• Proceedings of the Korea Information Processing Society Conference
• /
• 2008.05a
• /
• pp.69-72
• /
• 2008

본 논문은 언어의 통계적 특징을 이용하여 범용의 문장경계 인식기를 제안한다. 제안하는 방법은 대량의 코퍼스 내에서 사용되고 있는 문장 경계를 기준으로 음절 및 어절 등의 자질을 이용하여 통계적 특징을 추출하고 다양한 기계학습 기법을 사용하여 문장경계를 인식하고자 하였다. 또한 특정 언어나 도메인에 제한적이지 않고 범용적인 자질만을 사용하려고 노력하였다. 언어의 특성상 문장의 구분이 애매한 경우 또는 잘못 사용 된 구두점 등의 경우에도 적용 가능하도록 다양한 자질을 사용하여 실험하였으며, 한국어와 영문 코퍼스에 대해서 동일한 자질을 적용하여 실험하여 본 논문에서 제시한 자질들이 한국어 및 다른 언어권의 언어에도 적용될 수 있는 범용적인 자질임을 확인할 수 있었다. 한국어 문장경계 인식을 위한 기계학습 및 실험을 위해서 세종계획 코퍼스를 사용하였으며, 성능척도로는 정확률과 재현율을 사용하였으며, 실험결과 제안한 방법으로 99%의 정확률과 99.2%의 재현율을 보였다. 영문의 경우는 Wall Street Journal 코퍼스를 사용하였으며, 동일한 자질을 적용하여 실험한 결과 98.9%의 정확률과 94.6%의 재현율을 보였다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2024.05a
• /
• pp.731-732
• /
• 2024

본 논문은 음원 메타데이터 임베딩 방법론을 기반으로 새로운 음원 추천 방법을 제안한다. 사용자 행동 데이터를 활용한 개인 맞춤형 음악 추천 모델은 신규 사용자의 데이터가 부족할 경우, 적절한 추천이 어려운 콜드스타트 현상을 초래할 수 있다. 본 연구에서는 플레이리스트의 음원 메타데이터를 Song sentence 로 구성하고, 고차원 벡터 공간에 임베딩하여 유사도를 계산한 추천 알고리즘을 구축한다. 사용자 행동 데이터가 아닌 음원의 자체적인 정보에 근거하기 때문에 콜드 스타트 현상을 보완하여 사용자에게 편리한 음악 감상 경험을 제공할 수 있을 것으로 기대된다.

• Asia-pacific Journal of Multimedia Services Convergent with Art, Humanities, and Sociology
• /
• v.8 no.6
• /
• pp.493-503
• /
• 2018

The present study is a conceptual replication of O'Grady, Lee, and Choo's (2003) earlier study designed to investigate two hypotheses (linear distance hypothesis vs. structural distance hypothesis) in relation to L2 Korean learners' processing of Korean subject and object relative clauses (RCs) in a scholarly attempt to explicate Keenan and Comrie's (1977) Noun Phrase Accessibility Hierarchy (NPAH). In addition, the current study is intended to explore any potential role of working memory capacity (WMC) in the processing of Korean subject and/or object RCs. Chinese-speaking learners of Korean taking a language course offered at a local university in Korea participated in this experimental study. Among those recruited, only 23 learners completed the experimental tasks appropriately according to the specific instructions provided on each task, and thus, subsequent statistical analyses were conducted on their data. Fifteen Korean NSs were also recruited for the control group. Two experimental tasks were administerd to the participants: one picture selection task containing the same test items used in O'Grady et al.'s study to measure their processing of subject-object RCs and an operation span (OSPAN) task to measure their WMC. Somewhat differently from O'Grady et al.'s findings, the participating Chinese learners of Korean performed significantly better on object RCs than on subject RCs, seemingly lending support to the linear distance hypothesis. Further analyses, however, suggested that the results in favor of, or relative ease of processing, object relative clauses were due, most likely, to the learners' excessive use of the canonical sentence strategy, which also led to nonsignificant correlations between WMC and learner performance on the picture selection task.

