• Phonetics and Speech Sciences
• /
• v.4 no.2
• /
• pp.127-134
• /
• 2012

The purpose of this study is to present evidence for a particular processing approach based on the language-specific characteristics of Korean. To compare individuals' sentence-comprehension abilities, this study measured the accuracy and reaction times (RT) of 12 aphasic patients (AP) and 12 normal controls (NC) during a sentence-picture matching task. Four versions of a sentence were constructed with the two types of voice (active/passive) and two types of word order (agent-first/patient-first). To examine the effects of increased processing demand, picture stimuli were manipulated in such a way that they appeared immediately after the sentence was presented. As expected, the AP group showed higher error rates and longer RT for all conditions than the NC group. Furthermore, Korean speakers with aphasia performed above a chance level in sentence comprehension, even with passive sentences. Aphasics understood sentences more quickly and accurately when they were given in the active voice and with agent-first order. The patterns of the NC group were similar. These results confirm that Korean adults with aphasia do not completely lose their knowledge of sentence comprehension. When the processing demand was increased by delaying the picture stimulus onset, the effect of increased processing demands on RT was more pronounced in the AP than in the NC group. These findings fit well with the idea that the computational system for interpreting sentences is intact in aphasics, but its ability is compromised when processing demands increase.

• Journal of Information Processing Systems
• /
• v.15 no.6
• /
• pp.1503-1516
• /
• 2019

In this paper, to improve the accuracy of long sentence similarity calculation, we proposed a sentence similarity calculation method based on a system similarity function. The algorithm uses word2vector as the system elements to calculate the sentence similarity. The higher accuracy of our algorithm is derived from two characteristics: one is the negative effect of penalty item, and the other is that sentence similar function (SSF) based on word2vector similar elements doesn't satisfy the exchange rule. In later studies, we found the time complexity of our algorithm depends on the process of calculating similar elements, so we build an index of potentially similar elements when training the word vector process. Finally, the experimental results show that our algorithm has higher accuracy than the word mover's distance (WMD), and has the least query time of three calculation methods of SSF.

• Journal of Information Processing Systems
• /
• v.16 no.4
• /
• pp.820-831
• /
• 2020

Extractive document summarization aims to select a few sentences while preserving its main information on a given document, but the current extractive methods do not consider the sentence-information repeat problem especially for news document summarization. In view of the importance and redundancy of news text information, in this paper, we propose a neural extractive summarization approach with joint sentence semantic clipping and selection, which can effectively solve the problem of news text summary sentence repetition. Specifically, a hierarchical selective encoding network is constructed for both sentence-level and document-level document representations, and data containing important information is extracted on news text; a sentence extractor strategy is then adopted for joint scoring and redundant information clipping. This way, our model strikes a balance between important information extraction and redundant information filtering. Experimental results on both CNN/Daily Mail dataset and Court Public Opinion News dataset we built are presented to show the effectiveness of our proposed approach in terms of ROUGE metrics, especially for redundant information filtering.

• KIPS Transactions on Software and Data Engineering
• /
• v.11 no.3
• /
• pp.125-132
• /
• 2022

Sentence compression is a natural language processing task that generates concise sentences that preserves the important meaning of the original sentence. For grammatically appropriate sentence compression, early studies utilized human-defined linguistic rules. Furthermore, while the sequence-to-sequence models perform well on various natural language processing tasks, such as machine translation, there have been studies that utilize it for sentence compression. However, for the linguistic rule-based studies, all rules have to be defined by human, and for the sequence-to-sequence model based studies require a large amount of parallel data for model training. In order to address these challenges, Deleter, a sentence compression model that leverages a pre-trained language model BERT, is proposed. Because the Deleter utilizes perplexity based score computed over BERT to compress sentences, any linguistic rules and parallel dataset is not required for sentence compression. However, because Deleter compresses sentences only considering perplexity, it does not compress sentences by reflecting the linguistic information of the words in the sentences. Furthermore, since the dataset used for pre-learning BERT are far from compressed sentences, there is a problem that this can lad to incorrect sentence compression. In order to address these problems, this paper proposes a method to quantify the importance of linguistic information and reflect it in perplexity-based sentence scoring. Furthermore, by fine-tuning BERT with a corpus of news articles that often contain proper nouns and often omit the unnecessary modifiers, we allow BERT to measure the perplexity appropriate for sentence compression. The evaluations on the English and Korean dataset confirm that the sentence compression performance of sentence-scoring based models can be improved by utilizing the proposed method.

• The Transactions of the Korea Information Processing Society
• /
• v.2 no.2
• /
• pp.229-236
• /
• 1995

The conceptual world of sentence is composed f substantives(nouns) and verbal. The verbal is a semantic center of sentence, the substantives are placed under control of verbal, and they are combined in a various way. In this paper, the structural relation of verbal and substantives are analyzed and the phrase unit sentence which is derived from the result of morphological analysis is interpreted by a variable pattern net. This variable pattern net analyzes the phrases syntactically and semantically and extracts conceptual units of clausal form. This paper expands the traditionally restricted Horn clause theory to the general sentence, separates a simple sentence from a complex sentence automatically, constructs knowledge base by clausal form of logical conceptual units, and applies it to a question-answering system.

