Method for finding new words on basis of directed and weighted graph

The invention discloses a method for finding new words on the basis of a directed and weighted graph. The method comprises the following steps: adopting a word-classifying source open tool for classifying words for a linguistic data; filtering the words which are not used any more for the word-classifying result; establishing an incidence relation and weight between word and word according to the word-classifying result and generating the directed and weighted graph; adopting an edge-weight threshold value for selecting the edges of the directed and weighted graph; reserving the lexical item collocation with higher co-occurrence frequency in the linguistic data; selecting isolated points and self-loops in the directed and weighted graph and generating a sub-graph; establishing a hypothesis testing model according to the edge weight between the adjacent nodes and the node strength in the sub-graph; selecting possible new words from the sub-graph; selecting the possible new words according to a part-of-speech t