Research on the Automatic Summarization Model based on Genetic Algorithm and Mathematical Regression

This work presents a type of method to process automatic summarization. And the method is a kind of trainable summarizer, in which the several characteristics considered such as sentence position, positive keyword, the center of negative keyword, title with similar sentence, sentence included in name entity, sentence included in numerical data, relative length of sentence, the comparability of the sentence and the aggregation with bushy path are the abstract each sentence generated. Firstly, the effect of the each sentence characteristic on the task is investigated. Then all the graded functions will be used to generate the modes of genetic algorithm (GA) and mathematical regression (MR), and to obtain a suitable combining characteristic weight. The proposed method is the thing to measure the 100 English religious articles composed by the database on several compressibility and the design method the result presented is promising.