• Korean Journal of Cognitive Science
• /
• v.20 no.2
• /
• pp.225-241
• /
• 2009

This paper presents an authorship attribution study in Korean conducted on a corpus of newspaper column texts. Based on the data set consisting of a total of 160 columns written by four columnists of Chosun Daily, the approach utilizes relative frequencies of various lexical units in Korean such as fully inflected words, morphemes, syllables and their bigrams in an attempt to establish authorship of a blind text selected from the set. Among these various lexical units, "the morpheme" is found to be most effective in predicting who among the four potential candidates authored a text, reporting accuracies of over 93%. The results indicate that quantitative and statistical techniques in authorship attribution and computational stylistics can be successfully applied to Korean texts.

• Journal of Information Processing Systems
• /
• v.16 no.2
• /
• pp.329-342
• /
• 2020

In recent years, emotional text classification is one of the essential research contents in the field of natural language processing. It has been widely used in the sentiment analysis of commodities like hotels, and other commentary corpus. This paper proposes an improved W-LDA (weighted latent Dirichlet allocation) topic model to improve the shortcomings of traditional LDA topic models. In the process of the topic of word sampling and its word distribution expectation calculation of the Gibbs of the W-LDA topic model. An average weighted value is adopted to avoid topic-related words from being submerged by high-frequency words, to improve the distinction of the topic. It further integrates the highest classification of the algorithm of support vector machine based on the extracted high-quality document-topic distribution and topic-word vectors. Finally, an efficient integration method is constructed for the analysis and extraction of emotional words, topic distribution calculations, and sentiment classification. Through tests on real teaching evaluation data and test set of public comment set, the results show that the method proposed in the paper has distinct advantages compared with other two typical algorithms in terms of subject differentiation, classification precision, and F1-measure.

• Journal of Biomedical Engineering Research
• /
• v.36 no.4
• /
• pp.128-134
• /
• 2015

Recently, the structured data format in RDF/OWL has played an increasingly vital role in the semantic web. We converted pediatric and oncologic nuclear medicine imaging reports in free text into RDF/OWL format and evaluated the usefulness of nuclear medicine imaging reports in RDF/OWL by comparing SPARQL query results with the manually retrieved results by physicians from the reports in free text. SPARQL query showed 95% recall for simple queries and 91% recall for dedicated queries. In total, SPARQL query retrieved 93% (51 lesions of 55) recall and 100% precision for 20 clinical query items. All query results missed by SPARQL query were of some inference. Nuclear medicine imaging reports in the format of RDF/OWL were very useful for retrieving simple and dedicated query results using SPARQL query. Further study using more number of cases and knowledge for inference is warranted.

• Journal of the Korea Safety Management & Science
• /
• v.17 no.4
• /
• pp.335-342
• /
• 2015

The increasing interests on patents have led many individuals and companies to apply for many patents in various areas. Applied patents are stored in the forms of electronic documents. The search and categorization for these documents are issues of major fields in data mining. Especially, the keyword extraction by which we retrieve the representative keywords is important. Most of techniques for it is based on vector space model. But this model is simply based on frequency of terms in documents, gives them weights based on their frequency and selects the keywords according to the order of weights. However, this model has the limit that it cannot reflect the relations between keywords. This paper proposes the advanced way to extract the more representative keywords by overcoming this limit. In this way, the proposed model firstly prepares the candidate set using the vector model, then makes the graph which represents the relation in the pair of candidate keywords in the set and selects the keywords based on this relationship graph.

• Annual Conference on Human and Language Technology
• /
• 1999.10e
• /
• pp.184-190
• /
• 1999

본 논문에서는 Text-to-Speech 시스템에서 사용할 억양 모델을 위해 음성 DB에서 모델 파라미터와 피치 검색테이블(lookup-table)을 추출하여 미리 구성하고, 합성시에는 이를 추정하여 최종 F0 값을 생성하는 자료기반 접근방식(data-driven approach)을 사용한다. 어절 경계강도(break-index)는 경계강도의 특성에 따라 고정적 경계강도와 가변적 경계강도로 세분화하여 사용하였고, 예측된 경계강도를 기준으로 억양구(Intonation Phrase)와 액센트구(Accentual Phrase)를 설정하였다. 특히, 액센트구 모델은 인지적, 음향적으로 중요한 정점(peak)을 정확하게 모델링하는 것에 주안점을 두어 정점(peak)의 시간축, 주파수축 값과 이를 기준으로 한 앞뒤 기울기를 추정하여 4개의 파라미터로 설정하였고, 이 파라미터들은 CART(Classification and Regression Tree)를 이용하여 예측규칙을 만들었다. 경계음조가 나타나는 조사, 어미는 정규화된(normalized) 피치값과 key-index로 구성되는 검색테이블을 만들어 보다 정교하게 피치값을 예측하였다. 본 논문에서 제안한 억양 모델을 본 연구실에서 제작한 음성합성기를 통해 합성하여 청취실험을 거친 결과, 기존의 상용 Text-to-Speech 시스템에 비해 자연스러운 합성음을 얻을 수 있었다.

