TOPSIS fuzzy inference based method for evaluating the level of intelligence

DOI:	10.11809/bqzbgcxb2023.01.029
Keywords:	intelligent system; intelligence; assessment method; fuzzy inference; TOPSIS
Abstract:	In order to improve the shortcomings of the existing intelligent level evaluation methods for intelligent systems, including coarse grained evaluation, poor generality, inexplicability, etc., this paper proposes a system intelligence level evaluation method based on deep TOPSIS with FIS. This method proposes a general intelligence level evaluation index system and builds a deep TOPSIS based on the index system. The model firstly generates a preliminary comprehensive evaluation based on fine grained indicators of multiple dimensions, then uses an ANFIS to perform fuzzy calculations, and finally outputs the intelligence level. The model adopts an evaluation strategy combining qualitative evaluation and quantitative evaluation with an integration of expert knowledge and fuzzy inference, which makes the evaluation results interpretable and the model strongly generalized. For different information systems, this paper conducts experiments based on the proposed intelligence level evaluation index system, relevant data, and expert knowledge. The results show that the method can stably evaluate the intelligence level, the evaluation results are interpretable and the evaluation model has good generality. The proposed method can provide solutions and references for the evaluation of the intelligence level of big data and artificial intelligence systems.
Published:	2023-01-28
PDF HTML