A MACHINE LEARNING BASED ABSTRACT IDEA FOR SOLVING INEFFICIENCY IN THE RECRUITMENT PROCESS
PROBLEM :
According to my knowledge, the current recruitment process goes like this- The recruiting firm advertises and the aspirants apply for the advertised role/position. Then the actual recruitment process starts wherein the aspirant is given a certain number of tests which may involve a physical on-site interview.
In case of giant companies like Google, it is very difficult to actually pass all levels of an interview and get hired because the each level is rigorous. One needs to have a lot of persistence to actually make it.
Let me list out the three things required because that is where the problem lies- Persistence, Associated skill set, Knowledge of how the process works.
Nowadays, the emphasis of work that an aspirant does to prepare for the process lies almost entirely on developing persistence and getting to know how the process works. All the aspirants do is read books based on how to crack the process and get hired.
But what the recruiting company really needs is the required skill set which makes the whole process inefficient.
So much time and money are spent by both the recruiting company as well as the aspirants wherein the only objective is to prove to the company that the particular aspirant is employable. Yes, the company does test the aspirant for the skill set. But being an aspirant, you can either work on actually improving your skill set or you can work on how you can crack the process with the skill set that you already possess. A typical aspirant would go for the latter option and would probably have higher chances of getting hired based on the current recruitment process.
Therefore, the time and money are essentially spent (by both the recruiter and the aspirant) on something that isn’t needed as an end product for the hired aspirant.
Also, this may mean there are possibilities that someone without the required skill set may somehow manage to get hired just because of the immense knowledge of the recruitment process. In this case, the company (having spent so much time and money) will end up with someone unworthy for the particular position. This will be a major loss to the company.
For example, say the company is recruiting for the position of an android developer wherein the essential skill set is the one which includes the ability to develop apps.
For testing for this skill, the company may have included a particular assignment or say, a series of questions. These questions will most probably be repetitive and the aspirant might have prepared in such a way that he can answer these questions without actually having the skill of developing apps or having intermediate expertise which is not of the level expected by the company.
SOLUTION :
This can be avoided if we have a more sensible recruiting process which relies more on modern aspects such as machine learning and less on the conventional principles.
One possible abstract idea based on machine learning would be :
We will be needing a dataset of the questions/criterion of the tests conducted during the recruitment process and the performance of the employees as input.
In the most general case, let’s say there is a pool of questions which assess whether a particular candidate for a particular skill set.
For every particular skill in the skill set, there has to be a mapped set of questions. Only a number of questions(say 1 or 2) are selected by the machine for each candidate based on the learning that has taken place.
The machine should learn from its shortcomings. For example, if a candidate who had answered the questions goes on to be incapable of performing actions (in the future) which needed the skill that the questions were mapped to. Then the machine should assess these questions and, if found unworthy, should be removed/altered from the pool of questions and possibly, ask for the user of the machine to input more worthy questions (or it could use its intelligence to figure out new questions if that’s possible!).
The machine has to ensure that the questions selected for a particular skill set are not familiar with the candidate. That is, the candidate will have to actually answer the questions by making use of the knowledge related to the skill rather than the knowledge acquired by studying previously asked questions or some other undesirable way.
The machine should continuously learn from the shortcomings, compare the datasets of the recruitment process and the corresponding performance datasets of the candidates hired.
This essentially keeps the recruitment process as efficient as possible.
