• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.26 no.5
• /
• pp.662-667
• /
• 2022

Semantic role labeling task(SRL) is to extract predicate and arguments such as agent, patient, place, time. In the previously SRL task studies, a pipeline method extracting linguistic features of sentence has been proposed, but in this method, errors of each extraction work in the pipeline affect semantic role labeling performance. Therefore, methods using End-to-End neural network model have recently been proposed. In this paper, we propose a neural network model using the Biaffine Average Attention model for SRL task. The proposed model consists of a structure that can focus on the entire sentence information regardless of the distance between the predicate in the sentence and the arguments, instead of LSTM model that uses the surrounding information for prediction of a specific token proposed in the previous studies. For evaluation, we used F1 scores to compare two models based BERT model that proposed in existing studies using F1 scores, and found that 76.21% performance was higher than comparison models.

• Korean Journal of Cognitive Science
• /
• v.3 no.2
• /
• pp.161-174
• /
• 1992

The present discussion is concerned with showing the development of natural language processing and how it is related to information and cognition.On the basis of the computeational model,in which humans are viewed as processors of linguistic structures that use stored knowledge-grammar, lexicon and structures representing the encyclopedic information of the world,such programs of natural language understanding as Winograd's SHRDLU came out.However,such pragmatic factors as contexts and the speaker's beliefs,internts,goals and intentions are not easy to process yet.Language,ingormation and cognition are argued to be closely interrelated,and the study of them,the paper argues,can lead to the development of science on general.

• Proceedings of the IEEK Conference
• /
• 2000.06d
• /
• pp.222-225
• /
• 2000

Human frequently communicate non-linguistic information with gesture. So, we must develop efficient and fast gesture recognition algorithms for more natural human-computer interaction. However, it is difficult to recognize gesture automatically because human's body is three dimensional object with very complex structure. In this paper, we suggest a method which is able to detect key frames and frame changes, and to classify image sequence into some gesture groups. Gesture is classifiable according to moving part of body. First, we detect some frames that motion areas are changed abruptly and save those frames as key frames, and then use the frames to classify sequences. We symbolize each image of classified sequence using Principal Component Analysis(PCA) and clustering algorithm since it is better to use fewer components for representation of gestures. Symbols are used as the input symbols for the Hidden Markov Model(HMM) and recognized as a gesture with probability calculation.

• Journal of the Korean BIBLIA Society for library and Information Science
• /
• v.8 no.1
• /
• pp.117-134
• /
• 1997

This study proves that the quoted legal text functions as the index which represents the contents of the text because of the characteristics of legal information, the automatic indexing in the secondary legal full-text databases can be possible without the assitance of the experts. In case of the establishment, amendment or repealing of law, change of words of index can be possible through revising the legal text quoted in the secondary legal full-text databases. Even when we dont input the full-text about retrospective documents, automatic indexing is also possible, and the establihment and the practice of expert knowledge and integrated databases are possible in case of the retrospective documents. This study indicates that it is necessary to have characteristic information the information experts recognize - that is to say, experimental and inherent knowledge only human being can have - built-in into the system rather than to approach the information system by the linguistic, statistic or structuralistic way, and it can be more essential and intelligent information system.

• Proceedings of the KSR Conference
• /
• 2004.06a
• /
• pp.1155-1162
• /
• 2004

This research proposes a new risk analysis method in order to guarantee successful performance of construction projects. The proposed risk analysis methods consists of four phases. First step, AHP model can help contractors decide whether or not they bid for a project by analysing risks involved in the project. Second step, the influence diagraming, decision tree and Monte Carlo simulation are used as tools to analyze and evaluate project risks quantitatively. Third step, Monte Carlo simulation is used to assess risk for groups of activities with probabilistic branching and calendars. Finally, Fuzzy theory suggests a risk management method for construction projects, which is using subjective knowledge of an expert and linguistic value, to analyze and quantify risk. The result of study is expected to improve the accuracy of risk analysis because three factors, such as probability, impact and exposure, for estimating membership function are introduced to quantify each risk factor. Consequently, it will help contractors identify risk elements in their projects and quantify the impact of risk on project time and cost.

