• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.16 no.10
• /
• pp.3230-3255
• /
• 2022

Causality mining in NLP is a significant area of interest, which benefits in many daily life applications, including decision making, business risk management, question answering, future event prediction, scenario generation, and information retrieval. Mining those causalities was a challenging and open problem for the prior non-statistical and statistical techniques using web sources that required hand-crafted linguistics patterns for feature engineering, which were subject to domain knowledge and required much human effort. Those studies overlooked implicit, ambiguous, and heterogeneous causality and focused on explicit causality mining. In contrast to statistical and non-statistical approaches, we present Bidirectional Encoder Representations from Transformers (BERT) integrated with Multi-level Feature Networks (MFN) for causality recognition, called BERT+MFN for causality recognition in noisy and informal web datasets without human-designed features. In our model, MFN consists of a three-column knowledge-oriented network (TC-KN), bi-LSTM, and Relation Network (RN) that mine causality information at the segment level. BERT captures semantic features at the word level. We perform experiments on Alternative Lexicalization (AltLexes) datasets. The experimental outcomes show that our model outperforms baseline causality and text mining techniques.

• Proceedings of the Korean Society of Near Infrared Spectroscopy Conference
• /
• 2001.06a
• /
• pp.3112-3112
• /
• 2001

Near-Infrared Spectroscopy (NIRS) was investigated aiming at the identification of falsified drugs. The identification is based on comparison of the NIR spectrum of a sample with a typical spectra of an authentic drug using multivariate modelling and classification algorithms (PCA/SIMCA). Two spectrophotometers (Brimrose - Luminar 2000 and 2030), based on acoustic-optical filter (AOTF) technology, sharing the same controlling computer, software (Brimrose - Snap 2.03) and the data acquisition electronics, were employed. The Luminar 2000 scans the range 850 1800 nm and was employed for transmitance/absorbance measurements of liquids with a transflectance optical bundle probe with total optical path of 5 mm and a circular area of 0.5 $\textrm{cm}^2$. Model 2030 scans the rage 1100 2400 nm and was employed for reflectance measurement of solids drugs. 300 spectra, acquired in about 20 s, were averaged for each sample. Chemometric treatment of the spectral data, modelling and classification were performed by using the Unscrambler 7.5 software (CAMO Norway). This package provides the Principal Component Analysis (PCA) and SIMCA algorithms, used for modelling and classification, respectively. Initially, NIRS was evaluated for spectrum acquisition of various drugs, selected in order to accomplish the diversity of physico-chemical characteristics found among commercial products. Parameters which could affect the spectra of a given drug (especially if presented as solid tablets) were investigated and the results showed that the first derivative can minimize spectral changes associated with tablet geometry, physical differences in their faces and position in relation to the probe beam. The effect of ambient humidity and temperature were also investigated. The first factor needs to be controlled for model construction because the ambient humidity can cause spectral alterations that should cause the wrong classification of a real drug if the factor is not considered by the model.

• Journal of the Korean Society of Safety
• /
• v.35 no.4
• /
• pp.64-73
• /
• 2020

As Korean government and safety-related organizations make continuous efforts to reduce the number of industrial accidents, accident rate has steadily declined since 2010, thereby recording 0.48% in 2017. However, the number of fatalities due to industrial accidents was 1,987 in 2017, which means that more efforts should be made to reduce the number of industrial accidents. As an essential activity for enhancing the system safety, accident analysis can be effectively used for reducing the number of industrial accidents. Accident analysis aims to understand the process of an accident scenario and to identify the plausible causes of the accident. Accident analysis offers useful information for developing measures for preventing the recurrence of an accident or its similar accidents. However, it seems that the current practice of accident analysis in Korean manufacturing companies takes a simplistic accident model, which is based on a linear and deterministic cause-effect relation. Considering the actual complexities underlying accidents, this would be problematic; it could be more significant in the case of human error-related accidents. Accordingly, it is necessary to use a more elaborated accident model for addressing the complexity and nature of human-error related accidents more systematically. Regarding this, HFACS(Human Factors Analysis and Classification System) can be a viable accident analysis method. It is based on the Swiss cheese model and offers a range of causal factors of a human error-related accident, some of which can be judged as the plausible causes of an accident. HFACS has been widely used in several work domains(e.g. aviation and rail industry) and can be effectively used in Korean industries. However, as HFACS was originally developed in aviation industry, the taxonomy of causal factors may not be easily applied to accidents in Korean industries, particularly manufacturing companies. In addition, the typical characteristics of Korean industries need to be reflected as well. With this issue in mind, we developed HFACS-K as a method for analyzing accidents happening in Korean industries. This paper reports the process of developing HFACS-K, the structure and contents of HFACS-K, and a case study for demonstrating its usefulness.

