Contextual feature selection for text classification

We present a simple approach for the classification of “noisy” documents using bigrams and named entities. The approach combines conventional feature selection with a contextual approach to filter out passages around selected features. Originally designed for call for tender documents, the method can be useful for other web collections that also contain non-topical contents. Experiments are conducted on our in-house collection as well as on the 4-Universities data set, Reuters 21578 and 20 Newsgroups. We find a significant improvement on our collection and the 4-Universities data set (10.9% and 4.1%, respectively). Although the best results are obtained by combining bigrams and named entities, the impact of the latter is not found to be significant.     

Spatial/temporal indexing and information visualization genre for environmental digital libraries

INTRODUCTION A large number of environmental digital librar- ies have been designed and implemented recently to effectively manage environmental data such as satel- lite photos and images, map images, cartographic materials, and so on. Support for storing, distributing, searching for, retrieving, and utilizing this informa- tion is imperative in order for it to be used to its maximum potential benefit and effectiveness (Hicks and Tochtermann, 1998). The variability of envi- ronmental data…     

Enhancing the effectiveness of concept inventories using textual analysis: investigations in an electrical engineering subject

ABSTRACT

Concept inventories (CIs) are assessment instruments designed to measure students’ conceptual understanding of fundamental concepts in particular fields. CIs utilise multiple-choice questions (MCQs), and specifically designed response selections, to help identify misconceptions. One shortcoming of this assessment instrument is that it fails to provide evidence of the causes of the misconceptions, or the nature of students’ conceptual understanding. In this article, we present the results of conducting textual analysis on students’ written explanations in order to provide better judgements into their conceptual understanding. We compared students’ MCQ scores in Signals and Systems Concept Inventory questions, with the textual analysis utilising vector analysis approaches. Our analysis of the textual data provided the ability to detect answers that students identified as a ‘guessed’ response. However, the analysis was unable to detect if conceptually correct ideas existed within the ‘guessed’ responses. The presented approach can be used as a framework to analyse assessment instruments that utilise textual, short-answer responses. This analysis framework is best suited for the restricted conditions imposed by the short-answer structure.     

The effect of text genre on parent use of joint book reading strategies to promote phonological awareness

This study compared the rates of different types of joint book reading behaviors of parents of preschool children with and without language impairments using two different text genres. Participants included 72 preschool children between the ages of 51 and 70 months. Fifty-five children were typically developing and 17 were diagnosed with language impairments. Parent–child dyads were videotaped during periods of joint book reading using two different types of text (one alphabet-rhyming and one narrative). Results indicated parents rarely used reading behaviors known to enhance their preschoolers’ phonological awareness. However, when used, there was a significant difference between the two types of texts. The alphabet book elicited a higher rate of phonological awareness and print concept behaviors, while the narrative book resulted in parents using more content behaviors. In addition, the parents of typically developing children used more phonological awareness behaviors than the parents of children with language impairments.     

一脉相承，各放异彩——从三个文本的比较看互文性和独创性的统一

《呼啸山庄》、《黑暗的心脏》和《了不起的盖茨比》这三部不同时代的著名小说在叙事手法上有着相似之处:它们都采用了相似的故事内容,"故事中的故事"的叙事结构,多重叙述者,第一人称叙事视角,以及二元对称结构。但它们在叙事手法上也存在着细节上的不同,而且在故事表现的主题等方面也各不相同。所以这三个经典的叙事文本既相似又相异,它们集中体现了互文性和独创性的统一。     

西文文献数据库的局限性与进一步的思考

本文以检索这一环节为解剖点,通过对大量实例的调研,分析了一个西文文献型数据库的标引等各环节的缺陷及它们造成的总体局限性,并就数据库建设提出了建议。     

基于文本挖掘和可视化技术的主题自动标引方法——以《英雄格萨尔》为例

[目的/意义]基于文本挖掘技术自动发现更具代表性的文献内容主题词，通过定位主题词在章节中的具体位置，并基于可视化技术进行主题标引，帮助读者直观高效发现文献主题间的潜在关系。[方法/过程]基于文本挖掘技术深入文献内容层挖掘主题词，并利用可视化工具直观呈现所获信息，在此基础上尝试构建可视化主题自动标引系统，并在格萨尔领域的多个主题中对该系统的自动标引效果进行验证。[结果/结论]研究结果显示，该标引方法在格萨尔领域实现了文献内容级的可视化主题自动标引，快速精准地定位到章节、段落和句子。标引相关信息获取过程直观可视，并且具有交互性，可提升用户体验和参与度。文章以《英雄格萨尔》为例完成系统验证，但该标引方法技术本身无领域限定，可应用于其他领域的文献。     

因特网经济学未登录词计算机辅助挖掘试验

近年来随着社会经济科技和因特网的迅速发展,文献中不断涌现出大量未登录词。未登录词的存在严重影响了汉语自动分词与自动标引的准确率和速率。本文对1000篇经济类网页的关键部位———题名、摘要、关键词、首段进行未登录词挖掘试验,侧重对未登录词挖掘步骤设计和处理方法的讨论。     

单汉字标引方法的改进研究

本文根据信息论中的交互信息,给出了相邻汉字相关度的测量方法,在此基础上提出了基于字串预分割的单汉字标引检索方法,对当前具有代表性的单汉字标引方法进行了改进研究。试验证明本文提出的方法具有较好的性能     

第一部音序检字法字书——论《龙龛手镜》的编排体例

历来对《龙龛手镜》的编排体例褒贬不一。它是第一部音序检字法字书 ,偏旁归部按敦煌写本而定 ,正俗说体依当时的实际用法为标准。