Page 151 - JOURNAL OF LIBRARY SCIENCE IN CHINA 2018 Vol. 42
P. 151

150 Journal of Library Science in China, Vol. 8, 2016


            3.1  Method definition description


            The definition of a method is an accurate and concise explanation of a methods feature or its
            connotation as well as extension. The description for the definition of a method usually has rules
            as follows, “method is a kind of ... method”,”... method is, refers to, is defined as, the definition of
            the method is”. We use these rules to recognize definition description. Although some sentences
            do follow the rules, they may not be about definition. These sentences are called pseudo-definition
            sentences. For example, “local analysis is computationally efficient and does not rely on external
            resources, it is an effective query expansion method” refers to features of a method. While “human-
            competitive intelligence network dynamic analysis is in nature a network analysis method” refers
            to the type of a method. Table 1 gives rules of method definition description and examples.


            Table1. Rules of method definition description and examples
                  Rule                                   Example
                             共词分析也称词语共现分析方法,是一种分析同一个文本主体中不同词语对共同出现的
                             形式,并以此来定量的衡量不同词语之间的关联关系
                             Co-word analysis is also called co-occurrence analysis. It is a method that analyzes the co-
                             occurrence of different words in the same text body and uses this information to quantitatively
             …|法|方法+是一       measure the association relation between different words
             种…的方法。
             *** is a method that   交叉共现算法DLG算法(direct large itemset generation)是一种基于图的关联规则挖掘算
             ***             法,算法通过扫描一次数据集构建项关联图并为所有项建立位向量,然后对位向量进行
                             正交运算即得到关联项
                             Cross co-occurrence DLG algorithm is a graph-based association rule mining algorithm. The
                             algorithm reads the dataset, forms item association graph and creates bit vectors for each item.
                             Then orthogonal computation is used on bit vectors to get associative items
                             跨语言信息检索中,我们将翻译概率的计算方法定义如下:P(fi|e)=F(fi)/(F(fi)+F(fi))  在该
                             公式中,P(fi|e)表示fi为检索词e的翻译的概率;F(fi)为在一定的样本空间中, fi为检索词e
             …方法+是|是指|指      的翻译的频率;F(fi)为在一定的样本空间中,检索词e的翻译不为fi的频率,即检索词e翻
             |的定义为|被定义       译为其他译项的频率
             为|定义如下…         In cross-language IR, we define translation probability as follows, P(fi|e)=F(fi)/(F(fi)+F(fi)). In
             *** method is,   this formula, P(fi|e) is the translation probability of fi given e. F(fi) is the frequency of fi being
             refers, is defined as
                             the translation of query word e; F(fi) is the frequency of fi notbeing the translation of query word e,
                             that is the frequency of query word e being the translation of other words


            3.2  Method relation description


            Relation descriptions between methods represent a-kind-of relations, which is the foundation of
            constructing a method system. From a space distribution’s perspective, relations include hyponyms,
            hypernyms and sibling relations. Different methods are on different levels and positions. From
            a time distribution’s perspective, relations include improvement, inheritance, evolvement and
            substitution.

            3.2.1  Static relations between methods
            Static relations between methods describe the types of methods or attributes of methods. We can
   146   147   148   149   150   151   152   153   154   155   156