Page 156 - JOURNAL OF LIBRARY SCIENCE IN CHINA 2018 Vol. 42
P. 156

HUA Bolin / Types and description rules of knowledge elements about methods in academic papers  155


                                                                                   Continued
                     Rule                                  Example
                …表明,…方法 实验表明,用SVM与修正的KNN组合算法进行专有名词抽取比单一SVM方法以及传统的
                [在…方面]+具有 SVM-KNN方法更具优越性,而且这种方法可以推广到其他非平衡分布样本的分类问题
                更…的…          The result shows that compared with SVM and traditional SVM-KNN, a composite method of
                ***shows that ***  SVM and amended KNN is superior in extracting nouns. And the method can be further expanded
                is superior over *** to other non-balance distribution sample classification problem

               3.3.2  Method feature comparison sentence structure
               The method feature knowledge element sentence includes description, one-to-one comparison,
               one-to-many comparison. Rules and examples are shown in Table 6. To understand and discover
               the types and rules of comparison sentences will benefit the extraction and mining of method
               knowledge elements. Description sentences are sentences that describe features of a certain method
               without comparing with other methods. One-to-one comparison usually has a certain method term
               to be compared. In the sentence there exist obvious feature words like “compared with..., ... is”


               Table 6. Method feature description rules and examples
                  Type       Rule                              Example
                         …方法…具有一 该方法对军事领域的情报分析或是企业管理领域的决策信息处理均具有一定的适
                         定的…         用性和参考意义
                         The method has  The method has certain applicability and reference meaning to the intelligence analysis in
                Pure     ***         military and decision information processing in enterprise management
                description …方法…适用 本文提出的新的相似度测量方法概念明确,逻辑严谨,计算过程相对简单,特别
                         于…          适用于混合型多准则决策
                         The method app- The similarity measure proposed in this paper has a clear definition, rigorous logic and
                         lies to…    an easy computation. The method applies to hybrid Multi-criteria decision making
                                     与需要不断迭代的FlokRank方法相比,基于网络结构的推荐方法计算复杂性要小得多
                                     Compared with FlokRank, recommendation based on network structure has a small
                                     computation complexity
                         与…相比,…方 该算法可以提高多值属性关联规则挖掘效率,与Maqa算法相比,降低了时间复杂
                         法…          度,为处理海量数据、生成有用的知识和信息、进行知识管理及决策提供一种有
                         Compared with  效的方法
                One-to-  ***, *** is   This algorithm can improve the efficiency of multi-attribute association rule mining.
                one                  Compared with Maqa algorithm, this method has a lower time complexity, and is more
                                     effective to process a large amount of data, generates useful knowledge and information,
                                     and serves for knowledge management and decision support
                                     相比SVM方法,决策树方法能够学习析取规则表达式,这些规则可以客观反映描
                         相比…,方法…
                         Compared with   述词的多种属性的关系
                                     Compared with SVM, thedecision tree can extract rule expression which describes
                         *** , *** is
                                     various relationships between description words
                                     与其他方法相比,MCLP更加简洁实用,它允许用户输入不同的参数对模型进行
                         与其他…相 调整,具有更好地灵活性
                         比,…方法 Compared with other methods, MCLP is more simple, practical and flexible and allows
                         +[在…方面]+更 users to input different parameters in order to tune the model
                One-to-
                many     加…          与其他语义相关性度量方法相比,本文方法没有涉及繁重的文本处理工作,计算
                         Compared with  量小且效果显著,尤其适合于实时性要求高的信息检索系统
                         ***. The method  Compared with other semantic association measurements, the method proposed in this
                         is more…    paper does not require taxing text processing, the amount of computation is low and suits
                                     IR system that requires real-time response
   151   152   153   154   155   156   157   158   159   160   161