| The information of history evolution is something that includes the process of changing or evolving of some concepts. In some domains that comprise of intricate or quick-changing concepts, such as the software engineering, the computer technology, it is very important to extract the information of history evolution and construct the database of such information. These information is very useful for acquirement, inference, and reuse of knowledge. These information also can be of great value of application for many other domains. At the same time, Wikipedia, as a very normal, standardized, and open network encyclopedia, is becoming more and more important for researchers.Aimed at the information extraction of history evolution, this paper proposes two method to solve this problem. Firstly, this paper introduces the method of hybrid convolution tree kernel, which focuses on the extraction of the relation of concepts that is related to the information of history evolution. This method extracts the verb-entity relation through the calculation of hybrid convolution tree kernel based on the construction of syntactic parsing tree. This method is very intensive, but susceptible to the standardization of the sentence. The other method merges the PAM model and topical TextRank. This method regards the information of history evolution as a special topic, and models the text to history evolution topic. According to the model of history evolution topic, this method extracts the words and entities which is much related with the history evolution topic, and combines these words and entities to the information of history evolution. This method is not restricted to the standardization of the sentence, and very suitable for the network text. These two methods extract the information of history evolution from different ways and have both advantages and disadvantages. But all of them is efficient. These two method also have successful experiences for the deeper research. |
PDF Full Text Request