基于P/T系统的关系词搭配算法

An Algorithm for Collocation of Relation Word Based on P/T System

基于统计模型的搭配强度算法存在诸多不足,构建了一种基于P/T系统的关系词搭配算法.算法首先基于P/T系统建立了关系词搭配的基本模型,并以变迁数量区分了描述单层有标复句的单层型和描述多层有标复句的多层型;随后讨论了多层型的分解,并借助分解的可行性及S_不变性论证了算法的可行性及可靠性;最后以链表形式储存了库所和变迁,并用C语言对算法进行了描述.结论说明P/T系统是构建算法合适的数学模型.

The collocation strength algorithms based on statistical models have many disadvantages, Thus this paper constructed an algorithm for collocation of relation word based on P/T system. The algorithm established a basic model for the collocation of a relation word based on P/T system. Number of transition was then applied to distinguish single-layer model which described simple complex sentence of Chinese from multi-layer model which described multiple complex sentence of Chinese. The decomposition of multi-layer model were further discussed. The feasibility and reliability of this algorithm was verified with the feasibility of the decomposition and invariance of the model. Finally, the algorithm stored the place and transition in the form of a linked list, and was described in C language. The conclusion shows that P/T system is indeed a suitable mathematical model to construct the algorithm.