A Study of Interactive Electronic Dictionary System Construction Based on Automatic Synthesis Function of Sign Language Articulators

构建一个支持多媒体的电子词典系统是本研究的目的。这个电子词典的功能是自动生成动画图像。具有嘴型的动画图像的发音器满足组合结构。在本研究中，通过对三维动画构建电子词典系统的研究，对手语会话语料库部分、语言分析部分、发音生成部分、图像数据库部分的各个功能进行了考察。在手语会话语料库部分，笔者考察了在语料库收集中提出的去除日语影响的方法。根据我提出的数据收集方法，我从KOSIGN Ver.2中选取了994个手语词汇，对约3800个手语会话语料库进行了收集和录像。在此基础上，构建支持系统对标识对话视频图像进行分割分析。该系统能够方便地在各种分析项目中设置更多的标签，在其表单上有框架单元。在语言分析部分，我使用该分析支持系统对一个手语会话语料库进行了分析。分析对象是对手语特征和语素的提取。从分析结果中提出了NVGS句子层次语素描述模型。在动画图像生成部分，我提出了从描述生成动画图像的发音器的方法。该发音器通过NVGS语素模型和sIGNDEX对非手动信号进行语言结构描述。此外，该系统还生成了15个基本的嘴型功能，以提高手语动画图像的识别精度。在电子词典部分，我构建了一个交互式手语电子词典。该词典具有手语例句数据库检索功能和带嘴型生成的动画图像功能。此外，本词典的总词汇量为1142个符号词。本研究的目的和执行计划基本达到。研究成果发表在Journal of IEICE， Transactions of Human Interface Society， International Conference等期刊上。该成果于2001年3月获得了人机界面学会科学鼓励奖。我觉得有必要开展这项研究。在此研究成果的基础上，通过对手语会话的分析，继续从认知科学的角度进行分析工作。少

It is for a purpose of this research to construct an electronic dictionary system supporting multi-media. Function of this electronic dictionary is automatic generation of an animated image. An articulator of the animated image attached with mouth shape satisfies syntagmatic structure. In this research, I examined it in each function of sign language conversation corpus part, linguistic analysis part, articulation generation part, image database part on construction of electronic dictionary system by 3D animation through research period.1. In sign language conversation corpus part, I examined proposed in the corpus collection method which removed influence from Japanese. With the data collection method which I proposed, I chose 994 sign words from KOSIGN Ver.2 and had been collection and a video recording of about 3,800 sign language conversation corpus. Furthermore, I constructed support system to analyze segmentation of sign conversation video image. This system is able to easily set … More a tag in various analyses item with frame unit on its sheet.2. In language analysis part, I analyzed a sign language conversation corpus with this analysis support system. The analysis subject matter is extract of distinctive feature and a morpheme of sign language. I proposed of NVGS morpheme descriptive model of sentence level from an analysis result.3. In animated image generation part, I proposed in the method that generated an articulator of an animated image from description. The articulator described language construction by NVGS morpheme model with sIGNDEX for non-manual signals. Furthermore, this system has 15 basic mouth-shaped function to generate to improve recognition accuracy of a sign language animated image.4. In electronic dictionary part, I constructed an interactive sign language electronic dictionary. In this dictionary, there are sign language example sentences database retrieval function and an animated image with the mouth shape generated function. In addition, total vocabulary number of this dictionary is 1,142 sign words.This research purpose and the execution plan were almost achieved as above. In addition, results of research are presented in Journal of IEICE, Transactions of Human Interface Society, International Conference, etc. " Construction of the Mouth Shape Animation System" (HIS2000) of these result got the sciences encouragement prize from Human Interface Society in March 2001.I feel necessity to develop this research. On the basis of result of this research, I continue analysis work from the side of cognitive science through conversation analysis of sign language. Less