Time Words and Their Regular Representation

Time words are a kind of complex words with their own unique rules of word formation. The number of time words in the open source word vector corpus is large, the structure is complex, and there is a significant long-tail phenomenon, that is, with the increase of word segmentation materials, the number of time words also increases simultaneously, and more low-frequency words appear, which gives the word vector Representation and application bring difficulties. In order to obtain a better representation of time words, this paper investigates the compositional characteristics of time words in open source word vector data, and proposes a rule-based method to classify time words. The rules can be used to analyze the cause of the difficulty in representing time words. In the above, a new time word representation method is proposed, which has the potential to improve time word representation and application difficulties.