一种汉语语音多级识别策略
[作者]邝继顺; 何鎏藻;
[摘要]本文提出了一个汉语语音识别的三级识别结构,在以半音节为识别基元的基础上,将基于语音生成模型的模式匹配法和基对“知识”的特征规则法有效地结合起来,从一个新的角度来研究汉语语音识别。通过对汉语语音发音方式和拼读规则的分析及实验,总结出用于声母分类的有效特征及各级判决中的若干判别规则。实验结果表明,系统达到了较高的性能指标。前半音节的识别率为94.7％,存貯量和计算量分别和一般全音节识别系统的1/10和1/36。
[Abstract]This paper presents a 3-stage recognition structure of spoken Chinese. Using demisyllable as recognition primitive (unit), the recognition strategy combines the pattern inarching approach based on speech production model with the feature rule approach based on "knowledge", and permits us to investigate the spoken Chinese recognition from a new point of view. Via experiments and analysis of utterance manner and spelling pronunciation rule, a series of features used in consonant classification and dicision ru...
[关键字]声母分类; 汉语语音识别; 特征规则; 半音节; 语音生成模型; 识别基元; 模式匹配法;
|
基于合一语法的通用句法分析器:设计与实施
[作者]沙新时; 吴立德; 周斌;
[摘要]本文从建立一个通用的基于合一语法的句法分析器的实际出发,就特征结构、合一算法、基于合一的语法形式、使用基于合一的语法的方法等方面展开讨论,对某些模糊的东西加以澄清,对现有的各种方式、方法进行分析比较,开介绍了我们的方法,最后给出了结论,
[Abstract]In this paper, we concentrate our attentions on feature structures, unification algorithms, unification-based grammar formalisms, and the uses of unification-based grammars on the view of building practice NLP systems. After discussing sevaral existingmethods, we give our approcabes in details. At last, we draw on the conclusion that we can build a Unification-Based Universal Parser.
[关键字]原子特征; 特征结构; 句法分析器; 语法规则;
|
拼音语句汉字输入系统InSun
[作者]王晓龙;
[摘要]本文提出了基于键盘的字词输入技术之上的语句输入的思想,讨论了拼音语句输入的知识的表达、获取和系统实现,给出了测试结果及应用情况。
[Abstract]This paper presents the idea of Chinese sentence input after developing the technique of character input and word input, and discusses the knowledge representation, knowledge acquisition and system realization of the pinyin sentence input. The results of test and application of the pinyin sentence input are also introduced in this paper. Key Words. Man machine interface, Natural language understanding,
[关键字]词类; 音字转换; 拼音汉字转换; 产生式规则; 正确率; 语句输入; 汉字操作系统; 机器学习; 汉字输入系统; 知识库;
|
基于神经网络的分词方法
[作者]徐秉铮; 詹剑; 贺前华;
[摘要]本文在分析目前自动分词各种方法的基础上,提出了一种新的神经网络分析方法。这一新方法是以模拟人脑并行,分布处理和建立数值计算模型工作的。它将分词知识所分散隐式的方法存入神经网内部.通过自学习和训练修改内部权值,以达到正确的分词结果。最后给出神经网络自动分词结果。
[Abstract]The paper presents a new method of neural network based on analysing Current methods in automatic words segmentation. The new way works on process of parallal and distribution in imitating human brain and establishes a mathematical model. The knowledge of words segmentation can be stored in neural network with hidden and distribution way. Through self-learning, Training and modifing weights in neural network, The correct results of words segmentation can be achieved. Finally, the simulations of words segmen...
[关键字]歧义字段; 神经网络; 神经元; 中文自动分词; 分词方法; 自学习;
|
FPY中的同音词智能识别方法
[作者]万建成;
[摘要]FPY是作者研制的以词为识别单元、以拼音为编码的汉字智能输入系统。本文介绍了该输入系统中的数种同音词的智能识别方法,其中包括部分汉语短语句法分析方法,和作者首先提出并实现的几种词法和语义相关分析方法。
[Abstract]FPY is an intelligent Chinese input system, which is based on word pioce ssing and Pinyin coding. In this paper, several of its major intelligent recognition methods of Chinese homophone words are presented. These include the Chinese morphological recognition, some phrase syntactical analysis, and semantic association analysis, of which the last two methods are proposed and have been implemensed in FPY by the author.
