Earticle

웹 마이닝은 사용자의 웹 이용 분석을 위해 웹에서 발생한 데이터를 대상으로 유용한 패턴을 찾 아내는 기법이다. 하지만 기존의 웹 마이닝은 웹 로그의 연속적인 특성을 충분히 반영하여 분석하지 못하였다. 이를 보완하기 위하여 본 연구에서는 프로세스 마이닝을 활용하여 프로세스 모델에 의한 순 차적인 관계에 따른 웹 접속 로그를 분석하는 프레임워크를 제시한다. 프로세스 모델에 기반한 웹 로 그 분석은 웹 페이지들을 이동한 사용자들의 행위를 이해하고 문제점과 개선방안을 도출하는 데 유용 하게 사용될 수 있다. 본 연구에서는 제안한 방법론을 이용하여 대학정보시스템의 웹 로그를 분석하 여 적용 가능성과 그 분석 결과를 제시하였다.

Web mining techniques are often used to discover useful patterns from data log generated by Web servers for the purpose of web usage analysis. Yet traditional Web mining techniques do not reflect sufficiently sequential properties of Web log data. To address such weakness, we introduce a framework for analyzing Web access log data by using process mining techniques. To illustrate the proposed framework, we show the analysis of Web access log in a campus information system based on the framework and discuss the implication of the analysis result.

포털 사이트의 핵심 서비스인 검색서비스는 입력되는 검색어 중에서 짧은 순간에 급상승하는 검색어를 대상으로 순간 검색빈도가 높은 것을 기준으로 순위별로 제시하는 것이므로 일정기간 동안 관심도가 높은 검색어를 곧바로 알려주기 는 힘들다. 따라서 이를 극복하고 검색어 변화에 대한 향상된 분석결과가 나오게 하여 보다 지능적인 정보검색 서비스를 제공하기 위한 노력이 필요하다. 이를 위하여 본 논문에서는 실시간검색어의 관심도와 지속도, 그리고 주목도를 측정할 수 있는 기준을 제시한다. 그리고 그 기준에 맞추어 일정기간 동안 시간, 일간, 주간, 월간 실시간검색어에 대한 변화의 측정과 집계를 하고 이를 통해 관심도가 높은 이슈, 관심이 길게 지속된 이슈, 변화가능성이 커서 앞으로 주목해야 할 이슈를 평가 한다.

The search service, which is a core service of the portal site, presents search queries that are rapidly increasing among the inputted search queries based on the highest instantaneous search frequency, so it is difficult to immediately notify a search query having a high degree of interest for a certain period. Therefore, it is necessary to overcome the above problems and to provide more intelligent information retrieval service by bringing improved analysis results on the change of the search queries. In this paper, we present the criteria for measuring the interest, continuity, and attention of real-time search queries. In addition, according to the criteria, we measure and summarize changes in real-time search queries in hours, days, weeks, and months over a period of time to assess the issues that are of high interest, long-lasting issues of interest, and issues that need attention in the future.

실시간검색어는 지금 바로 이슈가 되는 검색어의 검색 증가율이 단기간에 급상승하는 것을 중심으로 하기 때문에 일정기간 지속적으로 관심도를 유지하고 있는 이슈를 나타내지 못하고 이들이 가까운 미래에 어떤 변화를 보이는지에 대한 것도 알 수 없는 한계를 가지고 있다. 본 논문에서는 이러한 한계를 극복할 수 있도록 일정기간 동안 상위 10위 안에 속한 적이 있는 실시간검색어에 대해 일자별, 시간별 지속성을 평가하여 꾸준히 관심을 받는 검색어를 추출한다. 그런 다음, 이들 중 상위에 속하는 검색어의 관심도가 어떻게 변화하는지를 알 수 있게 하는 시계열 분석과 신경망을 이용하는 방법을 제시하고 이를 통해 도출한 실제 예를 통해 가까운 미래의 변화량을 예측한 결과를 보인다. 일자별로는 시계열 분석을, 시간별로는 인공신경망의 학습을 통해 예측하는 것이 좋은 결과를 보인다는 것을 알 수 있다.

Since realtime search words are centered on the fact that the search growth rate of an issue is rapidly increasing in a short period of time, it is not possible to express an issue that maintains interest for a certain period of time. In order to overcome these limitations, this paper evaluates the daily and hourly persistence of the realtime words that belong to the top 10 for a certain period of time and extracts the search word that are constantly interested. Then, we present the method of using the time series analysis and the neural network to know how the interest of the upper search word changes, and show the result of forecasting the near future change through the actual example derived through the method. It can be seen that forecasting through time series analysis by date and artificial neural networks learning by time shows good results.