• The Transactions of the Korea Information Processing Society
• /
• v.7 no.4
• /
• pp.1132-1140
• /
• 2000

The purpose of this paper was to present the method of progress efficiency of morphosyntaxical analyzer for French information processing in the view of multilingual information processing. This study indicated that he method prohibit the additional useless word type by decomposing word type by decomposing works which were created by morphological amalgamation. Findings also suggested the need of preprocessing in order to decrease the overload of morphosyntaxical analysis algorithm. In addition, general rules were proposed to divide word form and to decompose complex sentence.

• Journal of Information Processing Systems
• /
• v.13 no.4
• /
• pp.752-777
• /
• 2017

This paper proposes an automatic method to summarize Bangla news document. In the proposed approach, pronoun replacement is accomplished for the first time to minimize the dangling pronoun from summary. After replacing pronoun, sentences are ranked using term frequency, sentence frequency, numerical figures and title words. If two sentences have at least 60% cosine similarity, the frequency of the larger sentence is increased, and the smaller sentence is removed to eliminate redundancy. Moreover, the first sentence is included in summary always if it contains any title word. In Bangla text, numerical figures can be presented both in words and digits with a variety of forms. All these forms are identified to assess the importance of sentences. We have used the rule-based system in this approach with hidden Markov model and Markov chain model. To explore the rules, we have analyzed 3,000 Bangla news documents and studied some Bangla grammar books. A series of experiments are performed on 200 Bangla news documents and 600 summaries (3 summaries are for each document). The evaluation results demonstrate the effectiveness of the proposed technique over the four latest methods.

• Smart Media Journal
• /
• v.10 no.1
• /
• pp.63-69
• /
• 2021

When a person sees a sentence and understands the sentence, the person understands the sentence by reminiscent of the main word in the sentence as an image. Text-to-image is what allows computers to do this associative process. The previous deep learning-based text-to-image model extracts text features using Convolutional Neural Network (CNN)-Long Short Term Memory (LSTM) and bi-directional LSTM, and generates an image by inputting it to the GAN. The previous text-to-image model uses basic embedding in text feature extraction, and it takes a long time to train because images are generated using several modules. Therefore, in this research, we propose a method of extracting features by using the attention mechanism, which has improved performance in the natural language processing field, for sentence embedding, and generating an image by inputting the extracted features into the GAN. As a result of the experiment, the inception score was higher than that of the model used in the previous study, and when judged with the naked eye, an image that expresses the features well in the input sentence was created. In addition, even when a long sentence is input, an image that expresses the sentence well was created.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2023.11a
• /
• pp.667-668
• /
• 2023

IT 기술의 발달로 인해 뉴스를 제공하는 플랫폼들이 다양해 졌고 최근 해외 인터뷰 영상, 해외 뉴스를 Youtube Shorts형태로 제작하여 화자의 의도와는 다른 자막을 달며 가짜 뉴스가 생성되는 문제가 대두되고 있다. 이에 본 논문에서는 Sentence-BERT를 활용한 YouTube 가짜 뉴스 탐지 시스템을 제안한다. 제안하는 시스템은 Python 라이브러리를 사용해 유튜브 영상에서 음성과 영상 데이터를 분류하고 분류된 영상 데이터는 EasyOCR을 사용해 자막 데이터를 텍스트로 추출 후 Sentence-BERT를 활용해 문자 유사도를 분석한다. 분석결과 음성 데이터와 영상 자막 데이터가 일치한 경우 일치하지 않은 경우보다 약 62% 더 높은 문장 유사도를 보였다.

• Journal of Korean Philosophical Society
• /
• v.139
• /
• pp.1-21
• /
• 2016

To quickly and accurately grasp the consistency and the true/false of sentence description, we may require the help of a computer. It is thus necessary to research and quickly and accurately grasp the consistency and the true/false of sentence description by computer processing techniques. This requires research and planning for the whole study, namely a plan for the necessary tables and those of processing, and development of the table of the five logic rules. In future research, it will be necessary to create and develop the table of ten basic inference rules and the eleven kinds of derived inference rules, and it will be necessary to build a DB of those tables and the computer processing of sentence logic using server programming JSP and client programming JAVA over its foundation. In this paper we present the overall research plan in referring to the logic operation table, dividing the logic and inference rules, and preparing the listed process sequentially by dividing the combination of their use. These jobs are shown as a variable table and a symbol table, and in subsequent studies, will input a processing table and will perform the utilization of server programming JSP, client programming JAVA in the construction of subject/predicate part activated DB, and will prove the true/false of a sentence. In considering the table prepared in chapter 2 as a guide, chapter 3 shows the creation and development of the table of the five logic rules, i.e, The Rule of Double Negation, De Morgan's Rule, The Commutative Rule, The Associative Rule, and The Distributive Rule. These five logic rules are used in Propositional Calculus, Sentential Logic Calculus, and Statement Logic Calculus for sentence logic.