March 2025
Advances in Nonlinear Variational Inequalities
A patent gives the inventors exclusive rights to their innovations, thus stimulating economic growth, technological advancement, and creativity. Valuable patent document retrieval is necessary for establishing novelty and patentability. Boolean search queries are built to perfectly fetch relevant patents but control complexity at the same time. The Brute Force method treats all keyword combinations and is complete but very expensive in terms of computation. Large-scale manual refining is needed for traditional Boolean methods, whereas AI-based approaches usually lack transparency. This work optimizes Boolean logic with beam search that enhances patent search queries by reducing the solution space to the best promising options. Precisely through the refining of terms in combination, Beam Search enhances precision and recall but does not lose its interpretability. Experimental results demonstrate a 0.94923 score in patent retrieval efficiency. Despite the difficulties such as noisy data and complexity in processing, the approach exhibits a consistent improvement in retrieval performance, thus being a potential solution for patent search optimization.
