Chinese Unknown Word Recognition Based on Functional Applications of Type Theory

We introduce a method based on type functional applications for unknown word recognition in Chinese texts. A major advantage of the method is that it has an ability to detect unknown words of unrestricted characters by calculating the results of type functional applications. The method is built upon the basic type theory. It can take more semantic information while analyzing sentential syntactic structures in this method. In our approach, each word is assigned a type. By type functional applications, union algorithm and type-revivification; we can get the right segmentations of unknown words. It was applied to a newspaper of people daily, achieving 89.96% recall for person names, 77.94% recall for place names and more than 73% recall for other unknown words.