• Journal of Intelligence and Information Systems
• /
• v.24 no.2
• /
• pp.59-83
• /
• 2018

With the increasing importance of sentiment analysis to grasp the needs of customers and the public, various types of deep learning models have been actively applied to English texts. In the sentiment analysis of English texts by deep learning, natural language sentences included in training and test datasets are usually converted into sequences of word vectors before being entered into the deep learning models. In this case, word vectors generally refer to vector representations of words obtained through splitting a sentence by space characters. There are several ways to derive word vectors, one of which is Word2Vec used for producing the 300 dimensional Google word vectors from about 100 billion words of Google News data. They have been widely used in the studies of sentiment analysis of reviews from various fields such as restaurants, movies, laptops, cameras, etc. Unlike English, morpheme plays an essential role in sentiment analysis and sentence structure analysis in Korean, which is a typical agglutinative language with developed postpositions and endings. A morpheme can be defined as the smallest meaningful unit of a language, and a word consists of one or more morphemes. For example, for a word '예쁘고', the morphemes are '예쁘(= adjective)' and '고(=connective ending)'. Reflecting the significance of Korean morphemes, it seems reasonable to adopt the morphemes as a basic unit in Korean sentiment analysis. Therefore, in this study, we use 'morpheme vector' as an input to a deep learning model rather than 'word vector' which is mainly used in English text. The morpheme vector refers to a vector representation for the morpheme and can be derived by applying an existent word vector derivation mechanism to the sentences divided into constituent morphemes. By the way, here come some questions as follows. What is the desirable range of POS(Part-Of-Speech) tags when deriving morpheme vectors for improving the classification accuracy of a deep learning model? Is it proper to apply a typical word vector model which primarily relies on the form of words to Korean with a high homonym ratio? Will the text preprocessing such as correcting spelling or spacing errors affect the classification accuracy, especially when drawing morpheme vectors from Korean product reviews with a lot of grammatical mistakes and variations? We seek to find empirical answers to these fundamental issues, which may be encountered first when applying various deep learning models to Korean texts. As a starting point, we summarized these issues as three central research questions as follows. First, which is better effective, to use morpheme vectors from grammatically correct texts of other domain than the analysis target, or to use morpheme vectors from considerably ungrammatical texts of the same domain, as the initial input of a deep learning model? Second, what is an appropriate morpheme vector derivation method for Korean regarding the range of POS tags, homonym, text preprocessing, minimum frequency? Third, can we get a satisfactory level of classification accuracy when applying deep learning to Korean sentiment analysis? As an approach to these research questions, we generate various types of morpheme vectors reflecting the research questions and then compare the classification accuracy through a non-static CNN(Convolutional Neural Network) model taking in the morpheme vectors. As for training and test datasets, Naver Shopping's 17,260 cosmetics product reviews are used. To derive morpheme vectors, we use data from the same domain as the target one and data from other domain; Naver shopping's about 2 million cosmetics product reviews and 520,000 Naver News data arguably corresponding to Google's News data. The six primary sets of morpheme vectors constructed in this study differ in terms of the following three criteria. First, they come from two types of data source; Naver news of high grammatical correctness and Naver shopping's cosmetics product reviews of low grammatical correctness. Second, they are distinguished in the degree of data preprocessing, namely, only splitting sentences or up to additional spelling and spacing corrections after sentence separation. Third, they vary concerning the form of input fed into a word vector model; whether the morphemes themselves are entered into a word vector model or with their POS tags attached. The morpheme vectors further vary depending on the consideration range of POS tags, the minimum frequency of morphemes included, and the random initialization range. All morpheme vectors are derived through CBOW(Continuous Bag-Of-Words) model with the context window 5 and the vector dimension 300. It seems that utilizing the same domain text even with a lower degree of grammatical correctness, performing spelling and spacing corrections as well as sentence splitting, and incorporating morphemes of any POS tags including incomprehensible category lead to the better classification accuracy. The POS tag attachment, which is devised for the high proportion of homonyms in Korean, and the minimum frequency standard for the morpheme to be included seem not to have any definite influence on the classification accuracy.

