of participation v.s non-participation. Also, even the study on participation has been studied from the point of limited optional participation. But, this study proved the existence of prosuming participation to design and produce products or services, besides optional participation. And this study empirically proved that optional participation and prosuming participation were the key determinant for user participation. Third, our study compliments traditional studies of TAM. According prior literature about of TAM, the constructs of network effect, social influence, and usefulness had effects on the technology adoption. This study proved that these constructs still are significant on Internet communities.

• 한국산림과학회지
• /
• 제104권3호
• /
• pp.337-351
• /
• 2015

한국임학회지(1962년의 1권부터 2013년 102권 4호)에 게재된 산림특용자원식물 논문에서 대상식물을 식 음용식물, 약용식물, 사료식물, 조경식물, 섬유식물, 산업용 식물, 밀원식물, 바이오에너지/환경정화 식물, 염료/도료 및 희귀 멸종위기 특산식물로 분류하고, 그 연구내용에 따라 입지환경, 생태, 생리, 번식, 조림(무육과 재배 포함), 유전 육종, 분류(또는 식별), 병 충해와 방제, 동물피해와 방제, 성분분석과 추출물, 식생 또는 자생지 정보, 생물공학, 경영과 경제, 휴양 및 산림치유 및 총설로 구분하였다. 총 2,433편의 논문 중에서 특용자원식물 관련 논문은 611편으로 25.1%를 나타내었다. 연구내용에서 보면, 조림분야의 논문이 14.9%로 가장 높은 비율을 나타냈으며, 다음으로는 생리, 번식, 분류 및 유전 육종 등의 순으로 나타났다. 식물의 이용 목적에 따라서는 식 음용으로 이용된 식물이 26.5%로 가장 높게 나타났으며, 다음으로는 산업용, 바이오에너지/환경정화, 조경, 약용과 희귀 멸종위기 특산식물 등의 순으로 나타났다. 게재된 식물 중 연구빈도가 가장 높은 식물은 포플러류(62편)였으며, 밤나무, 잣나무, 아까시나무, 은행나무 등의 순으로 나타났다. 연구의 분야, 내용 및 대상식물 결과 등을 종합하여 보면, 1) 산림 내 생물자원에 대한 자원적 가치 인식 제고, 2) 산림과학분야 다양한 영역에서의 생산, 경영 및 이용에 관한 연구, 3) 식물정보의 데이터베이스 구축 및 타 과학 및 산업 분야의 흐름에 부응 및 4) 전통 식물 발굴 등이 산림특용자원식물 분야의 연구에서 깊이 고려되어야 할 것이다.

• Asia pacific journal of information systems
• /
• 제19권2호
• /
• pp.157-178
• /
• 2009

Corporate credit rating is a very important factor in the market for corporate debt. Information concerning corporate operations is often disseminated to market participants through the changes in credit ratings that are published by professional rating agencies, such as Standard and Poor's (S&P) and Moody's Investor Service. Since these agencies generally require a large fee for the service, and the periodically provided ratings sometimes do not reflect the default risk of the company at the time, it may be advantageous for bond-market participants to be able to classify credit ratings before the agencies actually publish them. As a result, it is very important for companies (especially, financial companies) to develop a proper model of credit rating. From a technical perspective, the credit rating constitutes a typical, multiclass, classification problem because rating agencies generally have ten or more categories of ratings. For example, S&P's ratings range from AAA for the highest-quality bonds to D for the lowest-quality bonds. The professional rating agencies emphasize the importance of analysts' subjective judgments in the determination of credit ratings. However, in practice, a mathematical model that uses the financial variables of companies plays an important role in determining credit ratings, since it is convenient to apply and cost efficient. These financial variables include the ratios that represent a company's leverage status, liquidity status, and profitability status. Several statistical and artificial intelligence (AI) techniques have been applied as tools for predicting credit ratings. Among them, artificial neural networks are most prevalent in the area of finance because of their broad applicability to many business problems and their preeminent ability to adapt. However, artificial neural networks also have many defects, including the difficulty in determining the values of the control parameters and the number of processing elements in the layer as well as the risk of over-fitting. Of late, because of their robustness and high accuracy, support vector machines (SVMs) have become popular as a solution for problems with generating accurate prediction. An SVM's solution may be globally optimal because SVMs seek to minimize structural risk. On the other hand, artificial neural network models may tend to find locally optimal solutions because they seek to minimize empirical risk. In addition, no parameters need to be tuned in SVMs, barring the upper bound for non-separable cases in linear SVMs. Since SVMs were originally devised for binary classification, however they are not intrinsically geared for multiclass classifications as in credit ratings. Thus, researchers have tried to extend the original SVM to multiclass classification. Hitherto, a variety of techniques to extend standard SVMs to multiclass SVMs (MSVMs) has been proposed in the literature Only a few types of MSVM are, however, tested using prior studies that apply MSVMs to credit ratings studies. In this study, we examined six different techniques of MSVMs: (1) One-Against-One, (2) One-Against-AIL (3) DAGSVM, (4) ECOC, (5) Method of Weston and Watkins, and (6) Method of Crammer and Singer. In addition, we examined the prediction accuracy of some modified version of conventional MSVM techniques. To find the most appropriate technique of MSVMs for corporate bond rating, we applied all the techniques of MSVMs to a real-world case of credit rating in Korea. The best application is in corporate bond rating, which is the most frequently studied area of credit rating for specific debt issues or other financial obligations. For our study the research data were collected from National Information and Credit Evaluation, Inc., a major bond-rating company in Korea. The data set is comprised of the bond-ratings for the year 2002 and various financial variables for 1,295 companies from the manufacturing industry in Korea. We compared the results of these techniques with one another, and with those of traditional methods for credit ratings, such as multiple discriminant analysis (MDA), multinomial logistic regression (MLOGIT), and artificial neural networks (ANNs). As a result, we found that DAGSVM with an ordered list was the best approach for the prediction of bond rating. In addition, we found that the modified version of ECOC approach can yield higher prediction accuracy for the cases showing clear patterns.

