Artificial Intelligence Tools for Recruitment

Terry Elliott Products & Technology

As Jobscience’ Chief Data Scientist I have published articles and blogs on process automation that embeds artificial intelligence in our future recruiting and workforce activity management systems.  The purpose of these articles is to show how these tools can be incorporated into our products to address specific customer issues and to promote suggestions.

Jobscience manages large amounts of recruiting and employment data which is attractive to vendors at the forefront of the artificial intelligence field.  The company has special relationships with IBM and Salesforce.  IBM is an investor and we have been a partner with Salesforce since 2007.   In addition to trying to capitalize on developments from IBM’s Watson program and the Salesforce’ Einstein program, we are also working with Amazon and Google and their artificial intelligence programs.

With this background, I would like to briefly describe the “Random Forest” tools that are incorporated in the Salesforce Einstein program.  Hopefully, this will prompt you to think about how these tools might be used with our systems to improve your business management processes.

In order to provide a context without getting bogged down in statistical and machine learning concepts to understand random forests, you need to realize that the end goal is machine learning, where programs apply algorithms that can learn from and make predictions on data through updating models as data inputs are collected.

Random forests use multiple learning algorithms with the goal of producing better predictive performance than from any of the learning algorithm components alone.  They employ a statistical flow-event logic with a series of branches with probabilities assigned to each branch called “decision trees.”  As feedback data changes the classification schemes and weightings (probabilities) the decision tree logic is updated.

Random forests use classification methods, statistical regression analysis and tools like kernel algorithms to identify patterns and predict outcomes.  As data is collected the classification values and the average prediction (regression) of the decision trees are updated (learn).

Scoring candidates would be an example of the application of random forest tools to the recruitment process.  Whereas a corporate recruiter might apply a classification scheme to score candidates based on factors related to their most successful employees in the positions, staffing agencies might apply a classification scheme to score candidates based on factors like ease of assignment and agency profitability.  In both examples, the classification scheme and prediction (score) would be updated (learn) as data continues to be collected.

Although this is a very abbreviated description of the Salesforce Einstein random forest tool, I should note that it takes time and it is costly to set up random forests and it takes data to “move the needle” and adapt the model (learn).  Small and even mid-size employers and agencies, on their own, may not be able to produce enough data so some form of customer cooperation and shared costs may be required to create large enough blind pools of comparable data to get the most benefit from technique.

With these caveats, I would appreciate your application suggestions.  If you are a Jobscience customer, please use the IDEAs community site.  If you are not my email is wte@jobscience.com.