• English Language & Literature Teaching
• /
• v.13 no.1
• /
• pp.235-259
• /
• 2007

The purpose of the present study is to develop a prototype automated scoring system for English writing. The system was developed for scoring writings of Korean middle school students. In order to develop the automated scoring system, following procedures have been applied. First, review and analysis of established automated essay scoring systems in other countries have been accomplished. By doing so, we could get the guidance for development of a new sentence-level automated scoring system for Korean EFL students. Second, knowledge base such as lexicon, grammar and WordNet for natural language processing and error corpus of English writing of Korean middle school students were established. Error corpus was established through the paper and pencil test with 589 third year middle school students. This study provided suggestions for the successful introduction of an automated scoring system in Korea. The automated scoring system developed in this study should be continuously upgraded to improve the accuracy of the scoring system. Also, it is suggested to develop an automated scoring system being able to carry out evaluation of English essay, not only sentence-level evaluation. The system needs to be upgraded for the improved precision, but, it was a successful introduction of an sentence-level automated scoring system for English writing in Korea.

• Journal of Multimedia Information System
• /
• v.5 no.3
• /
• pp.209-214
• /
• 2018

Today is the era of artificial intelligence. With the development of artificial intelligence, machines have begun to impersonate various human characteristics today. Chatbot is one instance of this interactive artificial intelligence. Chatbot is a computer program that enables to conduct natural conversations with people. As mentioned above, Chatbot conducted conversations in text, but Chatbot, in this study evolves to perform commands based on speech-recognition. In order for Chatbot to perfectly emulate a human dialogue, it is necessary to analyze the sentence correctly and extract appropriate response. To accomplish this, the sentence is classified into three types: objects, actions, and preferences. This study shows how objects is analyzed and processed, and also demonstrates the possibility of evolving from an elementary model to an advanced intelligent system. By this study, it will be evaluated that speech-recognition based Chatbot have improved order-processing time efficiency compared to text based Chatbot. Once this study is done, speech-recognition based Chatbot have the potential to automate customer service and reduce human effort.

• Speech Sciences
• /
• v.13 no.4
• /
• pp.187-200
• /
• 2006

The speech synthesis technology is widely used and its application area is also being broadened to an automatic response service, a learning system for handicapped person, etc. However, the sound quality of the speech synthesizer has not yet reached to the satisfactory level of users. To make a synthesized speech, the existing synthesizer generates rhythms only by the interval information such as space and comma or by several punctuation marks such as a question mark and an exclamation mark so that it is not easy to generate natural rhythms of people even though it is based on mass speech database. To make up for the problem, there is a way to select rhythms after processing language from a higher level information. This paper proposes a method for generating tags for controling rhythms by analyzing the meaning of sentence with speech situation information. We use the Systemic Functional Grammar (SFG) [4] which analyzes the meaning of sentence with speech situation information considering the sentence prior to the given one, the situation of a conversation, the relationship among people in the conversation, etc. In this study, we generate Semantic Speech Control Tag (SSCT) by the result of SFG's meaning analysis and the voice wave analysis.

• KIPS Transactions on Software and Data Engineering
• /
• v.4 no.12
• /
• pp.577-584
• /
• 2015

In general, Q&A system retrieves passages by matching terms of a question in order to find an answer to the question. However it is difficult for Q&A system to find a correct answer because too many passages are retrieved and matching using terms is not enough to rank them according to their relevancy to a question. To alleviate this problem, we introduce a topic for a sentence, and adopt it for ranking in Q&A system. We define a set of person-related topic class and a clue expression which can indicate a topic of a sentence. A topic classification system proposed in this paper can determine a target topic for an input sentence by using clue expressions, which are manually collected from a corpus. We explain an architecture of the topic classification system and evaluate the performance of the components of this system.

• The KIPS Transactions:PartB
• /
• v.13B no.2 s.105
• /
• pp.149-154
• /
• 2006

The main purpose of document summarization is to reduce the complexity of documents that are consisted of sub-themes. Also it is to create summarization which includes the sub-themes. This paper proposes a summarization system which could extract any salient sentences in accordance with sub-themes by using graph division. A document can be represented in graphs by using chosen representative terms through term relativity analysis based on co-occurrence information. This graph, then, is subdivided to represent sub-themes through connected information. The divided graphs are types of sentence clustering which shows a close relationship. When salient sentences are extracted from the divided graphs, summarization consisted of core elements of sentences from the sub-themes can be produced. As a result, the summarization quality will be improved.