• 제어로봇시스템학회:학술대회논문집
• /
• 1993.10a
• /
• pp.737-741
• /
• 1993

In this paper, we design the fuzzy logic controller(FLC) for a nonlinear multivariable steam generating unit. Based on the knowledges of operator, the self-organizing controller(SOC) - a kind of FLC - is developed and tested. Both FLC and SOC based on linguistic rules have the advantages of not needing of some exact mathematical model for plant to be controlled. Beside, the SOC modifies the existing control rules by monitoring the control performance. The computer simulations have been carried out for the 200MW steam generating unit to show the usefulness of the proposed method and the effects of disturbances and parameter variations are considered.

• Korean Journal of English Language and Linguistics
• /
• v.4 no.1
• /
• pp.83-102
• /
• 2004

The purpose of this paper is to find out some major characteristics of intermediate-advanced Korean learners' English by corpus- linguistically analyzing their essays in comparison with native speakers'. We construct a corpus of CBT TOEFL essays by Korean learners, NNS1 (94076 words in 402 texts), and its sub-corpus, NNS2 (14291 words in 45 texts), and then a corpus of model essays written or meticulously edited by native speakers, NS (14833 words in 35 texts). We compare NNS1 and NNS2 with NS, and with some other corpora, in terms of high-frequency words, and show that Korean learners' writings have more features of informal writing than those of formal writing, which is in accord with the reports in Granger (1998) that EFL writings by European advanced learners are characterized by informality.

• Phonetics and Speech Sciences
• /
• v.7 no.4
• /
• pp.49-55
• /
• 2015

It is argued in this paper that, in determining semantic similarity, Korean words should be recategorized with a focus on the semantic relation to ontology in light of cross-linguistic morphological variations. It is proposed, in particular, that Korean semantic similarity should be measured on three tracks, human judgements track, relatedness track, and cross-part-of-speech relations track. As demonstrated in Yang et al. (2015), GloVe, the unsupervised learning machine on semantic similarity, is applicable to Korean with its performance being compared with human judgement results. Based on this compatability, it was further thought that the model's performance might most likely vary with different kinds of specific relations in different languages. An attempt was made to analyze them in terms of two major Korean-specific categories involved in their lexical and cross-POS-relations. It is concluded that languages must be analyzed by varying methods so that semantic components across languages may allow varying semantic distance in the vector space models.

• 제어로봇시스템학회:학술대회논문집
• /
• 1992.10a
• /
• pp.382-386
• /
• 1992

In this paper, the author proposed FLVSC(Fuzzy Logic Variable Structure Controller), of which control rules are extracted from the concepts of VSC(Variable Structure Control). FLC(Fuzzy Logic Controller) based on linguistic rules has the advantages of not needing of some exact mathematical model for plant to be controlled. The proposed method has the characteristics which are viewed in conventional VSC, e.g. insensitivity to a class of disturbances, parameter variations and uncertainties in sliding mode. In addition, the method has the properties of FLC - noise rejection capability etc. The computer simulations have been carried out for a DC servo motor to show the usefulness of the proposed method and the effects of disturbances and parameter variations are considered.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.12 no.1
• /
• pp.248-268
• /
• 2018

The performance of smartphone applications are usually constrained in user interactions due to resource limitation and it promises great opportunities to improve the performance by exploring the smartphone built-in and embedded sensing techniques. However, heterogeneity in techniques, semantic gap between sensor data and usable context, and complexity of contextual situations keep the techniques from seamless integration. Relevant studies mainly focus on feasibility demonstration of emerging sensing techniques, which rarely address both general architectures and comprehensive technical solutions. Based on a proposed functional model, this investigation provides a general architecture to deal with the dynamic context for context-aware automation and decision support. In order to take advantage of the built-in sensors to improve the performance of mobile applications, an ontology-based method is employed for context modelling, linguistic variables are used for heterogeneous context presentation, and semantic distance-based rule matching is employed to customise functions to the contextual situations. A case study on mobile application authentication is conducted with smartphone built-in hardware modules. The results demonstrate the feasibility of the proposed solutions and their effectiveness in improving operational efficiency.