In the past decade, recommendation systems have witnessed rapid advancements and have found extensive applications across various domains. Nevertheless, as industry require- ments continually shift and diversify, the necessity for personalized recommendation systems has become more pronounced. The dissertation presented here focuses on the evolving land- scape of recommendation systems to address the growing demand for tailored solutions in an ever-expanding era of information. This study delves into the innovative approach of de- signing custom neural networks that adapt to the unique preferences of individual users. In an information-rich environment, one size does not fit all, and the ability to cater to specific user needs is of paramount importance. The objective of this dissertation is to explore novel techniques and methodologies for personalized recommendation systems, in which lots of advanced technologies are involved, such as attention mechanisms, graph neural networks, quantum cognition theory, quantum jump, etc. Finally, the experiments validate the supe- riority of the proposed methods.