인터넷의 전자상거래에서 대규모 정보 저장소에 있는 원하는 정보를 신속하게 검색하기란 어렵다. 대부분의 전자상거래 사이트에 있어서 정보를 제공하는 방법으로는 통계적 분석이나 분류별 지향의 간단한 과정을 통해 생성된다. 그러나 이러한 것은 생성 정보들 사이의 다양한 상호관계를 표현할 수 없고 사용자의 정확한 구매 패턴을 반영하기 어렵다. 본 논문에서는 전자상거래에서 효과적인 클러스터링을 이용한 다양한 관심정보 추출을 위한 효율적인 웹 마이닝을 제안한다. 이러한 방법은 분류별 항목에서 순차 패턴과 상관 규칙을 이용하여 생성 정보들 사이의 보다 적합한 상관관계를 구성하고 제안된 방법을 통해 효율적인 실험 결과를 나타낸다. 그리고 효과적인 클러스터링을 이용하여 신속한 검색을 제안한다.

In internet being a repository of massive information, we easily may not find our desired information, this issue also exists in e-commerce which gets rapid growth. In most of e-commerce sites, the methods furnishing information have been made use of statistical analysis or simple process by category-oriented, but these can't represent diverse correlation among products information and also hardly reflect users' purchasing patterns precisely. In this thesis, we propose more efficient web mining ways for discovering interesting information using effective clustering in e-commerce, which get achieved more suitable relationship among products information using both sequential patterns and association rules in category-independent, and experiments show the efficiency of our proposed methods. And we propose search using effective clustering rapidly.

우리나라는 기후변화 위기에 대응하기 위해 2050 탄소중립을 선언하였으며, 이를 위해 다양한 감축 계획 및 입법화 과정을 진행 중이다. 탄소중립의 실현은 산업기술 전반에서의 근본적 변화를 필요로 하기 때문에 이를 위한 구체적 대응체계 마련이 매우 중요하다. 본고는 탄소중립 관련 산업기술 확보 경쟁에서 선제적으로 대비하기 위하여 글로벌 탄소중립 기술분 야의 현황과 발전 트렌드를 파악하고자 한다. 이를 위해, 탄소중립 관련 온라인 뉴스기사 데이터를 웹 크롤링하여 수집하였 고, 미래신호분석방법론과 인공신경망 딥러닝 기술인 Word2Vec알고리즘을 적용하여 탄소중립 기술 트렌드를 분석 및 예측 하였다. 분석결과, 탄소 과배출 업종인 철강업 및 석유화학 분야의 기술고도화가 요구되고 있었으며, 전기차 분야에의 투자 타당성 확보와 기술 고급화가 추세인 것으로 드러났다. 이에 대한 정부의 적극적인 지원과 글로벌한 기술협력/인프라 조성이 밑받침되어야 할 것으로 보인다. 그 외에도 탄소중립 관련 인력양성이 시급한 것으로 나타났으며, 기업에서 필요한 탄소중립 인력을 양성할 수 있도록 간접지원정책 마련의 필요성을 확인할 수 있었다.

Carbon neutrality is the concept of reducing greenhouse gases emitted by human activities and making actual emissions zero through removal of remaining gases. It is also called “Net-Zero” and “carbon zero”. Korea has declared a “2050 Carbon Neutrality policy” to cope with the climate change crisis. Various carbon reduction legislative processes are underway. Since carbon neutrality requires changes in industrial technology, it is important to prepare a system for carbon zero. This paper aims to understand the status and trends of global carbon neutrality technology. Therefore, ROK’s web platform “www.naver.com.” was selected as the data collection scope. Korean online articles related to carbon neutrality were collected. Carbon neutrality technology trends were analyzed by future signal methodology and Word2Vec algorithm which is a neural network deep learning technology. As a result, technology advancement in the steel and petrochemical sectors, which are carbon over-release industries, was required. Investment feasibility in the electric vehicle sector and technology advancement were on the rise. It seems that the government's support for carbon neutrality and the creation of global technology infrastructure should be supported. In addition, it is urgent to cultivate human resources, and possible to confirm the need to prepare support policies for carbon neutrality.

