Earticle

According to Genetic algorithms principle, the new hybrid evolutionary algorithm (HEA) is proposed in this paper by combining the Immune algorithm, Genetic algorithm and Pareto optimal solutions. The HEA has high convergence precision and improved the diversity of population. Multiple near optimization paths can be developed by the algorithm with multi‐objective restriction, and satisfy to minimize the routing of transportation and the numbers of the vehicles. The HEA has been used to solve the vehicle routing problem, the results of simulation experiment show that the HEA can gain higher global convergence rate and higher speed.

Expert Finding is a frequently faced problem in enterprises and organizations. However, due to the variety of types, of data and sparse expertise information, it is very difficult to find information relevant to experts. In this paper, we proposed an approach automatically extracting experts’ relevant information from documents in the corpus and then aggregating to an expert description document, which is based on the assumption that the knowledge of expert is representable. In TREC 2005 enterprise track, our method achieves the best performance among the participants and makes the improvement nearly 50% upon the traditional approaches.

The Secrete History of Mongols (SHM) is a special Mongolian historical document transliterated with Chinese characters as phonetic symbols, which features caused many interested researches in later ages. The discussion of this paper is mostly about the used Chinese characters, the frequencies, and the rules of transliteration on SHM. All the statistic data and the analysis is based on the electronic text of SHM, which supports providing those important data in an all‐round way, including the whole transliterated Chinese characters, aligned Chinese glosses by the side of characters, and translated original Chinese paragraphs. Furthermore, four types of Chinese characters as phonetic symbols and their information have been discussed, which are type C, xC, Cy, and xCy as in original forms: 安, 中豁, 阿勒, and 舌魯黑. Shortly, this statistics is so far a most completed inspection on the text of the Secrete History of Mongols.

In the non‐linear faint signal measurement, multiplications of amplifier are often needed to be adjusted according to the signal intensity extent and the non‐linear region continually, however the measurement circuit is complicated, of low precision, and difficult to adjust. Using PGA, the amplifier is designed to realize high‐accuracy, high multiplications; this amplifier and the singlechip circuit constitute the humidity detecting system of adjusting gain voluntarily. This system has excellent noise immunity and strong overcoming non‐linear performance.

多语种合成（TTS）语料库的构建过程，包括语料设计、数据收集和语音学标注等。其中语料设计和语音标注都涉及了大量的语音学知识。通过标准化的数据管理方式，多语言合成语料库可以采用统一架构满足基于功能的选择策略和基于服务器及嵌入式平台的语音合成模块的使用需求。

Endpoint detection is the first essential technology which speech recognition system meets in pre‐processing stage. This algorithm which based Tibetan vowel/consonant frequency spectrum characteristic, separately is processed again through the pronunciation signal minute high/low‐frequency band, conforms to Tibetan pronunciation clear/muddy opposition information distribution characteristic, then separately withdraws but actually is scored the cepstral coefficient to take the endpoint detection characteristic, because the cepstral coefficient actually scores the information which the characteristic contains compared to other parameters many, can attribute the better attribute pronunciation signal, the pronunciation quality is good, the recognition accuracy is high; When examination adopt the auto‐adapted noise parameter to estimate that, decided beginning/end vertex according to ceptrum distantce, the simulation result indicated its superiority.

We’ve passed through several phases and tried on different methods for the study of Mongolian language (both as a source language and a target language) machine translation. As far as the Chinese‐Mongolian MT research is concerned, we’ve implemented a rule‐based system and an example‐based system as well. Statistics‐based method has been used in the implementation of the Chinese‐Mongolian MT system lately. We also did research on English‐Mongolian, Mongolian‐Japanese, Japanese‐Mongolian and Mongolian‐Chinese MT, using rule‐based method and template‐based method. Mongolian language cannot only be a source language but a target language in a machine translation system, and according to this, we can divide the study on Mongolian language machine translation into two parts: X‐Mongolian MT and Mongolian‐X MT(Xstands for any natural language except for Mongolian language). Generally, we made much more research on X‐Mongolian MT than Mongolian‐X MT. Reviewing the history of the research on Mongolian language machine translation will be very helpful for the researchers who are doing the similar studies of minority language machine translation.

Yi Word segmentation has not been well studied yet. This paper introduces the research of language’s automatic word segmentation at first , on this foundation and according to Yi language’s characteristic , puts forward the way which are suiting and designed algorithm choosing for the Yi automatic word segmentation ‘s structure frame and realizing the method , and expounds the appraisal of every function module of Yi automatic word segmentation, test parameter, and discusses the realistic meaning on the development trend of Yi automatic word segmentation .

The evaluation of texture similarity is very important in content‐based image retrieval systems. Firstly the rotation invariance of grayprimitive co‐matrix was proved in this paper, then the texture image retrieval technique based on grey‐primitive co occurrence matrix was presented, The result of experiment indicates that the ability of this method to resist noise is better than those traditional methods.

