• Journal of Korea Multimedia Society
• /
• v.13 no.12
• /
• pp.1786-1797
• /
• 2010

Automatic understanding of contents in document image is a very hard problem due to involvement with mathematically challenging problems originated mainly from the over-determined system induced by document segmentation process. In both academic and industrial areas, there have been incessant and various efforts to improve core parts of content retrieval technologies by the means of separating out segmentation related issues using semi-structured document, e.g., invoice,. In this paper we proposed classification models for text lines on invoice document in which text lines were clustered into the five categories in accordance with their contents: purchase order header, invoice header, summary header, surcharge header, purchase items. Our investigation was concentrated on the performance of machine learning based models in aspect of linear-discriminant-analysis (LDA) and non-LDA (logic based). In the group of LDA, na$\"{\i}$ve baysian, k-nearest neighbor, and SVM were used, in the group of non LDA, decision tree, random forest, and boost were used. We described the details of feature vector construction and the selection processes of the model and the parameter including training and validation. We also presented the experimental results of comparison on training/classification error levels for the models employed.

• The KIPS Transactions:PartB
• /
• v.12B no.1 s.97
• /
• pp.57-68
• /
• 2005

It is important to segment a text, which is independent upon any text-embedded auxiliary information. This paper presents a technique for dividing the text into field-coherent passages. The presented method is based upon extracting field-associated terms from the text measuring how the topics grow, shrink and shift from sentence to sentence. We propose measures of topic continuity and of topic transition and suggest how those could be used to find the boundaries among passages. After collecting 12,500 documents, we obtain $88{\%}$ for average precision and $78{\%}$ for recall in Korean training set.

• Journal of KIISE:Computing Practices and Letters
• /
• v.5 no.5
• /
• pp.611-618
• /
• 1999

An efficient and effective access to needed information becomes an important factor in the modern information society. Many people have developed information retrieval (IR) systems that retrieve needed information from a large amount of data at a given time. However, most freely available IR systems have been developed for English text rather than for Korean text. In this research, we have analyzed the IR system freeWAIS-sf, and localized it with the Korean morphological analyzer, namely HAM. The localized freeWAIS-sf can handle both English text and Korean text simultaneously. We have also modified the weighting scheme of freeWAIS-sf. The experimental result shows that the modified weighting scheme outperforms the original one in terms of retrieval effectiveness.

• Journal of Korean Institute of Industrial Engineers
• /
• v.31 no.1
• /
• pp.68-78
• /
• 2005

A new emotion-based image retrieval method is proposed in this paper. The research was motivated by Soen's evaluation of human emotion on color patterns. Thirteen pairs of adjective words expressing emotion pairs such as like-dislike, beautiful-ugly, natural-unnatural, dynamic-static, warm-cold, gay-sober, cheerful-dismal, unstablestable, light-dark, strong-weak, gaudy-plain, hard-soft, heavy-light are modeled by 19-dimensional color array and $4{\times}3$ gray matrix in off-line. Once the query is presented in text format, emotion model-based query formulation produces the associated color array and gray matrix. Then, images related to the query are retrieved from the database based on the multiplication of color array and gray matrix, each of which is extracted from query and database image. Experiments over 450 images showed an average retrieval rate of 0.61 for the use of color array alone and an average retrieval rate of 0.47 for the use of gray matrix alone.

• Proceedings of the Korea Contents Association Conference
• /
• 2003.05a
• /
• pp.393-398
• /
• 2003

Video which contains the multiple data such as text, images, audio and motion of objects is typical multimedia data. Multimedia retrieval system using XML is essential for efficient rep. of multimedia data. Therefore, multimedia retrieval system for retrieval and structural understanding is needed to retrieve the multimedia data. This Paper Proposes the multimedia retrieval system based on XML Meta-data modeling techniques.

• Journal of Information Management
• /
• v.38 no.1
• /
• pp.1-19
• /
• 2007

This study investigates the effect of data fusion on the retrieval effectiveness by performing an experiment combining multiple representations of Web documents. The types of document representation combined in the study include content terms, links, anchor text, and URL. The experimental results showed that the data fusion technique combining document representation methods in Web environment did not bring any significant improvement in retrieval effectiveness.

• Journal of Information Processing Systems
• /
• v.11 no.3
• /
• pp.450-464
• /
• 2015

In linguistics, stemming is the operation of reducing words to their more general form, which is called the 'stem'. Stemming is an important step in information retrieval systems, natural language processing, and text mining. Information retrieval systems are evaluated by metrics like precision and recall and the fundamental superiority of an information retrieval system over another one is measured by them. Stemmers decrease the indexed file, increase the speed of information retrieval systems, and improve the performance of these systems by boosting precision and recall. There are few Persian stemmers and most of them work based on morphological rules. In this paper we carefully study Persian stemmers, which are classified into three main classes: structural stemmers, lookup table stemmers, and statistical stemmers. We describe the algorithms of each class carefully and present the weaknesses and strengths of each Persian stemmer. We also propose some metrics to compare and evaluate each stemmer by them.

• Journal of Computing Science and Engineering
• /
• v.5 no.2
• /
• pp.150-160
• /
• 2011

Automatic text classification has a long history and many studies have been conducted in this field. In particular, many machine learning algorithms and information retrieval techniques have been applied to text classification tasks. Even though much technical progress has been made in text classification, there is still room for improvement in text classification. In this paper, we will discuss remaining issues in improving text classification. In this paper, three improvement issues are presented including automatic training data generation, noisy data treatment and term weighting and indexing, and four actual studies and their empirical results for those issues are introduced. First, the semi-supervised learning technique is applied to text classification to efficiently create training data. For effective noisy data treatment, a noisy data reduction method and a robust text classifier from noisy data are developed as a solution. Finally, the term weighting and indexing technique is revised by reflecting the importance of sentences into term weight calculation using summarization techniques.

• Proceedings of the Korean Information Science Society Conference
• /
• 2004.10a
• /
• pp.799-801
• /
• 2004

인터넷 정보의 축적량이 증가함으로 인해 사용자는 원하는 정보를 찾기가 더욱 어려워졌다 따라서 수많은 문서들 중에서 원하는 정보를 효과적으로 찾아주는 정보검색 시스템의 중요성이 증가하게 되었으며 이에 대한 연구도 활발히 진행되었다. 인터넷 문서에서 추출할 수 있는 정보들은 링크 정보, Anchor Text 정보, Title Text 정보, 본문 Text 정보 등이 있으며, 이런 정보들을 이용한 수많은 정보검색 시스템이 개발되거나 모델이 연구되고 있다 본 논문에서는 기존에 이용되어 왔던 일반적인 추출 점보들을 정제 및 처리를 통해 성능을 높일 수 있는 방안을 연구했던 선행 연구를 기반으로 한 실험 결과 및 사이트 가중치를 추가한 모델을 제시한다.

• Journal of the Korean Society for Library and Information Science
• /
• v.32 no.1
• /
• pp.129-154
• /
• 1998

In these days the online full-text database are increasing, but conventional full-text information retrieval system has been proved with high recall ratio and low precision ratio. One of the disadvantages of full-text IR system is that it is not designed to reflect the user's information need it is due to the fact that full-text IR system has been designed based on physical and logical structure of document without considering the content of document. Therefore, the purpose of the study examined feasibility of document content structure in full-text IR system by resolving such disadvantages of conventional system. 180 Journal articles have been analyzed to find common structure of document content and finally general model of the structure of journal articles were developed. The result shows that have relation to between user's cogntive schema structure, user's information need and contents structure of document. Thus it is concluded that full-text IR system need to be designed by using document content structure in order to meet user's information need more effectively.