In an era when technology becomes more and more available and computer processing power can be purchased at lower cost, automation by utilizing artificial intelligence in a vast amount of industries has become important to achieve a competitive advantage. Especially in the field of headhunting and recruitment.
For many companies, finding the correct person for their position with the key competence can be the bottleneck for success. In the last decade, there has already been developed a great number of different AI recruitment tools to assist companies in their hiring process. A human capital management research director at Gartner stated that there’s a greater level of maturity in AI tools in the recruiting space than in any other area of HR. Tools as artificial assistants.\citep{zielinski}
The background for this project started when we got contacted by an IT company, looking for assistance around the possibilities for building a recruitment module.They were curious regarding the possibility of gathering candidate data automatically from the most common “Web 2.0 sites”: LinkedIn, Twitter, Facebook etc, and use this data to provide a ranking, based on both personality traits and skills. More detailed, this module based on input parameters such as skills, areas of work, and knowledge, products etc, could present a list that contained the top candidates from the corpus, and a descriptive profile about the candidate. Further discussion brought about a great desire if the candidate profile was able to suggest a strategy for preparing a job offer/interview based on the candidate's interests, and/or personality traits.
Since we have a great interest in web development and artificial intelligence and the combination of both, we saw this as a great opportunity to dive into a very interesting research topic. To accommodate this idea we provided a strategic model for scoping out which sub fields we had to investigate further and research.
Back in 2011 a research group from Greece delivered a paper on a personality mining system for automate the applicant ranking in online recruitment systems. Their approach relied on the linguistic analysis of data that were available for job applicants in order to infer the applicant's’ personality traits and do a ranking on them. This was employed in a web based e-recruitment system. In order to perform a ranking, the team used Analytic Hierarchy Process.\citep{tzimas}
Following up the same year, the research group delivered a paper that discloses a novel approach for recruiting and ranking job applicants in online recruitment systems. The objective was to automate applicant pre-screening. The team used a compiled corpus of 100 applicants that had both a LinkedIn profile and a personal blog. Applicants that did not have a sufficient blog able to describe their personality traits, were excluded from the study. Personality traits were automatically extracted from his/her social presence by using linguistic analysis. The findings from this study showed that the automated pre-screening performed consistently compared to human recruiters, with the exception of people in senior positions. \citep{articlea}
In 2012 the same research team presented an approach for evaluation job applicants in online recruitment systems by leveraging machine learning instead of AHP to solve the candidate ranking problem. They proposed a prototype system that extracted a set of objective criteria from applicants’ LinkedIn profiles and inferred their personality characteristics using the linguistic analysis on their blog posts. The conclusion of the study was that the system performed consistently to what the human recruiters did. \citep{tsakalidis2012}
Based on the study in the paper described above, the research group presented a new paper the year later in 2013. In this paper they also included semantic matching techniques. They proposed a system her that extracted a set of objective criteria from the applicant’s LinkedIn profile as in the previous paper, but now they compared them semantically to the job’s prerequisites. Personality characteristics using the linguistic analysis on the applicants blog posts was also implemented as in the previous paper. \citep{article}
The study above has a strong focus on sorting through existing application to job openings. We propose a system that automatically sources possible candidates from the internet using semantic data extracted from a job description. We will propose some new techniques for ranking the candidates.
Research question proposals:
- RQ 1: Can you find a candidate for a senior position using only the data available on linkedin and other social media?
- RQ 2: Can the above be automated using AI?
- RQ 3: Can the proposed system generate a ranked report that includes both qualification and personality traits based on the questions above?
- RQ 4: Can a report generated from all available online information about a candidate be used to aid HR in recruiting the candidate?
Alternative research questions:
- RQ 1: Can a system use data mining in order to determine a person's personality traits using information available in social media?
- RQ 2: Can a person's personality traits affect how the person does his job?
- RQ 3: Is it possible to produce a list of potential candidates more efficiently by automating the process?
The main product of the research will be a program with the purpose of assisting HR personnel in recruiting for senior positions. An experiment will also be designed and conducted to determine the effectiveness of the program. A product of this research will be the experiment design, the data collected and the answers to our hypotheses. Previous studies in the field have uncovered a gap when it comes to hiring for senior positions, some of our efforts will go into testing if this gap can be covered using the current advances in both AI and knowledge representation.