Hyperautomate HR processes with Business Entity Recognition: AI Business Services to the rescue

Need for hyperautomation in the recruitment process

A study from Boston Consulting Group titled “Realizing the Value of People Management” highlighted that recruiting and talent management is one of the most important functions of human resources. Another study revealed that 35% of the recruiting and selection process involves going through paper resumes and cover letters, making the process prone to manual errors. Each error in the recruiting process costs time for multiple parties involved including HR, interview panels, recruiting managers, and so on.

Using a combination of Business Entity Recognition and iRPA tools, we aim to automate the resume screening process so that recruiters can use all the relevant information about candidates to make wise decisions about shortlisting or rejecting them.

Information extraction from resumes through AI Business Services

The first and most important step in talent acquisition is shortlisting candidates based on careful screening of their resumes. Studies show that variations in resume formats and content leads to a resume screening time of around 23 hours for just one hire — and in fact, 52% of Talent Acquisition leaders say that the hardest part of recruitment is selecting candidates from a large applicant pool.

SAP’s solution to this tedious yet important process is Business Entity Recognition (BER). BER is a powerful AI Business Service that helps detect and identify defined elements in unstructured text. Combining BER with another AI Business Service, specifically BOCR, allows a candidate’s resume to be scanned and all the relevant information extracted in a matter of seconds. These AI Business Services are available on SAP’s Business Technology Platform.

The BER service comes with a set of simple yet powerful APIs that you can use to create an end-to-end flow for extracting information that is relevant to specific job openings.

BER in action

As part of the Hyperautomation Webinar Series, I presented a guided demo on how to use Business Entity Recognition to create a custom model to train that will ultimately enable a bot to use to extract data elements from a candidate’s resume. Here is a link to the video:

And here are the steps to follow:

1. Create a dataset and upload training documents. These documents must be annotated with the elements that you want to extract.
2. Trigger training using this dataset and provide a model name of your choice.
3. Once the training is completed, a BER model is generated and you can check the model’s accuracy and supported capabilities.
4. Deploy the trained model to use for inference.
5. Use Inference APIs to get predictions about a candidate’s resume.

The following diagram shows the important steps involved in creating a BER model and using it for predictions.

End-to-end flow for screening resumes

You can create a simple application for resume screening. The workflow for this is as follows:

1. Upload the candidate’s resume.
2. Use BOCR to convert PDF or other document content to text.
3. Once text is extracted, the BER service extracts the entities that you defined in your dataset when training the model.
4. Once you have all the information handy, extracted within seconds, you can further integrate it with a tool of your choice to either shortlist or reject a candidate.

Below is a screenshot of such an application.

By using Business Entity Recognition with a combination of iRPA, the recruitment process can be automated and become more efficient. You can try the Business Entity Recognition service on the BTP trial account and achieve optimization for your manual processes.