The purpose of operating the main homepage of University is to provide the related information about University resources to site visitors. In this study, we analyze website browsing patterns and extract characteristics of users in order to improve its utilization. The access log files to main homepage were used to analyze the browsing patterns and converted to process log files adaptable to a process mining tool, ProM. Finally we provide useful information about user friendly homepage design and suggest plans for improving its utilization to website operators.

The distinguished prosperity of information technologies from great progress of e-business during the last decade has unavoidably made software development for active data mining to discovery hidden predictive information regarding business trends and behavior from vary large databases. Therefore this paper develops an active mining object(ADMO) component, which provides real-time predictive information from web databases. The ADMO component is to extended ADO(ActiveX Data Object) component to active data mining component based on COM(Component Object Model) for application program interface(API). ADMO component development made use of window script component(WSC) based on XML(eXtensible Markup Language). For the purpose of investigating the application environments and the practical schemes of the ADMO component, experiments for diverse practical applications were performed in this paper. As a result, ADMO component confirmed that it could effectively extract the analytic information of classification and aggregation from vary large databases for Web services.

Product recommender system is one of the most popular techniques for customer relationship management. In addition, collaborative filtering (CF) has been known to be one of the most successful recommendation techniques in product recommender systems. However, CF has some limitations such as sparsity and scalability problems. This study proposes hybrid cluster analysis and case-based reasoning (CBR) to address these problems. CBR may relieve the sparsity problem because it recommends products using customer profile and transaction data, but it may still give rise to scalability problem. Thus, this study uses cluster analysis to reduce search space prior to CBR for scalability problem. For cluster analysis, this study employs hybrid genetic and K-Means algorithms to avoid possibility of convergence in local minima of typical cluster analyses. This study also develops a Web-based prototype system to test the superiority of the proposed model.

웹소설과 라이트노벨은 완전히 다른 특징을 가진 소설군임에도 서적화를 통해 교집합이 생겨나는 독특한 관계를 구축하고 있다. 웹소설의 활발한 서적화는 일본 웹소설만이 갖는 차별점이다. 본고는 이러한 일본의 웹소설과 라이트노벨에 대해 웹소설 플랫폼 ‘소설가가 되자’의 작품을 중심으로 분석했다. 본 연구는 기존의 연구가 관념적 검토에 그쳤던 한계를 극복하고 방대한 작품 데이터를 텍스트마이 닝 분석을 통해 정량분석함으로써 두 소설의 특성을 종합적으로 규명하고자 했다. 그 결과, 웹소설의 서적화는 판타지와 연애 장르 외에 문예, SF장르에서 거의 진행되지 않음을 알 수 있었다. 실제로는 다양한 장르의 웹소설이 투고되지만 서 적화는 일부 장르만을 중심으로 진행된다. 또 웹소설의 서적화에는 투고 시기 및 작품의 분량 등의 요소가 영향을 미치는 것으로 나타났다. 라이트노벨 그룹과 그 외의 서적화 그룹을 비교함으로써 라이트노벨의 특징을 규명했다. 비현실 요소가 포함된 판타지 장르의 인기 작품이자 연재 형태로 많은 분량을 가진 작품이 라이트노벨로 출판되는 경향을 발견했다. 이는 라이트노벨의 주된 출판 형태가 시리즈 간행이라는 특성을 반영한다. 그리고 키워드와 줄거리 분석을 통해 라이트노벨이 주로 남성 독자를 상정하고 있음을 확인했다. 반면, 다 른 분야로의 서적화는 연애 요소를 포함한 인기 작품이 많이 진행되었다. 그러나 라이트노벨 만큼의 많은 분량은 아니더라도 연재 작품이라면 서적화 가능성이 높 은 것으로 나타났다. 또 키워드와 줄거리 분석을 통해 라이트노벨 외의 서적화는 주로 여성을 대상으로 한 작품이 중심을 이루고 있음을 알 수 있었다.

This study analyzed Japanese web novels and light novels, focusing on works from the web novel platform ‘Shousetsuka ni Narou’. This research aimed to overcome the limitations of previous studies that were mainly theoretical by using text mining to quantitatively analyze a vast amount of work data to comprehensively identify the characteristics of both types of novels. The results revealed that the publication of web novels rarely occurs in literary and science fiction genres outside of fantasy and romance genres. Although various genres of web novels are submitted, the actual publication focuses mainly on certain genres. Additionally, factors such as the submission period and the length of the work were found to influence the publication of web novels. By comparing the light novel group with other publication groups, the characteristics of light novels were identified. It was found that popular works in the fantasy genre with unrealistic elements and serialized with a large volume tend to be published as light novels. Furthermore, keyword and plot analysis confirmed that light novels are mainly targeted at male readers. On the other hand, publications in other fields mainly involved popular works with romance elements. However, it was found that serialized works are more likely to be published, even if they do not have as much volume as light novels. Keyword and plot analysis also showed that publications other than light novels mainly target women.

