Research of Attribute Value Rough Equality Based-on the Hopfield Neural Network and Rough Set Theory

The problem of Attribute Value Rough Equality (AVRE) is a fundamental problem in the fields of Text Classification and Information Retrieval. However,challenge still exists. In practical application, the processed attribute values are often data/information set based on semantics. This situation is very difficult to be handled by the traditional theories and methods.To address the challenge, this paper proposes the Hopfield Neural Network Architecture (AVRE-HNNA) to solve the problem of attribute value rough equality by combining the Rough Set Theory with the Classification Mechanism of Neural Network. The neural network model (HNNM), the energy function,and learning algorithm are also presented in AVREHNNA.The HNNM is illustrated, evaluated and analyzed through using a simulation instance. The analyses show that the proposed AVRE-HNNA performed better in robustness and stability, and had higher matching precision and field independency. The corresponding algorithms had lower complexity than the traditional methods. The proposed AVRE-HNNA can well solve AVRE problem.