• Language and Information
• /
• v.18 no.2
• /
• pp.59-79
• /
• 2014

This paper aims at exploring the processing mechanism of filler-gap dependency resolution and feature checking in Korean wh-constructions. Based on their findings on Japanese sentence processing, Aoshima et al. (2004) have argued that the parser posits a gap in the embedded clause in head-final languages, unlike in head-initial languages, where the parser posits a gap in the matrix clause. In order to verify their findings in the Korean context, and to further explore the mechanisms involved in processing Korean wh-constructions, the present study replicated the study done by Aoshima et al., with some modifications of problematic areas in their original design. Sixty-four Korean native speakers were presented Korean sentences containing a wh-phrase in four conditions, with word order and complementizer type as the two main factors. The participants read sentences segment-by-segment, and the reading times at each segment were measured. The reading time analysis showed that there was no such slowdown at the embedded verb in the scrambled conditions as observed in Aoshima et al. Instead, there was a clear indication of the wh-feature checking process in terms of a major slowdown at the relevant region.

• Journal of Educational Research in Mathematics
• /
• v.26 no.4
• /
• pp.635-662
• /
• 2016

Combinatorics, the basis of probabilistic thinking, is an important area of mathematics and closely linked with other subjects such as informatics and STEAM areas. But combinatorics is one of the most difficult units in school mathematics for leaning and teaching. This study, using the designed combinatorial models and executable expression, aims to analyzes the eye movement of graduate students when they translate the written combinatorial problems to the corresponding executable expression, and examines not only the understanding process of the written combinatorial sentences but also the degree of difficulties depending on the combinatorial semantic structures. The result of the study shows that there are two types of solving process the participants take when they solve the problems : one is to choose the right executable expression by comparing the sentence and the executable expression frequently. The other approach is to find the corresponding executable expression after they derive the suitable mental model by translating the combinatorial sentence. We found the cognitive processing patterns of the participants how they pay attention to words and numbers related to the essential informations hidden in the sentence. Also we found that the student's eyes rest upon the essential combinatorial sentences and executable expressions longer and they perform the complicated cognitive handling process such as comparing the written sentence with executable expressions when they try the problems whose meaning structure is rarely used in the school mathematics. The data of eye movement provide meaningful information for analyzing the cognitive process related to the solving process of the participants.

• Journal of KIISE:Software and Applications
• /
• v.32 no.5
• /
• pp.406-415
• /
• 2005

Prosperity of Internet led to the web containing huge number of documents. Thus increasing importance is given to the web information retrieval technology that can provide users with documents that contain the right information they want. This paper proposes several techniques that are effective for the improvement of web information retrieval. Similarity between a document and the query is a major source of information exploited by conventional systems. However, we suggest a technique to make use of similarity between a sentence and the query. We introduce a technique to compute the approximate score of the sentence-query similarity even without a mature technology of natural language processing. It was shown that the amount of computation for this task is linear to the number of documents in the total collection, which implies that practical systems can make use of this technique. The next important technique proposed in this paper is to use stratification of documents in re-ranking the documents to output. It was shown that it can lead to significant improvement in performance. We furthermore showed that using hyper links, anchor texts, and titles can result in enhancement of performance. To justify the proposed techniques we developed a large scale web information retrieval system and used it for experiments.

• KIPS Transactions on Software and Data Engineering
• /
• v.5 no.1
• /
• pp.21-28
• /
• 2016

The performance of an English-Korean machine translation system depends heavily on its English parser. The parser in this paper is a part of the rule-based English-Korean MT system, which includes many syntactic rules and performs the chart-based parsing. The parser generates too many structures due to many syntactic rules, so much time and memory are required. The rule-based parser has difficulty in analyzing and translating the long sentences including the commas because they cause high parsing complexity. In this paper, we propose the 3-phase parsing method with sentence segmentation to efficiently translate the long sentences appearing in usual. Each phase of the syntactic analysis applies its own independent syntactic rules in order to reduce parsing complexity. For the purpose, we classify the syntactic rules into 3 classes and design the 3-phase parsing algorithm. Especially, the syntactic rules in the 3rd class are for the sentence structures composed with commas. We present the automatic rule acquisition method for 3rd class rules from the syntactic analysis of the corpus, with which we aim to continuously improve the coverage of the parsing. The experimental results shows that the proposed 3-phase parsing method is superior to the prior parsing method using only intra-sentence segmentation in terms of the parsing speed/memory efficiency with keeping the translation quality.