• Proceedings of the KSRS Conference
• /
• 2008.10a
• /
• pp.348-357
• /
• 2008

The classification of the effect of ionospheric disturbances on the radio occultation signal amplitude has been introduced based on an analysis of more than 2000 seances of radio occultation measurements per formed with the help of the CHAMP German satellite. The dependence of the histograms of variations in the radio occultation signal amplitude on the IMF variation index has been revealed. It has been indicated that it is possible to introduce the radio occultation index characterizing the relation between ionospheric disturbances and solar activity. An amplitude radio occultation (RO) method is proposed to study connection between the ionospheric and solar activity on a global scale. Sporadic amplitude scintillation observed in RO experiments contain important information concerning the seasonal, geographical, and temporal distributions of the ionospheric disturbances and depend on solar activity. The probability of strong RO amplitude variations (RO $S_4$ index greater than 0.2) in the CHAMP RO signals diminishes sharply with the weakening of solar activity from 2001 to 2008. The general number of RO events with strong amplitude variations can be used as an indicator of the ionospheric activity. We found that during 2001-2008 the daily globally averaged RO $S_{4a}$ index depends essentially on solar activity. The maximum occurred in January 2002, minimum has been observed in summer 2008. Different temporal behavoir of $S_{4a}$ index has been detected for polar (with latitude greater than $60^{\circ}$) and low latitude (moderate and equatorial) regions. For polar regions $S_{4a}$ index is slowly decreasing with solar activity. In the low latitude areas $S_{4a}$ index is sharply oscillating, depending on the solar ultraviolet emission variations. The different geographical behavoir of $S_{4a}$ index indicates different origin of ionospheric plasma disturbances in polar and low latitude areas. Origin of the plasma disturbances in the polar areas may be connected with influence of solar wind, the ultraviolet emission of the Sun may be the main cause of the ionospheric irregularities in the low latitude zone. Therefore, the $S_{4a}$ index of RO signal is important radio physical indicator of solar activity.

• Journal of the Korean Chemical Society
• /
• v.53 no.5
• /
• pp.570-584
• /
• 2009

The purpose of this study was to find out high school students' perception on psychologicallearning environment generated by science teachers and their attitude change related to science. The subjectsconsisted of 539 freshmen in a boys' high school pre-applied of common school group in S city. This study wasconducted with students' perception survey and classification of teachers' features according to it. The surveyabout science-related attitude was also made in early 1st semester and 2nd semester, and the students showingthe great attitude change related to science were interviewed. The results of this study revealed that statistically,students had a more positive perception on female teachers than on male ones and that according to their teachers,there were clear different in the psychological learning environment perceived by students. As for the relation of teachers' features and students' attitude change, it showed the negative effect only when the teacher was incharge of only one class, but in most of the cases, there was no meaningful correlation. The semi-structuredinterview with students with great attitude change related to science indicated that the main cause of the changewas the achievement they made in class. The interview showed that the change related to science happenedunder the indirect influence of teachers rather than direct influence. Furthermore, students wanted scienceteachers to meet the science class possessing various instruction behaviors and support behaviors. Therefore,science teachers playing an important role in students' choice of career should make efforts to realize thelearner-centered curriculum and change students' science-related attitude into a positive direction.