Web Page Recommendation using Random Forest with Fire Fly Algorithm in Web Mining

Nowadays, internet has become the easiest way to obtain more information from the web and millions of users search internet to find out the information. The continuous growth of web pages and users interest to search more information about various topics increases the complexity of recommendation. The user's behavior is extracted by using the web mining techniques, which are used in web server log. The main aim of this research study is to identify the navigation pattern of users from the log files. There are three major steps in the web mining process namely pre-processing the data, classification of pattern and users discovery. In recent periods, the web page articles are classified by the researchers before recommending the requested page to users. However, every category size is too large or manual labors are often needed for classification tasks. A high time complexity issues are faced by some existing clustering methods or according to the initial parameters, these techniques provides the iterative computing that leads to insufficient results. To address the above issues, a recommendation for web page is developed by initializing the margin parameters of classification techniques which considers both effectiveness and efficiency. This research work initializes the Random Forest's (RF) margin parameters by using the FireFly Algorithm (FFA) for reducing the processing time to speed up the process. A large volume of user's interest data is processed by these margin parameters, which provides a better recommendation than existing techniques. The experimental results show that RF-FFA method achieved 41.89% accuracy and recall values, when compared with other heuristic algorithms.