Access Log Analysis for Increasing the Number of Applicants Using LCS on a Recruitment Site

摘要

The total number of advertisement publications in job recruitment media in Japan is currently 1.4 million and has been increasing annually [1]. Analysis of trends by medium indicates that job advertisement on Internet recruitment sites is growing significantly. On these sites, publication fees are the main source of income for companies posting Internet recruitment advertisements; thus, it is important for recruitment sites that membership registrants apply for companies. In addition, based on the information that 90% or more of member registrants do not apply and that users who apply once are likely to apply again, we focused on the abandonment of member registrants on the recruitment site in this study. By applying Longest Common Subsequence (LCS) to extract common parts in the access logs of users who applied and those of users who did not apply, we can determine the difference in characteristics of those access patterns. As an experiment, we investigate whether the proposed method is useful and give concrete examples of improvement points of the website based on our analysis.