Building a text collection for Urdu information retrieval

Urdu is a widely spoken language in the Indian subcontinent with over 300 million speakers worldwide. However, linguistic advancements in Urdu are rare compared to those in other European and Asian languages. Therefore, by following Text Retrieval Conference standards, we attempted to construct an extensive text collection of 85 304 documents from diverse categories covering over 52 topics with relevance judgment sets at 100 pool depth. We also present several applications to demonstrate the effectiveness of our collection. Although this collection is primarily intended for text retrieval, it can also be used for named entity recognition, text summarization, and other linguistic applications with suitable modifications. Ours is the most extensive existing collection for the Urdu language, and it will be freely available for future research and academic education.     

一种基于小字典不对等语料的跨语言词嵌入方法

双语词嵌入通常采用从源语言空间到目标语言空间映射,通过源语言映射嵌入到目标语言空间的最小距离线性变换实现跨语言词嵌入。然而大型的平行语料难以获得,词嵌入的准确率难以提高。针对语料数量不对等、双语语料稀缺情况下的跨语言词嵌入问题,该文提出一种基于小字典不对等语料的跨语言词嵌入方法,首先对单语词向量进行归一化,对小字典词对正交最优线性变换求得梯度下降初始值,然后通过对大型源语言(英语)语料进行聚类,借助小字典找到与每一聚类簇相对应的源语言词,取聚类得到的每一簇词向量均值和源语言与目标语言对应的词向量均值,建立新的双语词向量对应关系,将新建立的双语词向量扩展到小字典中,使得小字典得以泛化和扩展。最后,利用泛化扩展后的字典对跨语言词嵌入映射模型进行梯度下降求得最优值。在英语—意大利语、德语和芬兰语上进行了实验验证,实验结果证明该文方法可以在跨语言词嵌入中减少梯度下降迭代次数,减少训练时间,同时在跨语言词嵌入上表现出较好的正确率。     

高正确率的双语语块对齐算法研究

高质量的自动对齐双语语块,对于机器翻译系统,特别是计算机辅助翻译系统的性能提高有重要作用,而且对于人工翻译以及辞典编纂也都有巨大的应用价值。该文提出基于单词间粘合度与松弛度的语块划分评分方法以及双语语块划分的双向约束算法,使得源语言和目标语言的语块的划分与对齐能相互促进。与传统方法相比,因为无需事先进行双语语块划分,而是在搜索最佳对齐时动态地考察划分效果,故可以减少边界划分错误对对齐结果的影响。该算法获得了远超过传统算法的高正确率。     

基于70年报刊语料的现代汉语历时稳态词抽取与考察

该文基于70年跨度的历时报刊语料库,使用九种统计方法计算了词语历年的使用情况,并通过对稳定性、覆盖度和时间区分性能的考察筛选获得了规模为3 013词的历时稳态词候选词集。该词集中动词与名词各占约三分之一(其余为形容词、副词与虚词),平均词长约1.7字,前密后疏地分布于历时语料库总频序表的前7 609位,覆盖了总语料的近九成。该部分词语中包含大量构造句子结构的核心词语。它们塑造了稳态词在词长和词类上的特性。稳态词的提取可以加深对语言生活底层与基础词汇的认识,对汉语教学、中文信息处理和语言规划都具有重要意义。     

Language supports for journal abstract writing across disciplines

Various writing assistance tools have been developed through efforts in the areas of natural language processing with different degrees of success of curriculum integration depending on their functional rigor and pedagogical designs. In this paper, we developed a system, WriteAhead, that provides six types of suggestions when non‐native graduate students of English from different disciplines are composing journal abstracts, and assessed its effectiveness. The method involved automatically building domain‐specific corpora of abstracts from the Web via domain names and related keywords as query expansions, and automatically extracting vocabulary and n‐grams from the corpora in order to offer writing suggestions. At runtime, learners' input in the writing area of the system actively triggered a set of corresponding writing suggestions. This abstract writing assistant system facilitates interactions between learners and the system for writing abstracts in an effective and contextualized way, by providing suggestions such as collocations or transitional words. For assessment of WriteAhead, we compared the writing performance of two groups of students with or without using the system, and adopted student perception data. Findings show that the experiment group wrote better, and most students were satisfied with the system concerning most suggestion types, as they can effectively compose quality abstracts through provided language supports from WriteAhead.     

