• Proceedings of the Korean Society for Language and Information Conference
• /
• 2002.02a
• /
• pp.1-1
• /
• 2002

This talk provides an overview of current work in my research group on the syntactic annotation of the T bingen corpus of spoken German and of the German Reference Corpus (Deutsches Referenzkorpus: DEREKO) of written texts. Morpho-syntactic and syntactic annotation as well as annotation of function-argument structure for these corpora is performed automatically by a hybrid architecture that combines robust symbolic parsing with finite-state methods ("chunk parsing" in the sense Abney) with memory-based parsing (in the sense of Daelemans). The resulting robust annotations can be used by theoretical linguists, who lire interested in large-scale, empirical data, and by computational linguists, who are in need of training material for a wide range of language technology applications. To aid retrieval of annotated trees from the treebank, a query tool VIQTORYA with a graphical user interface and a logic-based query language has been developed. VIQTORYA allows users to query the treebanks for linguistic structures at the word level, at the level of individual phrases, and at the clausal level.

• Genomics & Informatics
• /
• v.10 no.1
• /
• pp.33-39
• /
• 2012

High-throughput genomic technologies (HGTs), including next-generation DNA sequencing (NGS), microarray, and serial analysis of gene expression (SAGE), have become effective experimental tools for cancer genomics to identify cancer-associated somatic genomic alterations and genes. The main hurdle in cancer genomics is to identify the real causative mutations or genes out of many candidates from an HGT-based cancer genomic analysis. One useful approach is to refer to known cancer genes and associated information. The list of known cancer genes can be used to determine candidates of cancer driver mutations, while cancer gene-related information, including gene expression, protein-protein interaction, and pathways, can be useful for scoring novel candidates. Some cancer gene or mutation databases exist for this purpose, but few specialized tools exist for an automated analysis of a long gene list from an HGT-based cancer genomic analysis. This report presents a new web-accessible bioinformatic tool, called CaGe, a cancer genome annotation system for the assessment of candidates of cancer genes from HGT-based cancer genomics. The tool provides users with information on cancer-related genes, mutations, pathways, and associated annotations through annotation and browsing functions. With this tool, researchers can classify their candidate genes from cancer genome studies into either previously reported or novel categories of cancer genes and gain insight into underlying carcinogenic mechanisms through a pathway analysis. We show the usefulness of CaGe by assessing its performance in annotating somatic mutations from a published small cell lung cancer study.

• The KIPS Transactions:PartD
• /
• v.14D no.1 s.111
• /
• pp.35-44
• /
• 2007

Recently, an increasing in quantity of multimedia data have brought a new problem that expected data should be retrieved fast and exactly. The adequate representation for the multimedia data is the key element for efficient retrieval. For this reason, MPEG-7 standard was established for description of multimedia data. In this paper, we propose a new approach to metadata production. The user can decompose a given content into units and easily annotate each unit by adding basic Information such as time, place, etc. as well as classification information such as event, relationship, etc. according to the MPEG-7 standard. The objective is to build automatically a pure semantic description; the nodes are the events and the links are the graphs which describe the relationships among the events. Finally, we have implemented an annotation tool(SMAT) for semantic description based on proposed technique and assess some of the experiment results. In conclusion, we ran say that the proposod annotation tool is characterized by two important proprieties : reusability and extendibility.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2019.11a
• /
• pp.162-164
• /
• 2019

본 논문에서는 물체인식 딥러닝 모델 생성에 필요한 라벨링(Labeling)과정에서 사용자가 다양한 기능을 활용하여 효과적인 학습 데이터를 구성할 수 있는 GUI 프로그램을 구현했다. 프로그램의 인터페이스는 파이썬 기반의 GUI 모듈인 Tkinter 를 활용하여, 실시간으로 이미지 데이터를 수집할 수 있는 크롤링(Crawling)기능과 미리 학습된 Retinanet 을 통해 이미지 데이터를 인식함으로써 자동으로 주석(Annotation) 과정을 수행할 수 있는 기능을 구성했다. 또한, 수집한 이미지 데이터를 다양한 효과와 노이즈, 변형 등으로 Augmentation 기능을 추가함으로써, 사용자가 모델을 학습하기 위한 데이터 전처리 단계를 하나의 GUI 프로그램에서 수행할 수 있도록 했다. 또한 사용자가 직접 학습한 모델을 추정 모델(Inference Model)로 변환하여 프로그램에 입력할 수 있도록 설계한다.

• Annual Conference on Human and Language Technology
• /
• 2018.10a
• /
• pp.84-87
• /
• 2018

한국어 감정분석에 대한 연구는 활발하게 진행되고 있다. 그렇지만 학습 및 평가 말뭉치 표현에 대한 논의가 부족하다. 본 논문은 한국어 감정분석에 대해 정의하고, 말뭉치 제작을 위한 가이드라인을 제시한다. 또한, 태깅 가이드라인에 따라 말뭉치를 구축하였으며 한국어 감정분석을 위한 반자동 태깅 도구를 구현하였다.