• Asia pacific journal of information systems
• /
• 제17권2호
• /
• pp.123-137
• /
• 2007

The advent of the Internet and related Web technologies has created an easily accessible link between a firm and its customers, and has provided opportunities to a firm to use information technology to support supplementary after-sale services associated with a product or service. It has been widely recognized that supplementary services are an important source of customer value and of competitive advantage as the characteristics of the product itself. Many of these supplementary services are information-based and need not be co-located with the product, so more and more companies are delivering these services electronically. Net-based customer service, which is defined as an Internet-based computerized information system that delivers services to a customer, therefore, is the core infrastructure for supplementary service provision. The importance of net-based customer service in delivering supplementary after-sale services associated with product has been well documented. The strategic advantages of well-implemented net-based customer service are enhanced customer loyalty and higher lock-in of customers, and a resulting reduction in competition and the consequent increase in profits. However, not all customers utilize such net-based customer service. The digital divide is the phenomenon in our society that captures the observation that not all customers have equal access to computers. Socioeconomic factors such as race, gender, and education level are strongly related to Internet accessibility and ability to use. This is due to the differences in the ability to bear the cost of a computer, and the differences in self-efficacy in the use of a technology, among other reasons. This concept, applied to e-commerce, has been called the "e-commerce divide." High Internet penetration is not eradicating the digital divide and e-commerce divide as one would hope. Besides, to accommodate personalized support, a customer must often provide personal information to the firm. This personal information includes not only name and address, but also preferences information and perhaps valuation information. However, many recent studies show that consumers may not be willing to share information about themselves due to concerns about privacy online. Due to the e-commerce divide, and due to privacy and security concerns of the customer for sharing personal information with firms, limited numbers of customers adopt net-based customer service. The limited level of customer adoption of net-based customer service affects the firm profits and the customers' welfare. We use a game-theoretic model in which we model the net-based customer service system as a mechanism to enhance customers' loyalty. We model a market entry scenario where a firm (the incumbent) uses the net-based customer service system in inducing loyalty in its customer base. The firm sells one product through the traditional retailing channels and at a price set for these channels. Another firm (the entrant) enters the market, and having observed the price of the incumbent firm (and after deducing the loyalty levels in the customer base), chooses its price. The profits of the firms and the surplus of the two customers segments (the segment that utilizes net-based customer service and the segment that does not) are analyzed in the Stackelberg leader-follower model of competition between the firms. We find that an increase in adoption of net-based customer service by the customer base is not always desirable for firms. With low effectiveness in enhancing customer loyalty, firms prefer a high level of customer adoption of net-based customer service, because an increase in adoption rate decreases competition and increases profits. A firm in an industry where net-based customer service is highly effective loyalty mechanism, on the other hand, prefers a low level of adoption by customers.