中文电子病历命名实体标注语料库构建

针对中文电子病历命名实体语料标注空白的现状,研究了中文电子病历命名实体标注语料库的构建。参考2010年美国国家集成生物与临床信息学研究中心(1282)给出的电子病历命名实体类型及修饰类型的定义,在专业医生的指导下制定了详尽的中文电子病历标注规范;通过对大量中文电子病历的分析,提出了一套完整的中文电子病历命名实体标注方案,而且采用预标注和正式标注的方法,建立了一定规模的中文电子病历命名实体标注语料库,其标注语料的一致性达到了92%以上。该工作对中文电子病历的命名实体识别及信息抽取研究提供了可靠的数据支持,对医疗知识挖掘也有重要意义。     

英汉平行语料中双语兼类词消歧研究

对于一部分目前统计处理消歧效果较差、但出现频率又很高的兼类词,手工编写针对性极强的消歧规则。在未经词汇对齐的平行语料中,实现了基于个性规则的词性消歧方法。本研究为5个典型兼类词（过去、计划、与、back、so）设计的平行消歧算法,在大规模平行语料中得到了验证,平均F值达到了98.45%。研究结果表明该规则具有不受上下文长度和模板数量限制、特别适合于双语平行处理、消歧效果好等优点。     

利用五度字调模型实现普通话到兰州方言的转换

方言语音的转换对于实现个性化的汉语语音合成系统具有重要的意义。为了实现普通话到兰州方言的转换,论文利用《方言调查字表》建立了兰州方言的文本语料库,录制了普通话和兰州方言平行的语音语料库。利用五度字调模型建立了普通话和兰州方言的基频模型,利用统计方法建立了时长转换模型和停顿时长转换模型。在转换时,通过利用STRAIGHT算法修改普通话语音实现普通话到兰州方言的转换。对转换结果的MOS评测表明,转换后的单字平均MOS得分4.17,双字平均MOS得分为4.22,语句的平均MOS得分为3.55。     

The “good” limb makes the “bad” limb worse: Experience-dependent interhemispheric disruption of functional outcome after cortical infarcts in rats.

Following stroke-like lesions to the sensorimotor cortex in rats, experience with the ipsi-to-lesion (ipsilesional), “nonparetic”, forelimb worsens deficits in the contralesional, “paretic”, forelimb. We tested whether the maladaptive effects of experience with the nonparetic limb are mediated through callosal connections and the contralesional sensorimotor cortex. Adult male rats with proficiency in skilled reaching with their dominant (for reaching) forelimb received ischemic bilateral sensorimotor cortex lesions, or unilateral lesions, with or without callosal transections. After assessing dominant forelimb function (the paretic forelimb in rats with unilateral lesions), animals were trained with their nonparetic/nondominant forelimb or underwent control procedures for 15 days. Animals were then tested with their paretic/dominant forelimb. In animals with unilateral lesions only, nonparetic forelimb training worsened subsequent performance with the paretic forelimb, as found previously. This effect was not found in animals with both callosal transections and unilateral lesions. After bilateral lesions, training the nondominant limb did not worsen function of the dominant limb compared with controls. Thus, the maladaptive effects of training the nonparetic limb on paretic forelimb function depend upon the contralesional cortex and transcallosal projections. This suggests that this experience-dependent disruption of functional recovery is mediated through interhemispheric connections of the sensorimotor cortex. (PsycINFO Database Record (c) 2010 APA, all rights reserved)