Design and Application of Intelligent Library Intelligence Authentication System Based on Information Entropy Early Warning Models

Abstract

Abstract Identifying the authenticity of library intelligence has consistently posed a significant challenge in intelligence management. This paper aims to optimize the D-S evidence theory to address the issues of uncertainty, imprecision, and high conflict in library intelligence evidence fusion. To achieve this, DSmT is employed to extend the D-S evidence theory by eliminating the process of assigning conflicting information with normalization coefficients. Then, through the fuzzy information pseudo-feature analysis method, the library intelligent intelligence information entropy feature quantity is extracted, and semantic feature matching is used to construct the library intelligent intelligence true and false information recognition system. Finally, the recognition performance of the system is examined using a dataset to analyze the intelligence authenticity recognition of mis-spliced bacterial colony genomic data from a biological experiment that has been entered into the library intelligence database. Finally, it was found that the average recognition rate of this paper's system in the four classical datasets is 96.57%, which is higher than the 89.61% and 90.02% of the two evidence fusion classification schemes. The results of this paper's algorithm for high-conflict evidence are 0.9506, 0.0558, and 0.0097, reflecting the accuracy of the evidence support results. Finally, 35 colony genome splicing error intelligence were identified, of which 32 were genuine errors, with a precision rate of 91% and a recall rate of 94%. The library intelligence authenticity recognition system designed in this paper has excellent performance and provides an effective path and usable algorithmic model for library intelligence authentication.