• Korean Journal of Cognitive Science
• /
• v.24 no.2
• /
• pp.95-109
• /
• 2013

With the rapid evolution of the personal device environment, the demand for natural language applications is increasing. This paper proposes a morpheme segmentation and part-of-speech tagging model, which provides the first step module of natural language processing for many languages; the model is designed for mobile devices with limited hardware resources. To reduce the number of morpheme candidates in morphological analysis, the proposed model uses a method that adds highly possible morpheme candidates to the original outputs of a conventional left-longest-match-preference method. To reduce the computational cost and memory usage, the proposed model uses a method that simplifies the process of calculating the observation probability of a word consisting of one or more morphemes in a conventional hidden Markov model.

• Journal of the Korean Geographical Society
• /
• v.45 no.2
• /
• pp.201-220
• /
• 2010

Geographical name is not only a spoken or written language that has been constructed as a linguistic element, but it is also a geographical phenomenon and a cultural element. Based upon this consciousness, the purpose of this paper is to examine current classification and management systems of the geographical names in Korea and to propose a new alternative classification and maintenance of the geographical names. In particular, the paper suggests three categories for the type classification of the geographical names: morpheme, linguistic change, and contestation types. In turn, this paper investigates the index of the geographical names contained in THE NATIONAL ATLAS OF KOREA (2007) on the basis of the preceding classification types in order to unveil the practical problems and limitations of the current classification articulated in the national atlas. This paper also proposes a new classification of the geographical naming that reflects the divisions of front and back morpheme of geographical names. Finally, from the discussions with the reinforcement of National Committee on Geographical Names, this paper invokes administrative and institutional protection and systematical management of the contesting and unofficial small scaled geographical names that have been set apart from the current geographical name standardization.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.21 no.2
• /
• pp.461-467
• /
• 2017

As the number of smart devices increases, e-book contents and services are proliferating. However, the text based e-book is difficult for a hearing-impairment person to understand. In this paper, we developed an android based application in which we can choose an e-book text file and each sentence is translated to sign-language elements which are shown in videos that are retrieved from the sign-language contents server. We used the korean sentence to sign-language translation algorithm based on the morpheme analysis. The proposed translation algorithm consists of 3 stages. Firstly, some elements in a sentence are removed for typical sign-language usages. Secondly, the tense of the sentence and the expression alteration are applied. Finally, the honorific forms are considered and word positions in the sentence are revised. We also proposed a new method to evaluate the performance of the translation algorithm and demonstrated the superiority of the algorithm through the translation results of 100 reference sentences.

• Journal of Convergence for Information Technology
• /
• v.10 no.12
• /
• pp.9-14
• /
• 2020

Currently, NTIS (National Technology Information Service) is building an interactive search service based on artificial intelligence technology. In order to understand users' search intentions and provide R&D information, an interactive search service is built based on deep learning models and morpheme analyzers. The deep learning model learns based on the log data loaded when using NTIS and interactive search services and understands the user's search intention. And it provides task information through step-by-step search. Understanding the search intent makes exception handling easier, and step-by-step search makes it easier and faster to obtain the desired information than integrated search. For future research, it is necessary to expand the range of information provided to users.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.17 no.4
• /
• pp.199-208
• /
• 2022

Korean is an agglutinative language, and one or more morphemes are combined to form a single word. Part-of-speech tagging method separates each morpheme from a word and attaches a part-of-speech tag. In this study, we propose a new Korean part-of-speech tagging method based on the Head-Tail tokenization technique that divides a word into a lexical morpheme part and a grammatical morpheme part without decomposing compound words. In this method, the Head-Tail is divided by the syllable boundary without restoring irregular deformation or abbreviated syllables. Korean part-of-speech tagger was implemented using the Head-Tail tokenization and deep learning technique. In order to solve the problem that a large number of complex tags are generated due to the segmented tags and the tagging accuracy is low, we reduced the number of tags to a complex tag composed of large classification tags, and as a result, we improved the tagging accuracy. The performance of the Head-Tail part-of-speech tagger was experimented by using BERT, syllable bigram, and subword bigram embedding, and both syllable bigram and subword bigram embedding showed improvement in performance compared to general BERT. Part-of-speech tagging was performed by integrating the Head-Tail tokenization model and the simplified part-of-speech tagging model, achieving 98.99% word unit accuracy and 99.08% token unit accuracy. As a result of the experiment, it was found that the performance of part-of-speech tagging improved when the maximum token length was limited to twice the number of words.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2021.10a
• /
• pp.373-375
• /
• 2021

In the case of Korean, the possibility of development is lower than that of English if tagging is done through the word tokenization like English. Although the form of tokenizing the corpus by separating it into morpheme units via KoNLPy is represented as a graph database, full separation of voice files and verification of practicality is required when converting the module from graph database to corpus. In this paper, morphology representation using STT API is shown in Raspberry Pi. The voice file converted to Corpus is analyzed to KoNLPy and tagged. The analyzed results are represented by graph databases and can be divided into tokens divided by morpheme, and it is judged that data mining extraction with specific purpose is possible by determining practicality and degree of separation.