• 한국세라믹학회지
• /
• 제41권11호
• /
• pp.834-841
• /
• 2004

우리 주변(경기도 이천, 양평, 용인)에서 쉽게 구할 수 있는 곡물대 10가지와 2가지의 곡물껍질(콩깍지, 왕겨), 4가지의 채소(시금치, 무우잎과 줄기, 호박잎과 줄기, 고추대)의 재를 만들어 한국적 천연 재유의 특성을 연구 분석하고 그 활용 방안을 제시하였다. 그 결과 곡물대재는 4가지로 분류되었다. 1그룹인 들깨대재, 참깨대재, 검정콩대재, 팥대재에는 다른재에 비하여 $SiO_2$ 성분이 적고 CaO와 P$_2$ $O_{5}$ 성분이 10%이상 함유되어 비교적 유백현상이 강하게 나타났으며 약한 녹색기가 도는 밝은 노랑색유로 나타나 유백유 기본유로 활용하기에 적합하였다. 2그룹인 고추대재, 시금치재, 콩깍지재, 무우잎ㆍ줄기재, 콩대재에는 모든 성분이 비교적 고르게 함유되어 있었고, 특히 MgO 함량이 평균 8.53%로 가장 많고 Fe$_2$ $O_3$ 성분이 2% 정도 함유하고 있어 채도가 높은 녹색기를 많이 띤 노랑색유로 나타났으며 유흐름 현상이 강하게 나타나 이라 보 유약의 기본유로 활용하기에 적합하였다. 3그룹인 옥수수대재, 흰콩대재, 호박잎ㆍ줄기재는 $SiO_2$, $Al_2$ $O_3$의 함량이 다른 재에 비하여 많았고 Fe$_2$ $O_3$가 3～5%로 가장 많이 함유되어 있어 UV에 대한 4분류 중 색상, 채도 변화가 가장 크게 나타났으며 짙은 노랑에서 짙은 고동색을 지닌 색상이 강하게 나타나 천목유 및 흑유의 기본유로 활용하기에 적합하였다. $SiO_2$ 성분이 45～82%로 함량이 제일 많아 잘 녹지 않고 다른 재에 비하여 Fe$_2$ $O_3$와 P$_2$ $O_3$ 성분의 함량이 적은 4그룹인 갈대재, 볏짚재, 수수대재, 왕겨재는 푸른색기 또는 녹색기가 도는 흰색에 가까운 미색으로 나타나 백유의 기본유로 활용하기에 적합하였다.

• 지능정보연구
• /
• 제16권4호
• /
• pp.99-112
• /
• 2010

앙상블 학습은 분류 및 예측 알고리즘의 성과개선을 위하여 제안된 기계학습 기법이다. 그러나 앙상블 학습은 기저 분류자의 다양성이 부족한 경우 다중공선성 문제로 인하여 성과개선 효과가 미약하고 심지어는 성과가 악화될 수 있다는 문제점이 제기되었다. 본 연구에서는 기저 분류자의 다양성을 확보하고 앙상블 학습의 성과개선 효과를 제고하기 위하여 유전자 알고리즘 기반의 범위 최적화 기법을 제안하고자 한다. 본 연구에서 제안된 최적화 기법을 기업 부실예측 인공신경망 앙상블에 적용한 결과 기저 분류자의 다양성이 확보되고 인공신경망 앙상블의 성과가 유의적으로 개선되었음을 보여주었다.

• 한국막학회:학술대회논문집
• /
• 한국막학회 1999년도 The 7th Summer Workshop of the Membrane Society of Korea
• /
• pp.39-42
• /
• 1999