• International Journal of Contents
• /
• v.3 no.3
• /
• pp.6-14
• /
• 2007

In this paper, we design and implement a XML Repository System(XRS) that exploits the advantages of DBMSs and IRSs. Our scheme uses BRS to support full text indexing and content-based queries efficiently, and ORACLE to store XML documents, multimedia data, DTD and structure information. We design databases to manage XML documents including audio, video, images as well as text. We employ the non-composition model when storing XML documents into ORACLE. We represent structured information as ETID(Element Type Id), SORD(Sibling ORDer) and SSORD(Same Sibling ORDer). ETID is a unique value assigned to each element of DTD. SORD and SSORD represent an order information between sibling nodes and an order information among the sibling nodes with the same element respectively. In order to show superiority of our XRS, we perform various experiments in terms of the document loading time, document extracting time and contents retrieval time. It is shown through experiments that our XRS outperforms the existing XML document management systems. We also show that it supports various types of queries through performance experiments.

• Journal of the Korea Society of Computer and Information
• /
• v.23 no.10
• /
• pp.103-109
• /
• 2018

In this paper, we propose a dynamic document classification method which breaks away from existing document classification method with artificial categorization rules focusing on suppliers and has changing categorization rules according to users' needs or social trends. The core of this dynamic document classification method lies in the fact that it creates classification criteria real-time by using topic modeling techniques without standardized category rules, which does not force users to use unnecessary frames. In addition, it can also search the details through the relevance analysis by calculating the relationship between the words that is difficult to grasp by word frequency alone. Rather than for logical and systematic documents, this method proposed can be used more effectively for situation analysis and retrieving information of unstructured data which do not fit the category of existing classification such as VOC (Voice Of Customer), SNS and customer reviews of Internet shopping malls and it can react to users' needs flexibly. In addition, it has no process of selecting the classification rules by the suppliers and in case there is a misclassification, it requires no manual work, which reduces unnecessary workload.

• International Journal of Advanced Culture Technology
• /
• v.7 no.4
• /
• pp.49-55
• /
• 2019

In the age of information, newly produced knowledge is mostly written in English. Therefore, there has been a strong demand for English language learning in the EFL context. However, most EFL learners possess a lack of interest and motivation in the text-based reading class. In this educational context, film is one of the most widely used materials in English reading classes considering that modern learners are predominantly familiar with various audiovisual materials. The purpose of this study is to investigate how Korean EFL learners experienced in the film-based reading class. Specifically, this study aims to analyze the EFL students' perceptions about the class and learning strategies that they used during the class. In order to comprehensively interpret the EFL learners' experiences in the classroom, a coding system consisting of five categories was developed: report, emotion, reflection, evaluation, future plans. The results of data analysis showed that the use of movies in English reading classes had positive effects on reading comprehension and inference of word meaning. The most frequently used learning strategies were affective strategies which helped them control their emotion, attitude, motivations and values, whereas memorization strategies were rarely used. In this respect, this study suggests that the use of movies in the EFL reading classroom encourage students' attention and help them obtain and activate schema which is useful in gaining a better understanding of text-based reading materials.

• Journal of Information Technology Services
• /
• v.14 no.3
• /
• pp.33-48
• /
• 2015

Recently, many users frequently share their opinions on diverse issues using various social media. Therefore, many governments have attempted to establish or improve national policies according to the public opinions captured from the various social media. In this paper, we indicate several limitations of traditional approaches for analyzing public opinions about science and technology and provide an alternative methodology to overcome the limitations. First of all, we distinguish science and technology analysis phase and social issue analysis phase to reflect the fact that public opinion can be formed only when a certain science and technology is applied to a specific social issue. Next, we apply a start list and a stop list successively to acquire clarified and interesting results. Finally, to identify most appropriate documents fitting to a given subject, we develop a new concept of logical filter that consists of not only mere keywords but also a logical relationship among keywords. This study then analyzes the possibilities for the practical use of the proposed methodology thorough its application to discovering core issues and public opinions from 1,700,886 documents comprising SNS, blog, news, and discussion.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.26 no.5
• /
• pp.1191-1199
• /
• 2016

To ensure the fairness, journal reviewers use blind-review system which hides the author information of the journal. Even though the author information is blinded, we could identify the author by looking at the field of the journal or containing words and phrases in the text. In this paper, we collected 315 journals of 20 authors and extracted text data. Bag-of-words were generated after preprocessing and used as an input of artificial neural network. The experiment shows the possibility of circumventing the blind review through identifying the author of the journal. By the experiment, we demonstrate the limitation of the current blind-review system and emphasize the necessity of robust blind-review system.