• Annual Conference on Human and Language Technology
• /
• 2018.10a
• /
• pp.167-171
• /
• 2018

개체명 인식은 문서에서 인명, 지명, 기관명 등의 고유한 의미를 나타내는 단위인 개체명을 추출하고, 추출된 개체명의 범주를 결정하는 작업이다. 최근 개체명 인식과 관련된 연구는 입력 데이터의 앞, 뒤를 고려하기 위한 Bi-RNNs와 출력 데이터 간의 전이 확률을 이용한 CRFs를 결합한 방식을 기반으로 다양한 변형의 심층학습 방법론이 제안되고 있다. 그러나 대부분의 연구는 입력 단위를 단어나 형태소로 사용하고 있으며, 성능 향상을 위해 띄어쓰기 정보, 개체명 사전 자질, 품사 분포 정보 등 다양한 정보를 필요로 한다는 어려움이 있다. 본 논문은 기본적인 학습 말뭉치에서 얻을 수 있는 문자 기반의 입력 정보와 Multi-Head Attention을 추가한 Bi-GRU/CRFs을 이용한 다국어 개체명 인식 방법을 제안한다. 한국어, 일본어, 중국어, 영어에 제안 모델을 적용한 결과 한국어와 일본어에서는 우수한 성능(한국어 $F_1$ 84.84%, 일본어 $F_1$ 89.56%)을 보였다. 영어에서는 $F_1$ 80.83%의 성능을 보였으며, 중국어는 $F_1$ 21.05%로 가장 낮은 성능을 보였다.

• The KIPS Transactions:PartB
• /
• v.17B no.3
• /
• pp.249-254
• /
• 2010

We propose an automatic spam filter for e-mail data using Support Vector Machines(SVM). We use a lexical form of a word and its part of speech(POS) tags as features and select features by chi square statistics. We represent each feature by TF(text frequency), TF-IDF, and binary weight for experiments. After training SVM with the selected features, SVM classifies each e-mail as spam or not. In experiment, the selected features improve the performance of our system and we acquired overall 98.9% of accuracy with TREC05-p1 spam corpus.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.23 no.4
• /
• pp.317-324
• /
• 2013

Twitter sentiment analysis is to classify a tweet (message) into positive and negative sentiment class. This study deals with SentiWordNet(SWN)-based twitter sentiment analysis. SWN is a sentiment dictionary in which each sense of an English word has a positive and negative sentimental strength. There has been a variety of SWN-based sentiment feature extraction methods which typically first determine the sentiment orientation (SO) of a term in a document and then decide SO of the document from such terms' SO values. For example, for SO of a term, some calculated the maximum or average of sentiment scores of its senses, and others computed the average of the difference of positive and negative sentiment scores. For SO of a document, many researchers employ the maximum or average of terms' SO values. In addition, the above procedure may be applied to the whole set (adjective, adverb, noun, and verb) of parts-of-speech or its subset. This work provides a comparative study on SWN-based sentiment feature extraction schemes with performance evaluation on a well-known twitter dataset.

• Journal of KIISE
• /
• v.44 no.3
• /
• pp.306-313
• /
• 2017

Named entity recognition (NER) seeks to locate and classify named entities in text into pre-defined categories such as names of persons, organizations, locations, expressions of times, etc. Recently, many state-of-the-art NER systems have been implemented with bidirectional LSTM CRFs. Deep learning models based on long short-term memory (LSTM) generally depend on word representations as input. In this paper, we propose an approach to expand word representation by using pre-trained word embedding, part of speech (POS) tag embedding, syllable embedding and named entity dictionary feature vectors. Our experiments show that the proposed approach creates useful word representations as an input of bidirectional LSTM CRFs. Our final presentation shows its efficacy to be 8.05%p higher than baseline NERs with only the pre-trained word embedding vector.

• The KIPS Transactions:PartB
• /
• v.10B no.1
• /
• pp.41-46
• /
• 2003

This thesis composes separated words in text for extracting keywords from Japanese, proposes extracting indexing keywords which consist of a compound noun using words and sentences information with the rules in the sentences. It constructs generative rules of compound nouns to be based In dependence as a result of analysing character of keywords in the text not the same way as before. To hold other extracting keywords and the content of sentence, and suggest how to decide importance concerned some restriction and repetition of words about generative rules. To verify the validity of keywords extracting, we have used titles and abstracts from Japanese thesis 65 files about natural language and/or voice processing, and obtain 63% in outputting one in the top rank.