• Journal of Intelligence and Information Systems
• /
• v.26 no.4
• /
• pp.1-25
• /
• 2020

Sentiment Analysis (SA) is a Natural Language Processing (NLP) task that analyzes the sentiments consumers or the public feel about an arbitrary object from written texts. Furthermore, Aspect-Based Sentiment Analysis (ABSA) is a fine-grained analysis of the sentiments towards each aspect of an object. Since having a more practical value in terms of business, ABSA is drawing attention from both academic and industrial organizations. When there is a review that says "The restaurant is expensive but the food is really fantastic", for example, the general SA evaluates the overall sentiment towards the 'restaurant' as 'positive', while ABSA identifies the restaurant's aspect 'price' as 'negative' and 'food' aspect as 'positive'. Thus, ABSA enables a more specific and effective marketing strategy. In order to perform ABSA, it is necessary to identify what are the aspect terms or aspect categories included in the text, and judge the sentiments towards them. Accordingly, there exist four main areas in ABSA; aspect term extraction, aspect category detection, Aspect Term Sentiment Classification (ATSC), and Aspect Category Sentiment Classification (ACSC). It is usually conducted by extracting aspect terms and then performing ATSC to analyze sentiments for the given aspect terms, or by extracting aspect categories and then performing ACSC to analyze sentiments for the given aspect category. Here, an aspect category is expressed in one or more aspect terms, or indirectly inferred by other words. In the preceding example sentence, 'price' and 'food' are both aspect categories, and the aspect category 'food' is expressed by the aspect term 'food' included in the review. If the review sentence includes 'pasta', 'steak', or 'grilled chicken special', these can all be aspect terms for the aspect category 'food'. As such, an aspect category referred to by one or more specific aspect terms is called an explicit aspect. On the other hand, the aspect category like 'price', which does not have any specific aspect terms but can be indirectly guessed with an emotional word 'expensive,' is called an implicit aspect. So far, the 'aspect category' has been used to avoid confusion about 'aspect term'. From now on, we will consider 'aspect category' and 'aspect' as the same concept and use the word 'aspect' more for convenience. And one thing to note is that ATSC analyzes the sentiment towards given aspect terms, so it deals only with explicit aspects, and ACSC treats not only explicit aspects but also implicit aspects. This study seeks to find answers to the following issues ignored in the previous studies when applying the BERT pre-trained language model to ACSC and derives superior ACSC models. First, is it more effective to reflect the output vector of tokens for aspect categories than to use only the final output vector of [CLS] token as a classification vector? Second, is there any performance difference between QA (Question Answering) and NLI (Natural Language Inference) types in the sentence-pair configuration of input data? Third, is there any performance difference according to the order of sentence including aspect category in the QA or NLI type sentence-pair configuration of input data? To achieve these research objectives, we implemented 12 ACSC models and conducted experiments on 4 English benchmark datasets. As a result, ACSC models that provide performance beyond the existing studies without expanding the training dataset were derived. In addition, it was found that it is more effective to reflect the output vector of the aspect category token than to use only the output vector for the [CLS] token as a classification vector. It was also found that QA type input generally provides better performance than NLI, and the order of the sentence with the aspect category in QA type is irrelevant with performance. There may be some differences depending on the characteristics of the dataset, but when using NLI type sentence-pair input, placing the sentence containing the aspect category second seems to provide better performance. The new methodology for designing the ACSC model used in this study could be similarly applied to other studies such as ATSC.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.20 no.7
• /
• pp.158-165
• /
• 2019

Although evaluation of writing disabilities identification and timely intervention are required, clinicians adopt a manual scoring method and there is a possibility of error due to subjective evaluation. In this study, the size ratio and position of letters are digitized and quantified through image processing of offline handwritten characters. We tried to evaluate objectively and accurately the performance of writing through comparison with existing methods. From November 12th to 16th, 2018, 20 adults without neurological injury were selected. They used a pencil to follow the 10 words, 2 sentence stimuli after keeping the usual habit, and we collected the writing test data. The results showed that the height of the word was 1.2 times larger than the width and it tilted to the lower left. The spacing interval was 9mm on average. In the Paired T test, a high correlation was showed between our system and existing methods in the word and sentence 2. This demonstrated the possibility as a testing tool. This study evaluated objectively and precisely writing performance of offline handwritten characters through image processing and provided preliminary data for performance standards. In the future, it can be suggested as a basic data on writing diagnosis of various ages.

• Journal of Internet Computing and Services
• /
• v.13 no.2
• /
• pp.99-108
• /
• 2012

Many methods had been developed to improve the accuracy of extracting information from a vast amount of data. This paper combined a number of natural language processing methods such as NER (named entity recognition), sentence extraction, and part of speech tagging to carry out text analysis. The data source is comprised of texts obtained from the web using a domain-specific data extraction agent. A framework for the extraction of information from unstructured data was developed using the aforementioned natural language processing methods. We simulated the performance of our work in the extraction and analysis of texts for the detection of organizational structures. Simulation shows that our study outperformed other NER classifiers such as MUC and CoNLL on information extraction.