Inclusive evaluation and design for HCI interface using backpropagation neural network (BPNN) and artificial bee colony (ABC) algorithm model

Abstract

In an era of pervasive information technology and widespread internet access, Human Computer Interaction (HCI) interfaces play a vital role across various sectors like education, healthcare, and public services. Inclusive design, accommodating diversity in abilities, gender, age, and skills, has become crucial for enhancing user experiences. However, existing HCI interface research faces challenges in meeting diverse user needs and capabilities, including neglecting inclusivity for user groups with reduced abilities and streamlining the design process for inclusiveness. The study identifies several research gaps in the field of HCI interfaces: there is a lack of a standardized and systematic approach for evaluating inclusiveness in HCI interfaces, existing methods often inadequate for inclusivity and the design process of inclusive HCI interfaces, especially for novice designers, who often face challenges for inclusivity in HCI interfaces. Addressing these gaps is crucial for advancing the development of inclusive HCI interfaces. Thus, this research aims to address gaps in HCI interface development by focusing on inclusiveness and user-friendliness. To achieve this, three main objectives are outlined. Firstly, to devise an evaluation method for HCI interface inclusiveness, by introducing a usability metric: usability inclusiveness average value (UIAV) and a user capability metrics: user capability inclusiveness degree (UCID). This method offers both qualitative and quantitative analysis, enabling designers to assess inclusiveness across different interface versions. Secondly, to identify influential inclusive features of HCI interfaces, drawing on cognitive load measurement concepts to develop comprehensive design criteria. Lastly, to propose intelligent algorithms through BPNN and ABC for constructing an inclusive design data source, as a resource for generating guidelines for designing HCI interfaces. This research employs three main methods to contribute to the development of HCI interfaces. Firstly, it introduces evaluation indicators, UCID and UIAV as mentioned earlier which will be fit into two proposed tools, the 'usability user capability function' and the 'usability user capability diagram.' These tools enable the computation of the HCI interface inclusion index and provide a method for interpreting and analysing inclusiveness. Secondly, a cognitive load measurement analysis method is proposed, involving experiments with CNC machines, websites, and smart TVs, to extract influential features of HCI interface inclusiveness. Lastly, a design application method based on BPNN and ABC algorithms is proposed to construct an inclusiveness database to help streamlining the design process while maintaining inclusiveness. Final results include detailed score values e.g., Root Mean Square Error (RMSE) = 0.0323 and R Square = 0.9735 with ABC-BPNN; UCID for Experimental Group (EG) achieved 98.26%, and Control Group (CG) achieved 87.81%, while results of UIAV also show that EG is superior to CG in various indicators, showcasing the effectiveness of proposed methods in enhancing HCI interface inclusiveness. In conclusion, this research addresses the identified research gaps and contributes to the knowledge by proposing inclusive evaluation, inclusive features analysis, and inclusive design methods for highly inclusive HCI interfaces.