• Journal of the Korean Society for information Management
• /
• v.35 no.3
• /
• pp.101-121
• /
• 2018

This study proposes the analysis method in sentence semantics that can be automatically identified and processed as appropriate items in the system according to the composition of the sentences contained in the data corresponding to the logical semantic structure metadata of the research papers. In order to achieve the purpose, the structure of sentences corresponding to 'Research Objectives' and 'Research Outcomes' among the semantic structure metadata was analyzed based on the number of words, the link word types, the role of many-appeared words in sentences, and the end types of a word. As a result of this study, the number of words in the sentences was 38 in 'Research Objectives' and 212 in 'Research Outcomes'. The link word types in 'Research Objectives' were occurred in the order such as Causality, Sequence, Equivalence, In-other-word/Summary relation, and the link word types in 'Research Outcomes' were appeared in the order such as Causality, Equivalence, Sequence, In-other-word/Summary relation. Analysis target words like '역할(Role)', '요인(Factor)' and '관계(Relation)' played a similar role in both purpose and result part, but the role of '연구(Study)' was little different. Finally, the verb endings in sentences were appeared many times such as '~고자', '~였다' in 'Research Objectives', and '~었다', '~있다', '~였다' in 'Research Outcomes'. This study is significant as a fundamental research that can be utilized to automatically identify and input the metadata element reflecting the common logical semantics of research papers in order to support researchers' scholarly sensemaking.

• The Journal of the Acoustical Society of Korea
• /
• v.24 no.8
• /
• pp.498-504
• /
• 2005

We introduce a Korean speech recognition platform (ECHOS) developed for education and research Purposes. ECHOS lowers the entry barrier to speech recognition research and can be used as a reference engine by providing elementary speech recognition modules. It has an easy simple object-oriented architecture, implemented in the C++ language with the standard template library. The input of the ECHOS is digital speech data sampled at 8 or 16 kHz. Its output is the 1-best recognition result. N-best recognition results, and a word graph. The recognition engine is composed of MFCC/PLP feature extraction, HMM-based acoustic modeling, n-gram language modeling, finite state network (FSN)- and lexical tree-based search algorithms. It can handle various tasks from isolated word recognition to large vocabulary continuous speech recognition. We compare the performance of ECHOS and hidden Markov model toolkit (HTK) for validation. In an FSN-based task. ECHOS shows similar word accuracy while the recognition time is doubled because of object-oriented implementation. For a 8000-word continuous speech recognition task, using the lexical tree search algorithm different from the algorithm used in HTK, it increases the word error rate by $40\%$ relatively but reduces the recognition time to half.

• The Journal of the Acoustical Society of Korea
• /
• v.23 no.4E
• /
• pp.112-119
• /
• 2004

This paper presents a method for the automatic recognition of pitch accents with no prior knowledge about the phonetic content of the signal (no knowledge of word or phoneme boundaries or of phoneme labels). The recognition algorithm used in this paper is a time-delay recurrent neural network (TDRNN). A TDRNN is a neural network classier with two different representations of dynamic context: delayed input nodes allow the representation of an explicit trajectory F0(t), while recurrent nodes provide long-term context information that can be used to normalize the input F0 trajectory. Performance of the TDRNN is compared to the performance of a MLP (multi-layer perceptron) and an HMM (Hidden Markov Model) on the same task. The TDRNN shows the correct recognition of $91.9{\%}\;of\;pitch\;events\;and\;91.0{\%}$ of pitch non-events, for an average accuracy of $91.5{\%}$ over both pitch events and non-events. The MLP with contextual input exhibits $85.8{\%},\;85.5{\%},\;and\;85.6{\%}$ recognition accuracy respectively, while the HMM shows the correct recognition of $36.8{\%}\;of\;pitch\;events\;and\;87.3{\%}$ of pitch non-events, for an average accuracy of $62.2{\%}$ over both pitch events and non-events. These results suggest that the TDRNN architecture is useful for the automatic recognition of pitch accents.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.6 no.8
• /
• pp.1283-1291
• /
• 2002

In this paper, the PLL driven by the DDFS is designed on the schematic using the Q-logic cell based library and is implemented using FPGA QL32 x16B. The measurement results of the frequency synthesizer switching speed were agreement with a register. The simulated results show that the clock delay was generated after eleven clock and if input is random, It has influence on output DA converter has to be very extensive. Therefore, the DDFS used noise shaper to drive PLL by regular interval for input state. Also the bandwidth of DA converter very extensive, the simulation shows that the variation of small input control word is better than the switching speed of PLL.

