A project study submitted in partial fulfillment of the requirements for the degree of Professional Master in Data Science and Artificial Intelligence Applications, School of Engineering and Technology

Professional Master in Data Science and Artificial Intelligence Applications - Asian Institute of Technology, 2022

Shortlisting the right candidate is an important job in the recruitment process of recruitment companies. Currently, the current process of this task remains manual and sometimes heavily depends on the skills of the recruiting staff. Firstly, the recruiting staff must carefully read all the information written in the candidate's CV and input them into the recruitment system or also an excel file. Secondly, the recruiting staff rely on a number of criteria in the job requirements and then filter candidates based on their personal experience to find a pull of potential candidates for the next round of interviews. Therefore, a system, Auto-reading CVs for ranking of candidates based on CV Job Match Score, is proposed to automate the previous process with the aim of saving both time and effort for recruitment organizations. In this study, I did some experiments to compare the results with studies LayoutML(Xu et al. 2020; Xu et al. 2020; Huang et al. 2022) and using experience-based input feature sets depending on the placement of text lines and keywords, then construct a rule basis and group information sources using a clustering approach to evaluate the efficiency of information extraction in CVs. Our key findings are that LayoutMLv3 represents the superiority of the current methods that the company is applying. The project can help the company have positive reviews about using LayoutMLv3 for potential projects in the future when our company faces flexible forms.