Mixing method for determining keyword context window

Abstract

A mixing method for determining a keyword context window is provided. An initializing statistical method module conduct normalization curve process through solving weight contribution values of each position of contexts adjacent to a keyword C input by a user and then calculates a left and right window n value of a context window in a context of the keyword C; and a left and right window value of the context window can be calculated according to the n value. Valuable quantified description is provided for determination of context effective range, and deficiencies of previous subjective description can be overcome; a relative position of description capacity to the keyword of the context is gradually decreased from the near to the distant, meeting normal knowledge of people; valuable theory foundation is laid for calculation of phrase and sentence similarity; greater weight contribution value linearity and signal to noise ratio is achieved and post calculation is facilitated; high weight contribution value normalization curve accuracy is achieved; and effects on context effective window definition by sentence constituent relationships of the keyword left and right window are considered.

Description

technical field [0001] The invention relates to the technical field of semantic network, in particular to a hybrid method for determining a keyword context window. Background technique [0002] Since entering the 21st century, the global Internet industry has entered a new period of rapid development, and various new technologies have emerged continuously. Natural language processing, an important technology connecting computers and people, has also made great progress. The calculation methods of semantic similarity of words at home and abroad can be roughly divided into two categories: first, the calculation method of semantic similarity of words based on semantic dictionary, this method is simple, effective and easy to understand, but it relies on a relatively complete A large-scale semantic dictionary organized by structural hierarchy; second, a corpus-based word semantic similarity calculation method, which uses a large-scale corpus and uses the context information of w...

AI Technical Summary

Problems solved by technology

The context of words is the resource and basis for acquiring natural language knowledge in corpus linguistics and solving various practical application problems in natural language processing. In the specific application problem, the final result accuracy rate defines the insufficient scope of the context and the influence of the sentence component relationship in the left and right windows of the keyword on the definition of the effective window of the context. The present invention provides a hybrid method for determining the context window of the keyword

Images