An Improved Model for the Evaluation of Search Engine Result Pages Using Hybridized Evaluation Techniques

Abstract

In this paper, the hybridised ranking system was developed by utilising WPR to evaluate the structural importance of web pages by assigning values to both the incoming and outgoing links and also other web pages it is connected to while incorporating CTR metrics to account for user activities and TF-IDF to reduce the weight of commonly used words by assigning lower values to words that appear in many documents hence combining both to measure content relevance.  The hybridised approach is compared with the standard PageRank (PR) using the sum of WPR, CTR and TF-IDF metrics for position ranking with 50 randomly selected query datasets which were ranked into less relevant (LR) and most relevant (MR) positions.  The hybridised approach reduced irrelevant pages by reshuffling the less relevant pages to the bottom of the list of web pages found while moving the more relevant pages to the top for the given queries, thereby improving the ranking efficiency of the search engine result pages. In Table 4.1, the process of ranking the 1^st, 2^nd,7^th and 8^th document for the PageRank is moved to the 23^rd 24^th 25^th and 26^th position based on their relevance to the queried data while the 3^rd, 4^th,5^th and 6th document maintained a high position based on their relevance. In Table 4.2, the 1^st 2^nd 3^rd and 4^th documents were reshuffled to the 7^th 8^th 9^th and 10^th positions while the 6^th 7^th and 8^th documentsmaintained a high-ranked position due to their relevance.This study has provided a scalable framework that enhances user experience by minimising irrelevant search results and prioritising pages of higher relevance.