A new extension of FDOSM based on Pythagorean fuzzy environment for evaluating and benchmarking sign language recognition systems
Mohammed S. Al-Samarraay,Mahmood M. Salih,Mohamed A. Ahmed,A. A. Zaidan,O. S. Albahri,Dragan Pamucar,H. A. AlSattar,A. H. Alamoodi,B. B. Zaidan,Kareem Dawood,A. S. Albahri
DOI: https://doi.org/10.1007/s00521-021-06683-3
2022-01-10
Neural Computing and Applications
Abstract:Many studies have recently developed real-time sign language recognition system (SLRS)-based DataGlove wearable electronic devices for deaf and dumb to assort hand gestures as having an identical meaning in spoken language. An evaluation and benchmarking of these systems are important towards understanding the most suitable for fulfilling all essential requirements. This process falls under the multi-criteria decision-making (MCDM) problem because of different issues, namely, multi-evaluation criteria, criteria importance and data variation. Therefore, the MCDM solution is necessary to solve complex problems. The latest MCDM method called the fuzzy decision by the opinion score method (FDOSM) and its extension are considered the most powerful and suitable methods. However, these methods still suffer from vagueness issues. According to the advantage of Pythagorean fuzzy numbers in solving such issues, this study extended FDOSM into Pythagorean fuzzy set based on the Interactive hybrid arithmetic mean (IHAM) operator (called PFDOSM-IHAM) to evaluate and benchmark effectively the real-time SLRS. The methodology is presented on the basis of the two phases. Firstly, a decision matrix is proposed on the basis of ‘performance evaluation criteria’ and ‘SLRS set’. Secondly, the development of the PFDOSM-IHAM method is provided considering the following two stages: data transformation and processing. The following results are presented. (1) Variations are observed in the individual benchmarking results of real-time SLRS depending on each decision maker. (2) The group benchmarking results indicate that the 29th real-time SLRS was the best, whereas the worst real-time SLRS was attributed to SLRS (6th). (3) In evaluation, the statistical test indicates that the benchmarked systems from PFDOSM-IHAM are undergoing a systematic ranking. (4) Comparative analysis confirmed the efficacy of the proposed PFDOSM-IHAM against of the other well-known MCDM methods running on Pythagorean fuzzy numbers.
computer science, artificial intelligence