Striving to improve information retrieval, this paper introduces a new method for optimal weight assignment in signature generation to minimize false-drop probability, addressing limitations of previous superimposed coding techniques. The optimal scheme considers both uniform and nonuniform occurrence and query frequencies. Unlike the conventional FSB method, this method is still based on hashing. Moreover, it introduces the fixed-weight block (FWB) method, which controls the weight of signatures to a constant, resulting in lower false-drop probability than FSB. FWB is analytically simpler and more efficient in calculating term signature weights. By assigning weights to document terms based on their occurrence and query frequencies, this new approach aims to reduce false-drop probability while improving the efficiency. These advancements offer significant improvement in information retrieval systems, specifically within contexts requiring precision in identifying relevant information.
Due to `journal_title` is related to computers, this paper's topic aligns well with the typical focus of ACM Transactions on Database Systems. By presenting innovative weight assignment techniques for signature generation, the work reflects the journal's commitment to advancing database efficiency and effectiveness, particularly in the field of information retrieval.