【摘要】 本文通过对已有的向量空间模型的分析与评价,提出一种在向量空间模型中考虑词相依性的一般理论,即将词关系矩阵G的计算转化为对标引词向量生成的欧氏空间中一组正交基的确定。通过这一途径将实际的检索问题转化为符合独立性假设而又事实上没有作出假设的检索问题,并将这一理论用于解释和评价K. M. Wong等人的工作。最后提出了一些新的设想。更多还原

【Abstract】 Through the analysis and evaluation of the traditional vector space model,the author presents a general theory of information retrieval based on term de-pendance in vector space. On the basis of this theory, the computation of a termrelation matrix can be transformed into the determination of the vertical car-tesian produced by the indexing term vector in the Euclidean space. By meansof this approach, a practical retrieval problem can be converted into a retri-eval problem which accords with the hypothesis of independance while in effectit does not make such a hypothesis. This theory is used to interpret andevaluate the work of K. M. Wong, etc. Some new ideas on this problem havealso been presented.更多还原