[关键字]智能识别; 相关分析; 动宾短语; 动补短语; 语义相关; 同音词; 句法分析; 数量短语;
|
歧义、系统歧义和语境
[作者]钱树人;
[摘要]本文对歧义现象,特别对语言片段的歧义理解进行了剖析,并研究了不同语境对理解歧义的影响。进而提出了默认语境,系统语境和系统歧义等概念,开简要地介绍了汉语语言片段歧义分析模型系统CAAMS。
[Abstract]This paper analyzes an understanding of language segment ambiguity in detail, and effect on ambiguity in different language environment. Furthermore, some conception such as defailt language environment, systematic language environment, Language environment condition, systematic ambiguity, are specified. And the CAAMS, Chinese segment Ambiguity Analysis Model System, is intrucluced.
[关键字]语法结构; 形式语言; 语境条件; 歧义理解; 真值理解; 自然语言理解; 语法规则;
|
基于规则的高效索引算法和排序算法
[作者]杨宪泽;
[摘要]本文提出产生式规则索引算法和排序算法,用以提高智能系统运行速度。索引算法和排序算法都缩短了规则匹配时间。而且,排序算法不仅适宜规则静态排序,也适宜规则动态排序。
[Abstract]For enhancing the speed of AI sty stem, We presented the efficient sort and index algorithms. Based on production rules in this paper. All the algorithms saved the time for rule matching.The experiment result has proved that the sort algorithms can not only be used in the static sort bwt also appropriated to dynamic.
[关键字]字符串长度; 索引算法; 排序算法; 基于规则;
|
汉语孤立字声调的模糊识别方法
[作者]徐士林; SamuelC.Lee;
[摘要]本文应用模糊集合来识别汉语孤立字的声调。孤独字的四声调可被描述成四种模式类的模糊集合。由于四声调的基音轮廓具有其固定模式,因之在此基础上可构成模糊集合的隶属函数。方法中使用隶属函数为模式分类的判别函数。这些隶属函数既简单又易于计算,故适宜实对执行。实验结果表明,总的识别率高于99％。
[Abstract]This paper presents a tone recognition method of Chinese isolated words using fuzzy sets. The four tones are considered as four fuzzy sets which describe the four pattern classes. Since the pitch contours of the four tones have their distinct patterns, they are used as the basis to construct the membership functions of the four fuzzy sets. The tone recognition algorithm uses these membership functions as their pattern classification decision functions. These functions are simple, easy to compute, and theref...
[关键字]隶属函数; 基音检测; 模糊识别方法; 声调识别;
|
以成语为范围的词汇支援系统
[作者]林联合; 吴杰; 吴亮;
[摘要]这篇论文介绍有关调用汉语成语的词汇支援系统。人们给系统一个话题后,该系统将针对该话题提供一批成语来帮助人们选择和寻找词语。文章还介绍了成语的知识表达方法和自动的语义检索过程。
[Abstract]This paper introduces a consultative system for the choice and use of Chinese idioms. People offer a topic to the system, the system will provide some idioms for their choice in order to get an appropriate expression. This paper also introduces the knowledge representation of Chinese idioms and the automatic process of semantic retrieval.
[关键字]汉语成语; 支援系统; 词汇; 语义检索; 检索词;
|
统计语言模型及汉语音字转换的一些新结果
[作者]郭进;
[摘要]汉语音字转换是一个重要而困难的问题。语料库语言学为我们提供了新思路。作者们通过建立统计语言模型,将基于语料库的方法与传统的基于规则的方法结合,研制了THED新一代音字转换系统。该系统对随机抽取的新华社新闻语料有不低于95％的带调音节和国标汉字的转换正确率。本文侧重报道该系统在汉语音字转换方面及与此相关的汉语切词和词性标注方面的一些实验结果,也简要介绍该系统在语料库应用方面的一些思路。
[Abstract]The problem of transcribing a sentence of Chinese syllables into a sentence of Chinese graphic words is an important and yet difficult one. A new technology is highlighted by the recently booming Corpus Linguistics. By combining corpus-based approaches with traditional rule-based ones, the author has developed a new generation Chinese syllables-to-words transcription system which is capable of transcribing tonal syllable sentence into standard simplified Chinese graphic word sentence with a character accura...
[关键字]语料库; 统计语言模型; 汉语音字转换;
|
共95页 当前第71页 