• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.14 no.3
• /
• pp.209-215
• /
• 2014

Sequence tagging is the task of predicting frame-wise labels for a given input sequence and has important applications to diverse domains. Conventional methods such as maximum likelihood (ML) learning matches global features in empirical and model distributions, rather than local features, which directly translates into frame-wise prediction errors. Recent probabilistic sequence models such as conditional random fields (CRFs) have achieved great success in a variety of situations. In this paper, we introduce a novel discriminative CRF learning algorithm to minimize local feature mismatches. Unlike overall data fitting originating from global feature matching in ML learning, our approach reduces the total error over all frames in a sequence. We also provide an efficient gradient-based learning method via gradient forward-backward recursion, which requires the same computational complexity as ML learning. For several real-world sequence tagging problems, we empirically demonstrate that the proposed learning algorithm achieves significantly more accurate prediction performance than standard estimators.

• Journal of KIISE:Computing Practices and Letters
• /
• v.15 no.5
• /
• pp.395-399
• /
• 2009

In this work, we describe the syntactic tag prediction system for Korean using the decision tree and CRFs. Generally they select features by their intuition. It depends on their prior knowledge. In this works, we combine features systematically using the decision tree. We also analyze errors and optimize features for the best performance. From the result of experiments, we can see that the proposed method is effective for the syntactic tag estimation and will be helpful for the syntactic analysis.

• Annual Conference on Human and Language Technology
• /
• 2008.10a
• /
• pp.23-28
• /
• 2008

한국어 처리에서 구문분석기에 대한 요구는 많은 반면 성능의 한계와 강건함의 부족으로 인해 채택되지 못하는 것이 현실이다. 본 연구는 구문분석을 레이블링 문제로 전환하여 성능, 속도, 강건함을 모두 실현한 시스템에 대해서 설명한다. 우리는 다단계 구 단위화(Cascaded Chunking)를 통해 한국어 구문분석을 시도한다. 각 단계에서는 어절별 품사 태그와 어절 구문표지를 자질로 사용하고 Conditional Random Fields(CRFs)를 이용하여 최적의 결과를 얻는다. 98,412문장 세종 구문 코퍼스로 학습하고 1,430문장(평균 14.59어절)으로 실험한 결과 87.30%의 구문 정확도를 보였다. 이 결과는 기존에 제안되었던 구문분석기와 대등하거나 우수한 성능이며 기존 구문분석기가 처리하지 못하는 장문도 처리 가능하다.

• Annual Conference on Human and Language Technology
• /
• 2010.10a
• /
• pp.100-104
• /
• 2010

개체명 인식은 정보 추출의 한 단계로서 정보검색 분야 뿐 아니라 질의응답과 요약 분야에서 매우 유용하게 사용되고 있다. 본 논문에서는 structural Support Vector Machines(structural SVMs) 및 수정된 Pegasos 알고리즘을 이용한 한국어 개체명 인식 시스템에 대하여 기술하고 기존의 Conditional Random Fields(CRFs)를 이용한 시스템과의 성능을 비교한다. 실험결과 structural SVMs과 수정된 Pegasos 알고리즘이 기존의 CRFs 보다 높은 성능을 보였고(신뢰도 99%에서 통계적으로 유의함), structural SVMs과 수정된 Pegasos 알고리즘의 성능은 큰 차이가 없음(통계적으로 유의하지 않음)을 알 수 있었다. 특히 본 논문에서 제안하는 수정된 Pegasos 알고리즘을 이용한 경우 CRFs를 이용한 시스템보다 높은 성능 (TV 도메인 F1=85.43, 스포츠 도메인 F1=86.79)을 유지하면서 학습 시간은 4%로 줄일 수 있었다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2010.04a
• /
• pp.505-507
• /
• 2010

웹(Web)상에 전문용어를 포함한 문서가 증가함에 따라 전문용어를 자동으로 추출하는 연구가 계속해서 이루어지고 있다. 기존 연구에서는 전문용어를 추출하는 단계에서 대부분 형태소 분석기를 이용한다. 그러나 전문용어의 특성으로 인해 형태소 분석 단계에서 오분석 되는 경우가 발생한다. 이러한 문제를 해결하기 위해서 본 논문에서는 음절 바이그램과 CRFs(Conditional Random Fields)를 이용하여 의학 전문 용어를 추출하는 방법을 제안한다. 네이버 지식인의 의사 답변 문서 2000개로부터 5-fold cross validation을 이용하여 실험하였다. 실험 결과 정확률은 평균 68.91%, 재현율은 평균 71.25%로 나타났으며 F-measure는 70.06%로 나타났다.