• Genomics & Informatics
• /
• v.7 no.4
• /
• pp.217-219
• /
• 2009

The primary clue for locating protein-coding regions is the open reading frame and the determination of ORFs (Open Reading Frames) is the first step toward the gene prediction, especially for prokaryotes. In this respect, we have developed a web-based ORF search tool called ORF Miner. The ORF Miner is a graphical analysis utility which determines all possible open reading frames of a selectable minimum size in an input sequence. This tool identifies all open reading frames using alternative genetic codes as well as the standard one and reports a list of ORFs with corresponding deduced amino acid sequences. The ORF Miner can be employed for sequence annotation and give a crucial clue to determination of actual protein-coding regions.

• The Journal of Korean Association of Computer Education
• /
• v.19 no.2
• /
• pp.73-85
• /
• 2016

When using computer-supported collaborative learning(CSCL) tool where learners with different perspective are discussing in the same place, it is difficult to catch connection between specific position of learning content and written message by learners, share each other knowledge and opinion. This study investigated the phenomena responsible for causing "Un-shared knowledge" by literature research and designed collaborative representation tool based on described principle. A version of the Mediated annotation tool was developed for each problem identified and was subsequently tested by software evaluation standard and F.G.I. Final mediated tool will contribute to overcome knowledge sharing barriers, facilitate communication among students and describe high quality solution.

• Genomics & Informatics
• /
• v.10 no.1
• /
• pp.58-64
• /
• 2012

Intron prediction is an important problem of the constantly updated genome annotation. Using two model plant (rice and $Arabidopsis$) genomes, we compared two well-known intron prediction tools: the Blast-Like Alignment Tool (BLAT) and Sim4cc. The results showed that each of the tools had its own advantages and disadvantages. BLAT predicted more than 99% introns of whole genomic introns with a small number of false-positive introns. Sim4cc was successful at finding the correct introns with a false-negative rate of 1.02% to 4.85%, and it needed a longer run time than BLAT. Further, we evaluated the intron information of 10 complete plant genomes. As non-coding sequences, intron lengths are not limited by a triplet codon frame; so, intron lengths have three phases: a multiple of three bases (3n), a multiple of three bases plus one (3n + 1), and a multiple of three bases plus two (3n + 2). It was widely accepted that the percentages of the 3n, 3n + 1, and 3n + 2 introns were quite similar in genomes. Our studies showed that 80% (8/10) of species were similar in terms of the number of three phases. The percentages of 3n introns in $Ostreococcus$ $lucimarinus$ was excessive (47.7%), while in $Ostreococcus$ $tauri$, it was deficient (29.1%). This discrepancy could have been the result of errors in intron prediction. It is suggested that a three-phase evaluation is a fast and effective method of detecting intron annotation problems.

• Journal of Computing Science and Engineering
• /
• v.2 no.2
• /
• pp.162-179
• /
• 2008

We have developed a tool for annotation of electronic health record (EHR) data. Currently we are in the process of manually annotating a corpus of Norwegian general practitioners' EHRs with mainly linguistic information. The purpose of this project is to attain a linguistically annotated corpus of patient histories from general practice. This corpus will be put to future use in medical language processing and information extraction applications. The paper outlines some of our practical experiences from developing such a corpus and, in particular, the effects of semi-automated annotation. We have also done some preliminary experiments with part-of-speech tagging based on our corpus. The results indicated that relevant training data from the clinical domain gives better results for the tagging task in this domain than training the tagger on a corpus form a more general domain. We are planning to expand the corpus annotations with medical information at a later stage.

• Genomics & Informatics
• /
• v.19 no.3
• /
• pp.31.1-31.9
• /
• 2021

Leptospirosis is a zoonotic disease caused by spirochetes from the genus Leptospira. In Thailand, Leptospira interrogans is a major cause of leptospirosis. Leptospirosis patients present with a wide range of clinical manifestations from asymptomatic, mild infections to severe illness involving organ failure. For better understanding the difference between Leptospira isolates causing mild and severe leptospirosis, illumina sequencing was used to sequence genomic DNA in both serotypes. DNA of Leptospira isolated from two patients, one with mild and another with severe symptoms, were included in this study. The paired-end reads were removed adapters and trimmed with Q30 score using Trimmomatic. Trimmed reads were constructed to contigs and scaffolds using SPAdes. Cross-contamination of scaffolds was evaluated by ContEst16s. Prokka tool for bacterial annotation was used to annotate sequences from both Leptospira isolates. Predicted amino acid sequences from Prokka were searched in EggNOG and David gene ontology database to characterize gene ontology. In addition, Leptospira from mild and severe patients, that passed the criteria e-value < 10e^-5 from blastP against virulence factor database, were used to analyze with Venn diagram. From this study, we found 13 and 12 genes that were unique in the isolates from mild and severe patients, respectively. The 12 genes in the severe isolate might be virulence factor genes that affect disease severity. However, these genes should be validated in further study.