Firstly the Decomposition Least Mean (D‐LMS) algorithm is analyzed deeply in theoretical, then a variable step size D‐LMS algorithm is proposed. By using the MATLAB based simulation model, the system identification results of the stable step size D‐LMS algorithm and the ones of the variable step‐size D‐LMS algorithm are both obtained. The simulation results show that compared with the stable step size D‐LMS algorithm, this variable step‐size D‐LMS algorithm based on wavelet transform has the better convergence ability and tracing ability, but the less state errors.

The speech endpoints detection will decide the speech recognition rate. A new method using Energy‐Frequency‐Value(EFV) which studied in speech endpoints dectection, had improved more accurate rate in speech endpoints dectection than conventional method which only integrated shortterm energy and zero‐crossing rate.

This paper will contrast isolated phonetics of Chinese and Tibetan consonants phoneme systems in the same period, which will categorize, summarize, and contrast the number in articulator position and techniques of consonants, as well as identify characteristics of each consonant amply. As the datum showing, Chinese consonants are fewer than Tibetan consonants, and Chinese categories in articulator position and techniques is almost the same as that of Tibetan, both of which have little consonants in accordant. Tibetan differs is in different areas, so is the consonant phoneme systems, while Chinese differs.

The Tibetan sound field investigation is the basic work of Tibetan phonetics study. According to contemporary phonetics theory, this article uses advanced equipment and instrument in studying native language. It has expatiated the meaning of the Tibetan Ando dialect field investigation. It has explaining the ways and processes of the Tibetan Ando dialect field investigation from laying down investigation outline, collecting related information, fixing pronouncing partnership, making investigation notice and depiction. It has elaborated recording operation flow, recording environment and hardware and software instrument and the ways to arrange audioinformation. It has analyzed the results of the Tibetan Ando dialect field investigation.

Proposed a modified Nelder‐Mead simplex algorithm for training radial basis function neural networks (RBF‐NN). Traditional simplex algorithm consists of order, reflect, expand, contract and shrink operations, and to construct vertices on the subspace near the constraints. In the improved simplex algorithm, considered the new operate method in the feasible fields near the constrained boundary. This modified method is simple, accuracy and higher efficiency. Numerical experiments show that this method is promising for optimization design of temperature compensation networks.

The vehicle identifying system based on the mode of electronic ID card is presented. The system can identify automatically vehicles by using the radio frequency identification technology, ultrasonic technology and shortdistance radio frequency receive‐transmitting technology. The identifying method of the system is hardly influenced by environment and can identify the vehicle correctly, in order to reduce the fake or colon license plate to happen, and decontaminate traffic surroundings.

A design of high power strobe lamp based on interleaved parallel DC‐DC converters is proposed in this paper. When more than two modules are parallelly operated, the peak current of power can be dramatically reduced. Finally, the result of four modules parallelly operated is presented.

This thesis is on the research conducted to help foreigners by matching them with a Korean language learning system. LCMS and SCORM upholding the IMS/AICC international standards, with its basis in LMS, and its possibility of Korean language education using these learning systems, the principle points and facts connect LMS/LCMS to being the standard management system, the Competency Model. The learner's individual competency, Competency Registry/Repository, and the Competency Model which is focused on the needs of the community is based on an intellectual management system. The E‐Survey system analyzing the educational needs provides mobile services emphasizing 7 areas of research to be applied to the education of the Korean language, so that it's contents can be mobilized. First we must match these systems to work for the Chinese applying them in major cities such as Beijing, Shanghai, and Yanji. After its application in the Chinese language, it will further be serviced in English, Japanese, and Russian.

To investigate how different are loanword phonology and learner speech, we analyze the Korean speakers’ production of English words with a word‐final stop. We investigate whether the phonological cause of vowel insertion in the loanwords (e.g., rib[u] vs. rip[=] ) would appear in the phonetic measurements of the learner speech. In our experiment, we not only counted the instances of vowel insertion, but also measured the duration of preceding vowels of a word‐final voiced and voiceless consonant (e.g., r[i]b vs. r[i]p). When we compared the production of 6 Korean learners with that of 6 English native speakers, the learner speech did not reflect the insertion contrast of the loanwords, but the duration contrast of the native speech. After instruction, the learner speech developed toward the native speech value in terms of both insertion and duration. The result suggests that loanword phonology is different from the learner speech in that the learner speech does not reflect its L1 phonology while the loanword phonology does so. Additional finding is that the loanword phonology is constant, while learner speech is changing toward the target value.

한국의 음성 정보 기술을 세계적 수준으로 향상시키기 위하여 음성정보기술산업 지원센터(SITEC)에서는 다양한 언어 자원을 구축하여 배포하고 있으며, 또한 다른 기관에서 구축된 자원들도 배포하고 있다. 본 논문에서는 관련 연구자들의 참고를 위하여 5년 동안의 정부 지원 기간에 SITEC에서 수행된 관련 활동과 언어 자원 관련 기관인 한국전자통신 연구원(ETRI), 언어자원은행 (BOLA) 및 국립국어원의 활동에 대해서도 요약하여 기술하고자 한다.

