• Journal of the Institute of Electronics and Information Engineers
• /
• v.51 no.8
• /
• pp.66-74
• /
• 2014

Recently a variety of study of Korean parsing system is carried out by many software engineers and linguists. The parsing system mainly uses the method of machine learning or symbol processing paradigm. But the parsing system using machine learning has long training time because the data of Korean sentence is very big. And the system shows the limited recognition rate because the data has self error. In this thesis we design system using feature module which can reduce training time and analyze the recognized rate each the number of training sentences and repetition times. The designed system uses the separated modules and sorted table for binary search. We use the refined 36,090 sentences which is extracted by Sejong Corpus. The training time is decreased about three hours and the comparison of recognized rate is the highest as 84.54% when 10,000 sentences is trained 50 times. When all training sentence(32,481) is trained 10 times, the recognition rate is 82.99%. As a result it is more efficient that the system is used the refined data and is repeated the training until it became the steady state.

• Proceedings of the Acoustical Society of Korea Conference
• /
• spring
• /
• pp.337-340
• /
• 2004

The exactly reproduced prosody of a TTS system is one of the key factors that affect the naturalness of synthesized speech. In general, rules about prosody had been gathered either from linguistic knowledge or by analyzing the prosodic information from natural speech. But these could not be perfect and some of them could be incorrect. So we proposed artificial neural network(ANN)s that can be trained to team the prosody of natural speech and generate it. In learning phase, let ANNs learn the pitch and energy contour of center phoneme by applying a string of phonemes in a sentence to ANNs and comparing the output pattern with target pattern and making adjustment in weighting values to get the least mean square error between them. In test phase, the estimation rates were computed. We saw that ANNs could generate the prosody of a sentence.

• Journal of KIISE
• /
• v.44 no.1
• /
• pp.57-62
• /
• 2017

Traditional Korean morphological analysis and POS tagging methods usually consist of two steps: 1 Generat hypotheses of all possible combinations of morphemes for given input, 2 Perform POS tagging search optimal result. require additional resource dictionaries and step could error to the step. In this paper, we tried to solve this problem end-to-end fashion using sequence-to-sequence model convolutional features. Experiment results Sejong corpus sour approach achieved 97.15% F1-score on morpheme level, 95.33% and 60.62% precision on word and sentence level, respectively; s96.91% F1-score on morpheme level, 95.40% and 60.62% precision on word and sentence level, respectively.

• Korean Journal of Child Studies
• /
• v.6 no.1
• /
• pp.17-40
• /
• 1985

This study investigated Korean children's early acquisition of negatives and focused on four research questions: 1) processing of negative variations; 2) the nature of negatives when negatives are completely acquired in Korean (in which meaning and form are matched in one to one mapping); 3) the validity of Bellugi's negative acquisition model in Korean; and 4) the cause of child's erroneous sentence production: limited ability or regularity in children's cognition. The language data of the five subjects (age span; 1.1 - 3.11) were collected by their parents in the natural setting of the home. The results showed that 1) the pivot form, was processed in many ways from a simple to a complicated form, such as <(X+X')+N> <(x+x')+N,Y> <(x+x') N,(y+y')>. It appeared that the children used a simple negative format to reach a one-step advanced negative format. 2) Korean negatives are divided into range of negation in the negative sentence (part or whole), strength of negation (absolute or general), functions of meaning (negation, absences, refusal, prohibition, impossibility). All five children acquired negative sentences in all functions and the complete range after 3 years of age. 3) In spite of the differences in age level, Bellugi's four stage model was in evidence; that is, Korean children's negative acquisition was almost identical with Bellugi's tour stage model in deep structure. 4) Analyses of children's error sentences showed that the sentences with errors were made not because of the children's limitation in cognitive ability but because of the strict application of regularity of rules from the original grammars. Consequently, the children produced negative sentences using two rules: the rule of additive complexity (from simple to complex) and the rule of division (from one to several).

• The KIPS Transactions:PartB
• /
• v.18B no.4
• /
• pp.241-248
• /
• 2011

Homographs can have multiple senses. In order to understand the meaning of a sentence, it is necessary to identify which sense isused for each word in the sentence. Previous researches on this problem heavily relied on the word co-occurrence information. However, we noticed that in case of verbs, information about subordinating cases of verbs can be utilized to further improve the performance of word sense disambiguation. Different senses require different sets of subordinating cases. In this paper, we propose the verb sense disambiguation using subordinating case information. The case information acquire postposition features in Standard Korean Dictionary. Our experiment on 12 high-frequency verb homographs shows that adding case information can improve the performance of word sense disambiguation by 1.34%, from 97.3% to 98.7%. The amount of improvement may seem marginal, we think it is meaningful because the error ratio reduced to less than a half, from 2.7% to 1.3%.