• Smart Media Journal
• /
• v.10 no.1
• /
• pp.16-24
• /
• 2021

The exiting word embedding methodology such as word2vec represents words, which only occur in the raw training corpus, as a fixed-length vector into a continuous vector space, so when mapping the words incorporated in the raw training corpus into a fixed-length vector in morphologically rich language, out-of-vocabulary (OOV) problem often happens. Even for sentence embedding, when representing the meaning of a sentence as a fixed-length vector by synthesizing word vectors constituting a sentence, OOV words make it challenging to meaningfully represent a sentence into a fixed-length vector. In particular, since the agglutinative language, the Korean has a morphological characteristic to integrate lexical morpheme and grammatical morpheme, handling OOV words is an important factor in improving performance. In this paper, we propose parallel Tri-LSTM sentence embedding that is robust to the OOV problem by extending utilizing the morphological information of words into sentence-level. As a result of the sentiment analysis task with corpus in Korean, we empirically found that the character unit is better than the morpheme unit as an embedding unit for Korean sentence embedding. We achieved 86.17% accuracy on the sentiment analysis task with the parallel bidirectional Tri-LSTM sentence encoder.

• Smart Media Journal
• /
• v.11 no.5
• /
• pp.17-25
• /
• 2022

Korean part-of-speech taggers decompose a compound morpheme into unit morphemes and attach part-of-speech tags. So, here is a disadvantage that part-of-speech for morphemes are over-classified in detail and complex word types are generated depending on the purpose of the taggers. When using the part-of-speech tagger for keyword extraction in deep learning based language processing, it is not required to decompose compound particles and verb-endings. In this study, the part-of-speech tagging problem is simplified by using a Head-Tail tokenization technique that divides only two types of tokens, a lexical morpheme part and a grammatical morpheme part that the problem of excessively decomposed morpheme was solved. Part-of-speech tagging was attempted with a statistical technique and a deep learning model on the Head-Tail tokenized corpus, and the accuracy of each model was evaluated. Part-of-speech tagging was implemented by TnT tagger, a statistical-based part-of-speech tagger, and Bi-LSTM tagger, a deep learning-based part-of-speech tagger. TnT tagger and Bi-LSTM tagger were trained on the Head-Tail tokenized corpus to measure the part-of-speech tagging accuracy. As a result, it showed that the Bi-LSTM tagger performs part-of-speech tagging with a high accuracy of 99.52% compared to 97.00% for the TnT tagger.

• 중국학논총
• /
• no.69
• /
• pp.1-23
• /
• 2021

所謂本義一般在中國文字學裡就是漢字所表示的本來意義。例如"初"字在 《說文解字》 裡說 : "始也。從刀從衣。裁衣之始也。"根據 《說文解字》 "初"字的本義是"開始"。可是從它的字素"從刀從衣"來看, 沒辦法了解"初"字的本義就是"開始"。"從刀從衣"只能表示"用刀裁衣"。如果沒有說"裁衣之始也", 很難理解"初"字的本義就是"開始"。我們這裡要把"構意"和"本義"區分開來, 才能了解到字的"本義"。我們確定"本義"時, 必須考慮三個概念1)字形結構, 2)字形表示的意義, 3)引申義, 才能掌握"字義"的演變和發展。掌握"字的本義"也很重要, 還有更重要的就是"詞素的引申義", 我們從"詞素的引申義", 可以看出"文化特質", 比如漢語說"油老虎", 可是韓國人習慣用"油河馬"的說法。

• MALSORI
• /
• no.37
• /
• pp.49-71
• /
• 1999

The Purpose of this paper is to gather pairs of heteronyms in modern Korean and transcribe them in the IPA and the Roman alphabet in order to propose that all of them should be differentiated in Hanngul orthography. More than a quarter of the whole Korean vocabulary consists of words with a long vowel and the number of minimal pairs distinguished only by the chroneme reaches nearly ten thousand (i.e. twenty thousand words). The letter h syllable-finally is used here to represent the long vowel in Romanization except the vowel '으‘[?:] which is transcribed by doubling the letter u (i.e. uu). Another factor bringing forth lots of heteronyms in Korean is the lack of full indication as to the non-automatic reinforcement in the initial consonant of a word (or a morpheme) when preceded by another within a phrase (or a word). These reinforced word-initial consonants are written with the letter c and an apostrophe (like c'g- , c'd- , c'b-, c's-, c'j-) in Romanization here. The reinforced morpheme-initial consonant within a word is written with the letters k t, p, ss and cz for ㄲ, ㄸ, ㅃ, ㅆ and ㅉ sounds respectively. The contrasted pronunciations of pairs of heteronyms beginning with ㅅ /s/sup h// and ㅆ /s/ sounds are transcribed here for exemplification.