WUM is the part of web mining that identifies usages data from the web log server ,in order to known and improved serve the requirements of the web applications. The WUM involves three types, Data Preprocessing , Pattern Discovery & Pattern Analysis. pattern discovery phase contains many web data mining methods are apply to process the data so as to discover patterns. Once the pattern discovered. Analysis is done using various operation with unique session and unique user. This paper gives a brief overview of WUM and its phases .

There are numbers of web tools available to spread and explore information on internet. With the increasing use of Blogging sites, people are able to share their opinions, interests, experiences, and their views with others. However, it is not so easy to fetch required information from various Blogs in available time, which normally is very short. From last few years our major concern is on web searching and web mining. In this paper, a searching and curating model is discussed which introduces an approach of fetching the relevant information automatically from various Blog sites and fetches the personalized required information. That exactly is the need of the hour.

Identification of expert to domain knowledge in any field of interest is essential for consulting in industry, academia and scientific community. The objective of this study is to address the expert-finding task in contemporary communities. We proposed Multifaceted Web Mining Architecture (MfWMA) and implemented a tool with data extracted from Growbag, dblpXML and web authors home page resource to identify personnel with specific expertise. We mined two thousand and five hundred author's personal web pages with the underlying criteria of a dozen of key parameters; while parsing on each page in pursuit of 8 thousand topics. This study corroborate this quantification in terms of a measure of expertise. The prototype provides its users to distinguish the level of expertise in a particular area; thus resulting in the capability to mark people with broader expertise. Through this extension to the web enabling technique, we have demonstrated that the proposed architecture presents a novel web mining approximation with realistic results.

Named entity (NE) translation is a fundamental task in machine translation (MT) and cross-language information retrieval (CLIR). Furthermore, Organization name (ON) translation is the most complex among all the NEs. A novel system for translating ONs from Chinese to English, with a translation model and web resources, is proposed. Firstly, we built a translation model with Chunk. Then query expansion was adopted with the translation model and term-subject co-occurrence. Thirdly, we extracted the Chinese Organization names with English sentences using the method of frequency shifting and adjacency information to find English fragments as translation candidates. Finally, we found the best translation by computing the trustworthiness of all candidates. The experimental results showed that the approach returned a better performance than machine translation-based systems.

Along with the rapid advancement of Internet technology and machine learning science, the data mining techniques have been widely applied on the web page information pattern analysis issues. To enhance the traditional mining algorithms theoretically and numerically, we propose the novel deep web data mining algorithm based on multi-agent information system and collaborative correlation rule in this manuscript. Firstly, we review the latest web mining methodologies to serve as the comparison objects. Then, we introduce the revised agent based algorithm. MAS consists of more than one agent, MAS using parallel distributed processing technology and modular design thought and the complex system is divided into relatively independent agent subsystem. Later, we combine the AdaBoost method to propose the collaborative correlation rule. As the combination, we use the mentioned two techniques to form the optimized and enhanced deep web data mining algorithm with the implementation of programming languages. The experimental result proves the feasibility of our approach and compared with other contemporary state-of-the-art algorithms, our method outperforms and achieves better accuracy with low time-consuming.

The exponential increase in information, users and number of Websites on WWW has given rise to number of challenges. The most important challenge is the effective and systematic management of this massive Web data. For Web users, it is very difficult to access relevant information quickly and efficiently. And for Web site owners, it is very difficult to satisfy their users’ information needs effectively. Web Usage Mining has been used to deal with aforesaid issues. The Web Usage Mining techniques are solely based on knowledge acquired through the analysis of the users’ navigational behavior. Hence, quality of discovered patterns is low. Recent studies show that, semantically enriched Web Usage Mining enhances the quality of discovered patterns. The semantically enriched Web is called as Semantic Web, and this new form of Web Usage mining is called as Semantic Web Usage Mining. It is also called as Ontology based Web Usage Mining, as Ontologies act as backbone for conceptual description of semantic knowledge in Semantic Web. In this paper, we have presented brief overview of conventional Web usage mining and performed an extensive survey of research work done in ontology based web usage mining.