• Journal of Digital Contents Society
• /
• v.19 no.6
• /
• pp.1161-1167
• /
• 2018

The development of machine learning has enabled machines to perform delicate tasks that only humans could do, and thus many companies have introduced machine learning based translators. Existing translators have good performances but they have problems in number translation. The translators often mistranslate numbers when the input sentence includes a large number. Furthermore, the output sentence structure completely changes even if only one number in the input sentence changes. In this paper, first, we optimized a neural machine translation model architecture that uses bidirectional RNN, LSTM, and the attention mechanism through data cleansing and changing the dictionary size. Then, we implemented a number-processing algorithm specialized in number translation and applied it to the neural machine translation model to solve the problems above. The paper includes the data cleansing method, an optimal dictionary size and the number-processing algorithm, as well as experiment results for translation performance based on the BLEU score.

• Journal of KIISE
• /
• v.43 no.12
• /
• pp.1376-1384
• /
• 2016

Computers require analytic and processing capability for all possibilities of human expression in order to process sentences like human beings. Linguistic information processing thus forms the initial basis. When analyzing a sentence syntactically, it is necessary to divide the sentence into components, find obligatory arguments focusing on predicates, identify the sentence core, and understand semantic relations between the arguments and predicates. In this study, the method applied a case frame dictionary based on The Korean Standard Dictionary of The National Institute of the Korean Language; in addition, we used a CRF Model that constructed subcategorization of predicates as featured in Korean Lexical Semantic Network (UWordMap) for semantic role labeling. Automatically tagged semantic roles based on the CRF model, which established the information of words, predicates, the case-frame dictionary and hypernyms of words as features, were used. This method demonstrated higher performance in comparison with the existing method, with accuracy rate of 83.13% as compared to 81.2%, respectively.

• Korean Journal of Cognitive Science
• /
• v.18 no.2
• /
• pp.91-112
• /
• 2007

Two experiments were conducted to examine psychological reality and incremental nature of thematic processing in Korean sentence comprehension. By using two different types of verbs (i.e., transitive and causative verbs), we manipulated necessity of the thematic reanalysis (i.e., consistent vs. inconsistent condition) in the coordinated sentence structures. In Experiment 1, there was no significant difference in the reading times of the verbs between the consistent and the inconsistent condition. However, there was significant differences in question answering times between the two conditions. In Experiment 2 in which we changed a noun phrase of the test sentences into inanimate one, we found significant thematic reanalysis effects in the reading times of the final verbs. Based on these results we discussed the theoretical importance and universality of the thematic processes.

• Journal of the Korea Convergence Society
• /
• v.11 no.11
• /
• pp.33-39
• /
• 2020

In the field of Natural Language Processing, Various research such as Translation, POS Tagging, Q&A, and Sentiment Analysis are globally being carried out. Sentiment Analysis shows high classification performance for English single-domain datasets by pretrained sentence embedding models. In this thesis, the classification performance is compared by Korean E-commerce online dataset with various domain attributes and 6 Neural-Net models are built as BOW (Bag Of Word), LSTM[1], Attention, CNN[2], ELMo[3], and BERT(KoBERT)[4]. It has been confirmed that the performance of pretrained sentence embedding models are higher than word embedding models. In addition, practical Neural-Net model composition is proposed after comparing classification performance on dataset with 17 categories. Furthermore, the way of compressing sentence embedding model is mentioned as future work, considering inference time against model capacity on real-time service.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.25 no.11
• /
• pp.1505-1511
• /
• 2021

Pretrained models with high performance in various tasks within natural language processing have the advantage of learning the linguistic patterns of sentences using large corpus during the training, allowing each token in the input sentence to be represented with appropriate feature vectors. One of the methods of constructing a corpus required for a pre-trained model training is a collection method using web crawler. However, sentences that exist on web may contain unnecessary words in some or all of the sentences because they have various patterns. In this paper, we propose a dataset construction method for filtering sentences containing unnecessary words using neural network models for corpus collected from the web. As a result, we construct a dataset containing a total of 2,330 sentences. We also evaluated the performance of neural network models on the constructed dataset, and the BERT model showed the highest performance with an accuracy of 93.75%.