• School Mathematics
• /
• v.6 no.3
• /
• pp.235-249
• /
• 2004

The purpose of this study was to analyze the error patterns and sentence types in word problems with respect to 1$\frac{3}{4}$$\div$$\frac{1}{2}$ which were made by the pre-service elementary teachers, and to suggest the clues to the education in pre-service. Korean elementary teachers in pre-service misunderstood 'divide with $\frac{1}{2}$' to 'divide to 2' by the Korean linguistic structure. And they showed a new error type of 1$\frac{3}{4}$$\times$2 by the result of calculation. Although they are familiar to 'inclusive algorithm' they are not good at dealing with the fractional divisor. And they are very poor at the 'decision the unit proportion' and the 'inverse of multiplication'. So, it is necessary to teach the meaning of the fractional division as 'decision the unit proportion' and 'inverse of multiplication' and to give several examples with respect to the actual situation and context.

• Journal of the Korea Society of Computer and Information
• /
• v.18 no.10
• /
• pp.173-182
• /
• 2013

Typographical mistakes made in the writing process of drafts of electronic documents are more common than any other type of errors. The majority of these errors caused by mistyping are regarded as consequently still typo-errors, but a considerable number of them are developed into the grammatical errors and the semantic errors. Pseudo semantic errors among these errors due to typographical errors have more noticeable peculiarities than pure semantic errors between senses of surrounding context words within a sentence. These semantic errors can be detected and corrected by simple algorithm based on the co-occurrence frequency because of their prominent contextual discrepancy. I propose a method for detection and correction based on the co-occurrence frequency in order to detect semantic errors due to typo-errors. The co-occurrence frequency in proposed method is counted for only words with immediate dependency relation, and the cosine similarity measure is used in order to detect pseudo semantic errors. From the presented experimental results, the proposed method is expected to help improve the detecting rate of overall proofreading system by about 2~3%.

• The KIPS Transactions:PartB
• /
• v.18B no.6
• /
• pp.405-414
• /
• 2011

Korean grammar checkers typically detect context-dependent errors by employing heuristic rules that are manually formulated by a language expert. These rules are appended each time a new error pattern is detected. However, such grammar checkers are not consistent. In order to resolve this shortcoming, we propose new method for generalizing error decision rules to detect the above errors. For this purpose, we use an existing thesaurus KorLex, which is the Korean version of Princeton WordNet. KorLex has hierarchical word senses for nouns, but does not contain any information about the relationships between cases in a sentence. Through the Tree Cut Model and the MDL(minimum description length) model based on information theory, we extract noun classes from KorLex and generalize error decision rules from these noun classes. In order to verify the accuracy of the new method in an experiment, we extracted nouns used as an object of the four predicates usually confused from a large corpus, and subsequently extracted noun classes from these nouns. We found that the number of error decision rules generalized from these noun classes has decreased to about 64.8%. In conclusion, the precision of our grammar checker exceeds that of conventional ones by 6.2%.

• Journal of rehabilitation welfare engineering & assistive technology
• /
• v.2 no.1
• /
• pp.35-43
• /
• 2009

In this paper, we introduce a reading training program offering visual-auditory cue with noise cancellation function (RT program) developed by us. The RT program provides some training sentences with visual-auditory cues. Motor speech disorder patients can use the visual and/or auditory cues for reading training. To provide convenient estimation of training result, we developed a noise cancellation algorithm. The function of the algorithm is to remove noise and auditory-cues which are recorded with reading speech at the same time while patient read the sentences in PC monitor. In addition, we developed a function for finding out the first starting time of reading sound after a patient sees a sentence and begins to read the sentence. The recorded speeches are acquired from six people(three male, three female) in four noisy environments (interior noise, white noise, car interior noise, babble noise). We evaluated the timing error for starting time between original recorded speech and processed speech in condition of executing noise cancellation function and not executing. The timing error was improved as much as $4.847{\pm}2.4235[ms]$ as the effect of noise cancellation. It is expected that the developed RT program helps motor speech disorder patient in reading training and symptom evaluation.

• Smart Media Journal
• /
• v.6 no.1
• /
• pp.16-21
• /
• 2017

Long time has passed since computers which used to be a means of research were commercialized and available for the general public. People used writing instruments to write before computer was commercialized. However, today a growing number of them are using computers to write instead. Computerized word processing helps write faster and reduces fatigue of hands than writing instruments, making it better fit to making long texts. However, word processing programs are more likely to cause spelling errors by the mistake of users. Spelling errors distort the shape of words, making it easy for the writer to find and correct directly, but those caused due to users' lack of knowledge or those hard to find may make it almost impossible to produce a document free of spelling errors. However, spelling errors in important documents such as theses or business proposals may lead to falling reliability. Consequently, it is necessary to conduct research on high-level spelling error correction programs for the general public. This study was designed to produce a system to correct sentence-level spelling errors to normal words with Korean alphabet similarity algorithm. On the basis of findings reported in related literatures that corrected words are significantly similar to misspelled words in form, spelling errors were extracted from a corpus. Extracted corrected words were replaced with misspelled ones to correct spelling errors with spelling error detection algorithm.