GA on IR

Abstract

Increasing the growth rates of websites’ number has led to the challenge of assisting Web customers in finding appropriate details from the Internet using an intelligent search engine. Information retrieval (IR) is an essential and useful strategy for Web users; thus, different strategies and techniques are designed for such purpose. Currently, the focus on the usefulness of Artificial Intelligence (AI) has been improved with IR. One AI area is Evolutionary Computation (EC), which is based on designs of natural selection. A traditional and important strategy in EC is Genetic Algorithm (GA); this paper adopts the GA technique to enhance the retrieval of HTML documents. This improvement is obtained by creating a modern evaluation function and applying a hybrid crossover operator. The proposed evaluation function is based on term proximity, keyword probability within the document, and HTML tag weight query. Experimental results are compared with two well known evaluation function functions applied in IR domain which are Okapi-BM25 and Bayesian interface network model. The results demonstrate a good level of enhancement to the recall and precision. In addition, the documents retrieved by the proposed system were more accurate and relevant to the queries than that retrieved by other models.