A New Method to Extract Text from Natural Scenes

This paper presents a new method for text detection, location and binarization from natural scenes. Several morphological steps are used to detect the general position of the text, including English, Chinese and Japanese characters. Next bonnding boxes are processed by a new “Expand, Break and Merge” （EBM） method to get the precise text areas. Finally, text is binarized by a hybrid method based on Otsu and Niblack. This new approach can extract different kinds of text from complicated natural scenes. It is insensitive to noise, distortedness, and text orientation. It also has good performance on extracting texts in various sizes.     

英语松散句和掉尾句

对英语松散句和掉尾句从定义、类型及修辞特点等方面进行了比较，探讨了其相互间的转换以及各自的语篇功能。     

基于主题发现的舆情分析系统的设计与实现

开发了一套由中文分词、向量空间模型建立(VSM)、特征提取、降维处理和文本聚类等模块组成的舆情分析系统,实现互联网舆情信息的主题发现;在舆情分析系统的实现过程中,提出了一种K-means改进算法。实验证明,改进后的算法在查全率和查准率上都有了一定的提高,能够得到较高且稳定的准确率。     

Reverse engineering of compact suffix trees and links: A novel algorithm

Invented in the 1970s, the Suffix Tree (ST) is a data structure that indexes all substrings of a text in linear space. Although more space demanding than other indexes, the ST remains likely an inspiring index because it represents substrings in a hierarchical tree structure. Along time, STs have acquired a central position in text algorithmics with myriad of algorithms and applications to for instance motif discovery, biological sequence comparison, or text compression. It is well known that different words can lead to the same suffix tree structure with different labels. Moreover, the properties of STs prevent all tree structures from being STs. Even the suffix links, which play a key role in efficient construction algorithms and many applications, are not sufficient to discriminate the suffix trees of distinct words. The question of recognising which trees can be STs has been raised and termed Reverse Engineering on STs. For the case where a tree is given with potential suffix links, a seminal work provides a linear time solution only for binary alphabets. Here, we also investigate the Reverse Engineering problem on ST with links and exhibit a novel approach and algorithm. Hopefully, this new suffix tree characterisation makes up a valuable step towards a better understanding of suffix tree combinatorics.     

Exploiting the performance of dictionary-based bio-entity name recognition in biomedical literature

Bio-entity name recognition is the key step for information extraction from biomedical literature. This paper presents a dictionary-based bio-entity name recognition approach. The approach expands the bio-entity name dictionary via the Abbreviation Definitions identifying algorithm, improves the recall rate through the improved edit distance algorithm and adopts some post-processing methods including Pre-keyword and Post-keyword expansion, Part of Speech expansion, merge of adjacent bio-entity names and the exploitation of the contextual cues to further improve the performance. Experiment results show that with this approach even an internal dictionary-based system could achieve a fairly good performance.     

发挥电脑的声色优势,激发学生的学习兴趣

学生在学习程序设计语言时,最感兴趣的话题之一就是如何让计算机发声、如何控制文本的输出颜色;然而目前常用的几种程序设计语言教材几乎都没有涉及这方面的内容。为此,本文总结并探讨了在BASIC、PASCAL、C和FoxPro这四种语言中让计算机发声、奏乐以及输出彩色文本的方法及应用实例。     

Photochemistry of Organic Compounds.From Concepts to Practice. Postgraduate Chemistry Series. By Petr Klán and Jakob Wirz.

John Wiley & Sons, Hoboken 2009. 582 pp., softcover € 49.90.—ISBN 978‐1405161732

     

基于主题词权重和句子特征的自动文摘

为获得高质量的自动文摘,在组合词识别算法的基础上,充分考虑词的频率、词性、词的位置、词长等因素,构建了一个词语权重计算公式,该公式能使表达主题的词和短语具有较高的权重.对句子权重的计算,则考虑了句子的内容、位置以及线索词的作用和用户偏好等.摘要的生成充分考虑了候选文摘句的相似性,避免了冗余信息的加入.对摘要的评估进行了从句子粒度到词语粒度的改进,提出了一种基于词语粒度的准确率和召回率计算方法.实验证明,该算法生成的自动文摘有着较高的质量,平均准确率达到77.1%.