For educational and research purposes, we design and evaluate a Korean speech recognition platform to build a decoder. The platform has an object‐oriented architecture so that researchers can modify the platform easily and evaluate the performance of a recognition algorithm of their interests. The platform has the following functionalities: Noise reduction, speech detection, feature extraction, hidden Markov model (HMM)‐based acoustic modeling, cross‐word modeling, ngram language modeling, n‐best search, word graph generation, and Korean‐specific language processing. The platform can handle both lexical search trees for large vocabulary speech recognition, and finite‐state networks for small‐tomedium vocabulary speech recognition. It performs the word‐dependent n‐best search algorithm with a bigram language model in the first forward search stage, then extracts a word lattice, and finally rescores the lattice with a trigram language model in the second backward search stage. In a large vocabulary continuous speech recognition task, we compare the performance of the platform with HTK and Julius.

지금의 이동휴대 통신단말기는 기존의 음성통화 뿐만 아니라 기술간 융복합을 통해 새로운 신기통신망 멀티미디어 단말기 형태로 변화하고 있다. 또한 새로운 시장 성장 가치를 제시함으로써 구체화된 상품화 전략과 기술 표준을 통한 선점의 중요성도 증대되고 있다. 최근 출시된 Apple사의 아이폰(iPhone), LG전자의 프라다 폰(FRADA phone)을 비롯한 외형적 변화를 시도하는 다양한 형태의 시작(試作)폰(Concept phone)은 이러한 움직임을 대변하는 예라 할 수 있다. 기존의 키보드 형 입력 장치를 터치스크린 기술을 적용하여 가변적으로 디스 플레이와 입력 구도를 하나로 묶음으로써 자판의 기능을 유지하면서 자판이 차지하는 구조적 제약을 넘어 점차 증대되는 멀티미디어 사용 환경을 우선 고려한 디스플레 이의 비중을 상대적으로 늘린 형태를 따르고 있다. 또한 제품 설계 디자인을 무엇보다도 우선 고려한 형태로 ‘선디자인 후 기술’ 적용의 형태로 탈 자판이 가능한 단말기로의 변화 움직임을 보인다 할 수 있다. 이러한 새로운 경향은 전체적 엄지손 자판 시장 변화에도 적지 않은 영향이 미칠 것이 예상되고 있어 새롭게 변화하는 이동 휴대 단말의 입력구도 형태 연구가 필요한 상태이다. 따라서 본 논문은 최근 엄지손 자판 기술변화를 분석하고 엄지손 자판의 정보화와 표준화를 위한 과제를 확인함으 로써 (한)국어정보학회를 중심으로 논의 중인 남북중 엄지손 자판의 표준화를 위한 고려해야할 엄지손 자판의 정보화 기준과 표준화를 위한 기술 기준을 확인하고자 한다.

1994년 이후 11년 동안 지속된 남북중 컴퓨터상의 언어 문자 관련 연구는 ICMIP2004 국제 학술대회를 통해 2008년 북경올림픽 대비 휴대용 통번역시스템 연구 남북중 합의 하는 등 다각적 학술교류가 이루어지고 있다. 본 논문은 이와 관련된 선행연구로 남한의 음성인식을 위한 삼분손익에 근거한 정격율과 북한의 AMM모델을 결합한 연구이다

The Digital TV switchover which brings the extra channels and additional services, is boosted all over the world, therefore Korean government legislates for the special enactment to achieve digital switchover effectively and efficiently. The digital signal is robust unlike analogue signal, which gradually fade the digital signal but, often provides data which is either perfect or non-existent at the receiving end. So the proper model on blank area and efficient field strength should be selected to improve the reception of digital signal. This paper would analyses the blank area of the National Capital Region through the survey on the field strength for Digital TV. The result shows that under ch 18 and over ch 29 in Digital TV should be devided, analysing efficient field strength related to urbanization and regional development, And Okumura-Hata model as the spectrum model shows 90% accuracy with the result of field test on field strength.

This paper having a purpose of dynamic frequency distribution and code allocation research by operator based on telecommunication and broadcasting convergency. To dynamic frequency distribution and code allocation, it analyze frequency status and define parameter for the effective frequency distribution. For the parameter of multimedia service optimization is data transmission rate(9.6K, 10K, 14.4K, 144K, 700K, 2Mbps), simultaneous access rate(10%), line usage rate(20%), threshold frequency rate (0.7) and population density(subscriber), Using adoption of frequency occupation rate and distribution rate by operator through HSDPA technology will allocate and distribute dynamic frequency allocation. Finally, without mutual interference in limited frequency bandwidth, it will show frequency usage rate through dynamic frequency allocation with achievement market establishment.

Classification of Korean characters is studied for keyboard design. Korean characters are classified as two sets type or three sets type. We suggest a modified model of current Korean Standard for key arrangement of keyboard. And we suggest 3 principles for design of three sets type Korean keyboard. This study can be used to design of input device of computer.

For the purpose of the Network in the home for multimedia service is one of the solution for delivery, processing, management and reuse of information. Home networking information devices in the home are required connection through wired and wireless network without time and location limitation. Normally, Home network system has been evolved using wired and wireless technologies to meet the technological requirement and installation cost. Home network system should be considered some kind of elements like multi physical layer and protocol support, security of QoS, transmission speed for multimedia service, easy to use at the existing premise and scalability.