• Journal of KIISE
• /
• v.43 no.6
• /
• pp.678-685
• /
• 2016

Named Entity Recognition and Classification (NERC) is a task for recognition and classification of named entities such as a person's name, location, and organization. There have been various studies carried out on Korean NERC, but they have some problems, for example lacking some features as compared with English NERC. In this paper, we propose a method that uses word embedding as features for Korean NERC. We generate a word vector using a Continuous-Bag-of-Word (CBOW) model from POS-tagged corpus, and a word cluster symbol using a K-means algorithm from a word vector. We use the word vector and word cluster symbol as word embedding features in Conditional Random Fields (CRFs). From the result of the experiment, performance improved 1.17%, 0.61% and 1.19% respectively for TV domain, Sports domain and IT domain over the baseline system. Showing better performance than other NERC systems, we demonstrate the effectiveness and efficiency of the proposed method.

• Journal of KIISE:Software and Applications
• /
• v.37 no.11
• /
• pp.837-844
• /
• 2010

Prediction of the prosodic phrase boundary is one of the most important natural language processing tasks. We propose, for the natural prediction of the Korean prosodic phrase boundary, a statistical approach incorporating efficient learning features. These new features reflect the factors that affect generation of the prosodic phrase boundary better than existing learning features. Notably, moreover, such learning features, extracted according to the hand-crafted prosodic phrase boundary prediction rule, impart higher accuracy. We developed a statistical model for Korean prosodic phrase boundaries based on the proposed new features. The results were 86.63% accuracy for three levels (major break, minor break, no break) and 81.14% accuracy for six levels (major break with falling tone/rising tone, minor break with falling tone/rising tone/middle tone, no break).

• Journal of KIISE:Software and Applications
• /
• v.35 no.5
• /
• pp.306-312
• /
• 2008

In a goal-oriented dialogue, speaker's intention can be approximated by a semantic structure that consists of a pair of a speech act and a concept sequence. Therefore, it is very important to correctly identify the semantic structure of an utterance for implementing an intelligent dialogue system. In this paper, we propose a model to efficiently analyze the semantic structures based on an active teaming method. To reduce the burdens of high-level linguistic analysis, the proposed model only uses morphological features and previous semantic structures as input features. To improve the precisions of semantic structure analysis, the proposed model adopts CRFs(Conditional Random Fields), which show high performances in natural language processing, as an underlying statistical model. In the experiments in a schedule arrangement domain, we found that the proposed model shows similar performances(92.4% in speech act analysis and 89.8% in concept sequence analysis) to the previous models although it uses about a third of training data.

• The KIPS Transactions:PartB
• /
• v.18B no.1
• /
• pp.45-50
• /
• 2011

Unknown morpheme errors in Korean morphological analysis are divided into two types: The one is the errors that a morphological analyzer entirely fails to return any morpheme sequences, and the other is the errors that a morphological analyzer returns incorrect combinations of known morphemes. Most previous unknown morpheme estimation techniques have been focused on only the former errors. This paper proposes a unknown morpheme estimation method which can handle both of the unknown morpheme errors. The proposed method detects Eojeols (Korean spacing units) that may include unknown morpheme errors using SVM (Support Vector Machine). Then, using CRFs (Conditional Random Fields), it segments morphemes from the detected Eojeols and annotates the segmented morphemes with new POS tags. In the experiments, the proposed method outperformed the conventional method based on the longest matching of functional words. Based on the experimental results, we knew that the second type errors should be dealt with in order to increase the performance of Korean morphological analysis.

• Proceedings of the Korean Society for Language and Information Conference
• /
• 2007.11a
• /
• pp.303-310
• /
• 2007

Named Entity Recognition (NER) is always limited by its lower recall resulting from the asymmetric data distribution where the NONE class dominates the entity classes. This paper presents an approach that exploits non-local information to improve the NER recall. Several kinds of non-local features encoding entity token occurrence, entity boundary and entity class are explored under Conditional Random Fields (CRFs) framework. Experiments on SIGHAN 2006 MSRA (CityU) corpus indicate that non-local features can effectively enhance the recall of the state-of-the-art NER systems. Incorporating the non-local features into the NER systems using local features alone, our best system achieves a 23.56% (25.26%) relative error reduction on the recall and 17.10% (11.36%) relative error reduction on the F1 score; the improved F1 score 89.38% (90.09%) is significantly superior to the best NER system with F1 of 86.51% (89.03%) participated in the closed track.