• The korean journal of orthodontics
• /
• v.31 no.2 s.85
• /
• pp.173-185
• /
• 2001

The orthodontic osseointegrated titanium implant, a kind of intraoral skeletal anchorage can be an alternative to tooth-borne anchorage, in case that the conventional tooth-borne anchorage is not available or the anchorage is critical. This study was conducted to elucidate the effect of early loading on the osseointegration of the orthodontic titanium implant and the healing process of the impaired bone at the site of implant after removing it. In two adult beagle dogs24 osseointegrated titanium implants were inserted into the alveolar bone, with 12 implants placed in each dog. In dog1, 6 out of 12 implants were loaded with 200-300gm of force immediately after placing, and the remaining 6 implants were not loaded for 4weeks. In dog2, all 12 implants had healing period of 4weeks, and then were loaded with 200-300gm of force for another 4weeks. Following an observation period of 4 and 8 weeks, the animals were sacrificed. Then the implants and the surrounding bone of dog1 and dog2 were removed, respectively. Undecalcified sections along the long axis of implant were made and the degree of osseointegration was examined under the light microscope. The results were as follows. 1. In the histologic features of tissues around implants anchored in dog1, there was no difference between immediately loaded implants and unloaded implants. Immature woven bone was ingrowing into the thread spaces from the original compacta and in direct contact with the implant surface in part. 2. The premature loading just after 4weeks healing period did not halt the progress of the osseointegration between bone and implant surface. The woven bone around the implants was maturing into the lamellar bone which resembled the structure of the original compacta at the end of 8weeks observation period. 3. Most implants with the inflammed surrounding mucosa were lost or mobile. The mobile implants were encapsulated by fibrous connective tissue which separated the implant surface from the bone. 4. The impaired bone at the site of the implant failed to anchor was showing recovery without inflammatory reaction 2weeks after removing, with the immaure woven bone lined by active osteoblasts and osteoid. Based on the results of this study, the integration of this orthodontic implant seemed to be impaired by the inflammation of the tissue surrounding the Implant rather than by early loading on implant, and increased with time lapsed after placing the implant. The use of implant described in this report can be recommended as an orthodontic anchorage unit immediately after insertion under the careful control of orthodontic force applied and plaque.

• Tuberculosis and Respiratory Diseases
• /
• v.53 no.6
• /
• pp.619-634
• /
• 2002

Background : Many inflammatory mediators and collagenases are involved in the pathogenesis of acute lung injury (ALI) and acute respiratory distress syndrome (ARDS). The increase of matrix metalloproteinase-9 (MMP-9, gelatinase-B) produced mainly by inflammatory cells was reported in many ALI models and connective tissue cells. In this study, the expression of MMP-9 in ventilator-induced lung injury (VILI) model and the effects of matrix metalloproteinase inhibitor (MMPI) on VILI were investigated. Methods : Eighteen Sprague-Dawley rats were divided into three groups: low tidal Volume (LVT, 7mL/Kg tidal volume, 3 $cmH_2O$ PEEP, 40/min), high tidal volume (HVT, 30mL/Kg tidal volume, no PEEP, 40/min) and high tidal volume with MMPI (HVT+MMPI) groups. Mechanical ventilation was performed in room air for 2 hours. The 20 mg/Kg of CMT-3 (chemically modified tetracycline-3, 6-demethyl 6-deoxy 4-dedimethylamino tetracycline) was gavaged as MMPI from three days before mechanical ventilation. The degree of lung injury was measured with wet-to-dry weight ratio and acute lung injury score. Expression of MMP-9 was studied by immunohistochemical stain with a mouse monoclonal anti-rat MMP-9 $IgG_1$. Results : In the LVT, HVT and HVT+MMPI groups, the wet-to-dry weight ratio was $4.70{\pm}0.14$, $6.82{\pm}1.28$ and $4.92{\pm}0.98$, respectively. In the HVT group, the ratio was significantly higher than other groups (p<0.05). Acute lung injury score measured by five-point scale was $3.25{\pm}1.17$, $12.83{\pm}1.17$ and $4.67{\pm}0.52$, respectively. The HVT group was significantly damaged by VILI and MMPI protects injuries by mechanical ventilation (p<0.05). Expression of MMP-9 measured by four-point scale was $3.33{\pm}2.07$, $12.17{\pm}2.79$ and $3.60{\pm}1.95$, respectively, which were significantly higher in the HVT group (p<0.05). Conclusion : VILI increases significantly the expression of MMP-9 and MMPI prevents lung injury induced by mechanical ventilation through the inhibition of MMP-9.

