Towards Natural Language Processing: A Well-Formed Substring Table Approach to Understanding Garden Path Sentence
Jia‐Li Du, Pingfang Yu · 2010
As computers have become more affordable and accessible, the theories and techniques of natural language processing (NLP) are increasingly used as a means for automatically decoding natural language. Well-formed substring table (WFST) is an efficient parsing algorithm used to decode natural language. The form of (START, FINISH, LABEL→FOUND. TO FIND) is accepted by system as its basic model, and its obvious decoding of natural language can be shown by chart parsing. GP sentence has a special syntactic structure which can bring processing breakdown and backtracking during the automatic decoding. By analyzing the GP sentence on the basis of the techniques of WFST, we reach the conclusions that WFST method is helpful for readers to understand GP sentence which has a complex syntactic structure, that it is necessary for an effective NLP system to have access to syntactic, cognitive and semantic knowledge, and that the integration between machine techniques and human being cognition is essential for system to decode natural language.