• 전자공학회논문지CI
• /
• 제41권2호
• /
• pp.43-50
• /
• 2004

웹과 같은 대량의 문서집단에서 단일 문서에 대한 자동 요약은 일반적으로 통계요약 방법을 이용한다. 그러나 단순한 통계 요약 방법은 문서내의 빈도수가 높은 단어를 포함하는 문장들이 중복되어 나타날 확률이 높다. 이러한 단점을 보완하기 위하여 본 논문에서는 통계기반 요약방법에 MMR 기법을 적용하여 요약의 질을 향상시켰다(약 λ=0.6에서 최고의 성능을 보임). 또한 본 논문에서는 인위적 요약을 수행하여 MMR 통계기반의 요약 결과의 성능을 평가하였다.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2003년도 하계종합학술대회 논문집 Ⅲ
• /
• pp.1319-1322
• /
• 2003

We present a web document summarizer, simpler more condense than the existing ones, of a search engine. This summarizer generates summaries with a statistic-based summarization method using Clustering or MMR technique to reduce redundancy in the results, and that generates summaries using Perfect Link Graph. We compare the results with the summaries generated by human subjects. For the comparison, we use FScore. Our experimental results verify the accuracy of the summarization methods.

• 한국산업정보학회논문지
• /
• 제10권2호
• /
• pp.26-31
• /
• 2005

본 논문에서는 웹 검색엔진에서 일반적으로 사용하는 문서요약에 대한 연구로써 문서 내에 있는 문장들의 꼭짓점을 연결하는 완전그래프기법을 도입하여 요약내용을 좀 더 간결하고 함축하게 하는 통계요약기법을 제안했다. 이 요약기술을 지금까지 통계 문서요약기술에서 우수하다고 판단된 클러스터링 기법과 MMR 기법 등과 비교하였다. 특히, 요약 성능을 평가하기 위하여 인위적으로 요약된 요약문을 기준으로 한 각 요약기법들의 FScore값들과 비교하였다. 이 기술들 중에서 완전그래프기법이 약 $30\%$정도 성능향상을 보였다.

• 한국수학교육학회지시리즈A:수학교육
• /
• 제44권3호
• /
• pp.457-475
• /
• 2005

We have inquired on what the statistical classes of the secondary schools had been aiming to, say the epistermlogical objects. And we now appreciate that the main obstacle to the systematic articulation is the lack of anticipation on what the statistical concepts are. This study focuses on the ingredients of the statistical concepts. Those are to be the ground of the systematic articulation of statistic courses, especially of the one for the school kids. Thus we required that those ingredients must satisfy the followings. i) directly related to the contents of statistics ii) psychologically developing iii) mutually exclusive each other as much as possible iv) exhaustive enough to cover all statistical concepts We examined what and how statisticians had been doing and the various previous views on these. After all we suggest the following three concepts are the core of conceptual developments of statistic, say the concept of distributions, the summarizing ability and the concept of samples. By the concepts of distributions we mean the frequency views on each random categories and that is developing from the count through the probability along ages. Summarizing ability is another important resources to embed his probe with the data set. It is not only viewed as a number but also to be anticipated as one reflecting a random phenomena. Inductive generalization is one of the most hazardous thing. Statistical induction is a scientific way of challenging this and this starts from distinguishing the chance with the inevitable consequences. One's inductive logic grows up along with one's deductive arguments, nevertheless they are different. The concept of samples reflects' one's view on the sample data and the way of compounding one's logic with the data within one's hypothesis. With these three in mind we observed Korean Statistic Curriculum from K to 12. Distributional concepts are dealt with throughout but not sequenced well. The way of summarization has been introduced in the 1 st, 5th, 7th and the 10th grade as a numerical value only. One activity on the concept of sample is given at the 6th grade. And it jumps into the statistical reasoning at the selective courses of ' Mathematics I ' or of ' Probability and Statistics ' in the grades of 11-12. We want to suggest further studies on the developing stages of these three conceptual features so as to obtain a firm basis of successive statistical articulation.