Article Preview
TopIntroduction
Who gets to our websites? What are they doing? Where are they from? responses to these questions are recorded in the raw log file called access logs (Atta-ur-Rahman et al., 2019). It provides knowledge required to enhance the efficiency of any Web based applications. Here, Web mining plays a predominant role in discovering patterns from these sources. To discover pattern from the Web, it uses Web mining techniques. Therefore, it is known as automatic discovery of useful patterns from the Web (Zhang, 2015). Here, the process of content discovery and pattern analysis focuses on data accessed by end-users. Therefore, the browsing habits of various users are recorded in the server access log. Many applications are using users’ access data as their primary source of the data for pattern discovery. Hence, the usage data in this research reflects the access log stored on the proxy server, which tracks user navigation information to various websites. Such patterns are used to better understand and meet Web-based system needs (Xiao, 2001; Wang, 2006; Cirqueira et al., 2020).
Sample access logs shown in Figure 1 contain raw data that requires certain kinds of preprocessing prior to any predictive mining activities. Most of the data recorded are meaningless and insufficient for current mining activities to be carried out. So, it requires some preprocessing operations that must be applied before performing mining (Zaiane, 2000; Zhijie, 2009; Radwan et al., 2019). Consequently, the function in Web mining is data preprocessing, which prepares data for patterns analysis.
To uncover hidden patterns, various mining techniques such as association rule mining, clustering, classification and so on can be applied on the cleaned log data (Venketesh and Venkatesan, 2009; Zehua et al., 2018; Hossein and Vimala,2018). The discovered pattern is then useful for applications such as network improvement, website reengineering, business intelligence, etc. The following sections describe comprehensive preprocessing tasks for prediction based recommendation which have been addressed and considered.
Sequences of fields in the log entry line are given below. It is following Common Log File format (CLF) for recording the below details. There are some Web servers having extended file format which is having browser information additionally:
- •
user_id
- •
Product_Url
- •
Product_Name
- •
Product_Category_Tree
- •
Pid
- •
Retail_Price
- •
Discounted_Price
- •
Image
- •
Is_FK_Advantage_Product
- •
Description
- •
Product_Rating
- •
Overall_Rating
- •
Brand
- •
Product_Specifications