• ETRI Journal
• /
• v.41 no.2
• /
• pp.235-241
• /
• 2019

In this paper, we propose a rank-weighted reconstruction feature to improve the robustness of a feed-forward deep neural network (FFDNN)-based acoustic model. In the FFDNN-based acoustic model, an input feature is constructed by vectorizing a submatrix that is created by slicing the feature vectors of frames within a context window. In this type of feature construction, the appropriate context window size is important because it determines the amount of trivial or discriminative information, such as redundancy, or temporal context of the input features. However, we ascertained whether a single parameter is sufficiently able to control the quantity of information. Therefore, we investigated the input feature construction from the perspectives of rank and nullity, and proposed a rank-weighted reconstruction feature herein, that allows for the retention of speech information components and the reduction in trivial components. The proposed method was evaluated in the TIMIT phone recognition and Wall Street Journal (WSJ) domains. The proposed method reduced the phone error rate of the TIMIT domain from 18.4% to 18.0%, and the word error rate of the WSJ domain from 4.70% to 4.43%.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.18 no.8
• /
• pp.2178-2198
• /
• 2024

Graph-based document models have good capabilities to reveal inter-dependencies among unstructured text data. Natural language processing (NLP) systems that use such models as an intermediate representation have shown good performance. This paper proposes a novel fuzzy graph-based document model and to demonstrate its effectiveness by applying fuzzy logic tools for text summarization. The proposed system accepts a text document as input and identifies some of its sentence level features, namely sentence position, sentence length, numerical data, thematic word, proper noun, title feature, upper case feature, and sentence similarity. The fuzzy membership value of each feature is computed from the sentences. We also propose a novel algorithm to construct the fuzzy graph as an intermediate representation of the input document. The Recall-Oriented Understudy for Gisting Evaluation (ROUGE) metric is used to evaluate the model. The evaluation based on different quality metrics was also performed to verify the effectiveness of the model. The ANOVA test confirms the hypothesis that the proposed model improves the summarizer performance by 10% when compared with the state-of-the-art summarizers employing alternate intermediate representations for the input text.

• Journal of KIISE
• /
• v.41 no.12
• /
• pp.1058-1065
• /
• 2014

In order to assist user's who are in the process of executing a search, a query expansion method suggests keywords that are related to an input query. Recently, several studies have suggested keywords that are identified by finding domains using a clustering method over the documents that are retrieved. However, the clustering method is not relevant when presenting various domains because the number of clusters should be fixed. This paper proposes a method that suggests keywords by finding various domains related to the input queries by using a community detection algorithm. The proposed method extracts words from the top-30 documents of those that are retrieved and builds communities according to the word graph. Then, keywords representing each community are derived, and the represented keywords are used for the query expansion method. In order to evaluate the proposed method, we compared our results to those of two baseline searches performed by the Google search engine and keyword recommendation using TF-IDF in the search results. The results of the evaluation indicate that the proposed method outperforms the baseline with respect to diversity.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.43 no.4 s.310
• /
• pp.31-38
• /
• 2006

In this paper, the inputting method of english pronunciation for a computer by the combining diacritical mark is studied. English pronunciation system and the methods of its notations are investigated and conditions to input english pronunciations easily are analysed. Therefore, the inputting method which can input 3, 4-level stress as well as 2-level stress is presented. By using this method, English pronunciation can be inputted to the spreadsheets, databases and presentations as well as word-processors, and each application program's data can have compatibility. In the result of experiments, every data can have the compatibility in all of application programs and inputting speed is increased highly compare with using the individual vowel method which has high speed than using the pre-existing functions of word processors.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.36 no.12C
• /
• pp.744-752
• /
• 2011

In intelligent surveillance system, various methods for detecting abnormal behavior were proposed recently. However, most researches are not robust enough to be utilized for actual reality which often has occlusions because of assumption the researches have that individual objects can be tracked. This paper presents a novel method to detect abnormal behavior by analysing major motion of the scene for complex environment in which object tracking cannot work. First, we generate Visual Word and Visual Document from motion information extracted from input video and process them through LDA(Latent Dirichlet Allocation) algorithm which is one of document analysis technique to obtain major motion information(location, magnitude, direction, distribution) of the scene. Using acquired information, we compare similarity between motion appeared in input video and analysed major motion in order to detect motions which does not match to major motions as abnormal behavior.

• Proceedings of the KIEE Conference
• /
• 2004.11c
• /
• pp.346-348
• /
• 2004

In this rarer, we introduce a new Fuzzy Polynomial Neural Networks (FPNNs)-like structure whose neuron is based on the Fuzzy Set-based Fuzzy Inference System (FS-FIS) and is different from that of FPNNs based on the Fuzzy relation-based Fuzzy Inference System (FR-FIS) and discuss the ability of the new FPNNs-like structurenamed Fuzzy Set-based Polynomial Neural Networks (FSPNN). The premise parts of their fuzzy rules are not identical, while the consequent parts of the both Networks (such as FPNN and FSPNN) are identical. This difference results from the angle of a viewpoint of partition of input space of system. In other word, from a point of view of FS-FIS, the input variables are mutually independent under input space of system, while from a viewpoint of FR-FIS they are related each other. In considering the structures of FPNN-like networks such as FPNN and FSPNN, they are almost similar. Therefore they have the same shortcomings as well as the same virtues on structural side. The proposed design procedure for networks' architecture involves the selection of appropriate nodes with specific local characteristics such as the number of input variables, the order of the polynomial that is constant, linear, quadratic, or modified quadratic functions being viewed as the consequent part of fuzzy rules, and a collection of the specific subset of input variables. On the parameter optimization phase, we adopt Information Granulation (IG) based on HCM clustering algorithm and a standard least square method-based learning. Through the consecutive process of such structural and parametric optimization, an optimized and flexible fuzzy neural network is generated in a dynamic fashion. To evaluate the performance of the genetically optimized FSPNN (gFSPNN), the model is experimented with using gas furnace process dataset.