• KIPS Transactions on Software and Data Engineering
• /
• v.8 no.2
• /
• pp.79-88
• /
• 2019

Machine translation (MT) system, especially Korean-Mongolian MT system, has recently attracted much attention due to its necessary for the globalization generation. Korean and Mongolian have the same sentence structure SOV and the arbitrarily changing of their words order does not change the meaning of sentences due to postpositional particles. The particles that are attached behind words to indicate their grammatical relationship to the clause or make them more specific in meaning. Hence, the particles play an important role in the translation between Korean and Mongolian. However, one Korean particle can be translated into several Mongolian particles. This is a major issue of the Korean-Mongolian MT systems. In this paper, to address this issue, we propose a method to use the combination of UTagger and a Korean-Mongolian particles table. UTagger is a system that can analyze morphologies, tag POS, and disambiguate homographs for Korean texts. The Korean-Mongolian particles table was manually constructed for matching Korean particles with those of Mongolian. The experiment on the test set extracted from the National Institute of Korean Language's Korean-Mongolian Learner's Dictionary shows that our method achieved the accuracy of 88.38% and it improved the result of using only UTagger by 41.48%.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.43 no.6 s.312
• /
• pp.43-51
• /
• 2006

This paper proposes a metrics for example matching under the example-based machine translation for English-Korean machine translation. Our metrics served as similarity measure is based on edit-distance algorithm, and it is employed to retrieve the most similar example sentences to a given query. Basically it makes use of simple information such as lemma and part-of-speech information of typographically mismatched words. Edit-distance algorithm cannot fully reflect the context of matched word units. In other words, only if matched word units are ordered, it is considered that the contribution of full matching context to similarity is identical to that of partial matching context for the sequence of words in which mismatching word units are intervened. To overcome this drawback, we propose the context-weighting scheme that uses the contiguity information of matched word units to catch the full context. To change the edit-distance metrics representing dissimilarity to similarity metrics, to apply this context-weighted metrics to the example matching problem and also to rank by similarity, we normalize it. In addition, we generalize previous methods using some linguistic information to one representative system. In order to verify the correctness of the proposed context-weighted metrics, we carry out the experiment to compare it with generalized previous methods.

• KIPS Transactions on Software and Data Engineering
• /
• v.3 no.6
• /
• pp.219-230
• /
• 2014

An analysis of dependency relation is a job that determines the governor and the dependent between words in sentence. The dependency relation of predicate is established by patterns and selectional restriction of subcategorization of the predicate. This paper proposes a method of analysis of Korean dependency relation using homograph predicate disambiguated in morphology analysis phase. The disambiguated homograph predicates has each different pattern. Especially reusing a stage transition training dictionary used during tagging POS and homograph, we propose a method of fixing the dependency relation of {noun+postposition, predicate}, and we analyze the accuracy and an effect of homograph for analysis of dependency relation. We used the Sejong Phrase Structured Corpus for experiment. We transformed the phrase structured corpus to dependency relation structure and tagged homograph. From the experiment, the accuracy of dependency relation by disambiguating homograph is 80.38%, the accuracy is increased by 0.42% compared with one of undisambiguated homograph. The Z-values in statistical hypothesis testing with significance level 1% is ${\mid}Z{\mid}=4.63{\geq}z_{0.01}=2.33$. So we can conclude that the homograph affects on analysis of dependency relation, and the stage transition training dictionary used in tagging POS and homograph affects 7.14% on the accuracy of dependency relation.

• Journal of the Korea Society of Computer and Information
• /
• v.26 no.4
• /
• pp.105-112
• /
• 2021

In the era of big data, interest in data is exploding. In particular, the development of the Internet and social media has led to the creation of new data, enabling the realization of the era of big data and artificial intelligence and opening a new chapter in convergence technology. Also, in the past, there are many demands for analysis of data that could not be handled by programs. In this paper, an analysis model was designed and verified for classification of unstructured data, which is often required in the era of big data. Data crawled DBPia's thesis summary, main words, and sub-keyword, and created a database using KoNLP's data dictionary, and tokenized words through morpheme analysis. In addition, nouns were extracted using KAIST's 9 part-of-speech classification system, TF-IDF values were generated, and an analysis dataset was created by combining training data and Y values. Finally, The adequacy of classification was measured by applying three analysis algorithms(random forest, SVM, decision tree) to the generated analysis dataset. The classification model technique proposed in this paper can be usefully used in various fields such as civil complaint classification analysis and text-related analysis in addition to thesis classification.