Extensible object oriented reasoning information filtering

Abstract

In this paper, we present a new systematic approach of extensible object oriented reasoning information filtering by combining the concept of object oriented, Extenics Theory with reasoning theory seamlessly. In the extensible object oriented reasoning filtering research, a user's filtering profile is constructed as a profile class including the attributes and values of these attributes of the class. The samples and documents to be filtered are defined as the extensible objects instead of ordinary objects. Using the novel extensible object oriented representation, the learning problem of the user's profile is converted to find the optimized attribute values of the profile class in order to produce the desired output. Principle Component Analysis (PCA) is used to select the representative characteristics and help to construct the profile without discarding information that is useful for representing filtering profile. The principle of filtering is to find the related extensible objects that belong to the profile class. The profile learning procedure is fulfilled by using the simplified neural network approach. Simulation experiments are used to show the effectiveness.