Query Expansion Information Retrieval using Customized Ontology Technique

Abstract

Information from online archives is now much more widely used and accessible than before. As a result, searching becomes more challenging and time-consuming. This vast data utilization is the focus of a significant area of research called information retrieval (IR) systems. The goal is to reduce retrieval time while also maintaining and improving answer relevancy. To solve the issues stated, it is necessary to provide an IR Model. The provided method takes care of indexing, similarity keyword extraction, semantic similarity, updating historical data, and updating. The effectiveness of the suggested strategy and the current methods are compared in terms of performance with distinguished parameters. A novel similarity estimate approach is used to group the texts and determine how similar works are based on the obtained score. It is compared to the existing methods to evaluate the findings and shows the usefulness of the suggested model on the scales of accuracy, mean absolute error, precision, recall, sensitivity, and specificity. The improvement of personalization performance, which is an update based on historical knowledge, is one of the main goals of this effort. The Impact Score Estimation technique is used to improve data extraction using semantic keyword extraction and indexing. To cluster documents based on computed scores, the algorithm is to evaluate similarity estimation which can improve searching by speeding up information retrieval and processing. Decision tree classifiers provide better results for class 3 which is 0.93. While the micro average ROC curve generates 0.87 accuracy.