Development of an Intelligent Job Recommender System for Freelancers using Client's Feedback Classification and Association Rule Mining Techniques

Most of the freelancer's time is killed in finding suitable jobs due to the huge number of freelance marketplaces. Freelancing sites send email notifications or show in newsfeed about posted jobs but most of them are irrelevant. Recommending relevant jobs to freelancers to minimize job finding time has drawn the attraction of researchers. Here, in this paper, we propose a recommender system to find out appropriate jobs for freelancers using client’s feedback classification and Association rule mining techniques. After collecting the previous work history of freelancers, we analyze the sentiment of client's feedback using Logistic Regression and Linear Support Vector Machine model to classify the completed jobs into two categories: positive and negative. We apply the Association rule mining technique to find out freelancer's frequent skillsets used in both categories of completed jobs. Then, we find out the jobs matched with the positive frequent skillsets using set operations. We also discard jobs that contain negative frequent skillsets. Finally, a collaborative filtering algorithm is applied considering the client's overall rating, the minimum budget/ hourly rate, deadline, re-hire, etc. to generate a more accurate recommendation. After extensive experiments on the real dataset collected from different online marketplaces, we are able to prove that our proposed method correctly recommends the appropriate jobs with 83.40% (Logistic Regression) and 84.03% (Linear SVM) accuracy.