| Automatic document summarization is a technology extract and abstract the key information from the original document via a series of processing of computer. Through automatic abstracting, users can effectively reach the information they need from the mass of information in a short period of time. At present, how to get the key information from the mass and long articles has become the hot issue for the researchers from the academic and the industry.In this paper, we focus on the issues of doucment extract and abstract, such as the poor contact between the abstracted content and the original content, curt incoherent and poor readability, etc. Due to solve the problem of document extract, we proposed an abstract automatic post processing algorithm based on eliminating redundant rules. The algorithm can extract the abstract aims to further polish, so that it has been further improved in the sentence is smooth, logical structure, abstract terms. The main work and innovations of this paper are as follows:Firstly, the sentence structure optimization algorithm will make the language of abstract not so bumpy or straggling in sentence combination, simplification and reference.Secondly, we will create the contrast sentence marking system in sentence weighting and improve the previous sentence weighting marking system so that the improved algorithm can do the graded abstract for the contrast sentence.Finally, we will use the redundancy algorithm to make the abstract sentence sort-able in the original order in generating and processing of the abstract. It will solve the problems of sentence perversion and unreasonable logic so that the abstract can be concise and smooth. Also, the abstract can conform to main idea of the original text and close to the nature language at the same time.The experimental results show that, the accuracy and integrity of the abstract has a certain improvement after using the method in this paper which outperforms the previously existing algorithms. |
PDF Full Text Request