International Journal of Science, Engineering and TechnologyAuthors: Kevin K. Karanja, Andrew Kipkebut
Abstract: Abstract- End-users and institutional buyers consistently purchase sub-optimal computer devices due to a critical technical comprehension barrier and reliance on biased, product-centric advice. This inefficiency leads to the prodigality of device specification wasted financial resources on incompatible or over-specified hardware. This study addresses the research gap by developing and validating a hyper-personalization model that translates non-technical user needs into precise hardware specifications. A mixed-method design (design science research) was employed, starting with an empirical survey (n=32) that confirmed 84% of users struggle with technical metrics. The solution a hybrid conversational model powered by BERT-based Natural Language Processing (NLP) was developed and tested. Validation demonstrated high Accuracy (91%) in matching user intent to specifications, alongside high user acceptance for transparency and ease of use. The model provides a non-biased, effective solution, significantly enhancing user satisfaction and resource utilization in the device procurement lifecycle.