Perfluoropolymers represent the ultimate resistance to hostile chemical environments and high service temperature, attributed to the presence of fluorine in the polymer backbone, i.e. to the high bond energy of C-F and C-C bonds of fluorocarbons. Copolymers of Tetrafluoroethylene (TEE) and 2, 2, 4Trifluoro-5Trifluorometoxy- 1, 3Dioxole (TTD), commercially known as HYFLON AD, are amorphous perfluoropolymers with glass transition temperature (Tg)higher than room temperature, showing a thermal decomposition temperature exceeding 40$0^{\circ}C$. These polymer systems are highly soluble in fluorinated solvents, with low solution viscosities. This property allows the preparation of self-supported and composite membranes with desired membrane thickness. Symmetric and asymmetric perfluoropolymer membranes, made with HYFLON AD, have been prepared and evaluated. Porous and not porous symmetric membranes have been obtained by solvent evaporation with various processing conditions. Asymmetric membranes have been prepared by th wet phase inversion method. Measure of contact angle to distilled water have been carried out. Figure 1 compares experimental results with those of other commercial membranes. Contact angles of about 120$^{\circ}$for our amorphous perfluoropolymer membranes demonstrate that they posses a high hydrophobic character. Measure of contact angles to hexandecane have been also carried out to evaluate the organophobic character. Rsults are reported in Figure 2. The observed strong organophobicity leads to excellent fouling resistance and inertness. Porous membranes with pore size between 30 and 80 nanometers have shown no permeation to water at pressures as high as 10 bars. However high permeation to gases, such as O2, N2 and CO2, and no selectivities were observed. Considering the porous structure of the membrane, this behavior was expected. In consideration of the above properties, possible useful uses in th field of gas- liquid separations are envisaged for these membranes. A particularly promising application is in the field of membrane contactors, equipments in which membranes are used to improve mass transfer coefficients in respect to traditional extraction and absorption processes. Gas permeation properties have been evaluated for asymmetric membranes and composite symmetric ones. Experimental permselectivity values, obtained at different pressure differences, to various single gases are reported in Tab. 1, 2 and 3. Experimental data have been compared with literature data obtained with membranes made with different amorphous perfluoropolymer systems, such as copolymers of Perfluoro2, 2dimethyl dioxole (PDD) and Tetrafluorethylene, commercialized by the Du Pont Company with the trade name of Teflon AF. An interesting linear relationship between permeability and the glass transition temperature of the polymer constituting the membrane has been observed. Results are descussed in terms of polymer chain structure, which affects the presence of voids at molecular scale and their size distribution. Molecular Dyanmics studies are in progress in order to support the understanding of these results. A modified Theodoru- Suter method provided by the Amorphous Cell module of InsightII/Discover was used to determine the chain packing. A completely amorphous polymer box of about 3.5 nm was considered. Last but not least the use of amorphous perfluoropolymer membranes appears to be ideal when separation processes have to be performed in hostile environments, i.e. high temperatures and aggressive non-aqueous media, such as chemicals and solvents. In these cases Hyflon AD membranes can exploit the outstanding resistance of perfluoropolymers.

• 지능정보연구
• /
• 제21권1호
• /
• pp.1-13
• /
• 2015

빅데이터 및 오피니언 마이닝 분야가 대두됨에 따라 정보 검색/추출, 특히 비정형 데이터에서의 정보 검색/추출 기술의 중요성이 나날이 부각되어지고 있다. 또한 정보 검색 분야에서는 이용자의 의도에 맞는 결과를 제공할 수 있는 검색엔진의 성능향상을 위한 다양한 연구들이 진행되고 있다. 이러한 정보 검색/추출 분야에서 자연어처리 기술은 비정형 데이터 분석/처리 분야에서 중요한 기술이고, 자연어처리에 있어서 하나의 단어가 여러개의 모호한 의미를 가질 수 있는 단어 중의성 문제는 자연어처리의 성능을 향상시키기 위해 우선적으로 해결해야하는 문제점들의 하나이다. 본 연구는 단어 중의성 해소 방법에 사용될 수 있는 말뭉치를 많은 시간과 노력이 요구되는 수동적인 방법이 아닌, 사전들의 예제를 활용하여 자동적으로 생성할 수 있는 방법을 소개한다. 즉, 기존의 수동적인 방법으로 의미 태깅된 세종말뭉치에 표준국어대사전의 예제를 자동적으로 태깅하여 결합한 말뭉치를 사용한 단어 중의성 해소 방법을 소개한다. 표준국어대사전에서 단어 중의성 해소의 주요 대상인 전체 명사 (265,655개) 중에 중의성 해소의 대상이 되는 중의어 (29,868개)의 각 센스 (93,522개)와 연관된 속담, 용례 문장 (56,914개)들을 결합 말뭉치에 추가하였다. 품사 및 센스가 같이 태깅된 세종말뭉치의 약 79만개의 문장과 표준국어대사전의 약 5.7만개의 문장을 각각 또는 병합하여 교차검증을 사용하여 실험을 진행하였다. 실험 결과는 결합 말뭉치를 사용하였을 때 정확도와 재현율에 있어서 향상된 결과가 발견되었다. 본 연구의 결과는 인터넷 검색엔진 등의 검색결과의 성능향상과 오피니언 마이닝, 텍스트 마이닝과 관련한 자연어 분석/처리에 있어서 문장의 내용을 보다 명확히 파악하는데 도움을 줄 수 있을 것으로 기대되어진다.