• Journal of Life Science
• /
• v.25 no.5
• /
• pp.523-532
• /
• 2015

A variety of previous pharmacological studies have suggested Liriope platyphylla (L. platyphylla) may exert beneficial biological effects on inflammation, diabetes, neurodegenerative disorder, obesity, constipation, and atopic dermatitis. In addition, hydrocolloid membranes (HCMs) have attracted attention in dermatological care, including in the treatment of scleroderma skin ulcers, cutaneous ulcers, permanent tympanic membrane perforations, pressure sores, and decubitus ulcers in the elderly. To investigate the therapeutic effects of HCM containing an aqueous extract of L. platyphylla (HCM-LP) on second-degree burn wounds, their physico-chemical properties were analyzed and the therapeutic effects were observed in SD rats after treatment with HCM-LP for 14 days. Significant declines in tensile strength (38.4%) and absorptiveness (46.3%), as well as an increase in surface roughness (38.1%) were detected in HCM-LP compared with that of HCM. In SD rats with burned skin, the wound diameter was shorter in the HCM-LP treated group than in the GZ group on post-surgical day 14, while the significant improvements in scar tissue reduction, epithelium regeneration, angiogenesis, and extracellular matrix deposition were observed in the HCM-LP-treated group during all experimental periods. Overall, these results suggest HCM-LP may accelerate the process of healing the burn injury skin of SD rats through the regulation of angiogenesis and connective tissue formation.

• Clinical and Experimental Reproductive Medicine
• /
• v.27 no.1
• /
• pp.31-37
• /
• 2000

This study was designed to identify the ultrastructural changes of mouse endometrium during peri-implantation period and obtain the fundamental information for the establishment of 3-dimensional culture system of mouse endometrial cells in vitro. The used female ICR mice ($6{\sim}8$ wks) were conducted on pregnant. The biopsies were obtained from whole uterus at cycle day 1 (D1) and day 5 (D5) after hCG injection and mating. The biopsies materials were fixed 2.5% glutaraldehyde and 1% osmium tetroxide. Subsequently, for observation using light and transmission electron microscopy (LM and TEM), they were dehydrated and embedded in Epon and the embedded biopsies were sectioned and stained. For scanning electron microscopy (SEM), the fixed specimens were dehydrated, dried and coated with gold. 1) For LM, the biopsied materials at D5 (late secretory phase) were appeared the extended stromal layer by increased connective tissues and the fully developed endometrial glands and vessels compared with D1 (early secretory phase). 2) For TEM, the mouse endometrium was consisted of 3-layers, a simple polarized columnar epithelial cells, basement membrane and stromal cells. At D5, the distribution of microvilli, endoplasmic reticulum, Golgi body, lipid and glycogen deposits, secretory granules and surface area of basement membrane were increased. 3) For SEM, the degree of folding and microvilli of surface of mouse epithelial cells was became more and more according to the process of secretory phase, and at D5, implantation time of mouse, the appearance of pinopodes as a specific marker of uterine receptivity was found. The uterine pinopodes of mouse were found in narrow sites at the luminal surface, irregularity and appeared the different stages in the same sample. Therefore, these results indicated that the mouse endometrium was experienced dramatic morphological changes during peri-implantation period.