As the county does a top-to-bottom rebuild of its Investigation Case Management System (ICMS), it's developing a program that can deepen its understanding of text through the use of artificial intelligence.
The additional AI functions should be ready by the middle of next year, but no vendor has been chosen. Both open-source and proprietary systems are being considered, and a system that can analyze text is the top priority. No RFIs or RFPs have been issued yet.
HR Chief Information Officer Roozan Zarifian told Techwire in an interview Wednesday that OpenText software is being used to analyze specific language in text fields, so the type of case being filed in the complaint system can be more readily identified.
Department leaders want to automate as many business processes as possible anything "that can be regimented using rules," Assistant HR Director Murtaza Masood said in the Techwire interview.
The county is also looking at how to use artificial intelligence over the applicant tracking system on NeoGov software. The department director, Lisa Garrett, is working to automate as many systems within HR as possible. The department just finished analyzing the time-to-hire process with Microsoft BI, its business intelligence system.
The HR leaders hope that analyzing behavior patterns and identifying specific concerns within cases will make it easier to choose appropriate training and interventions before a problem occurs.
"It's our approach to applying lexicon analysis and predictive analysis to time sensitive workforce morale and legal issues," Masood said.
The goal is to deepen that learning over time, making the filing work more automated and allowing people to spend more time analyzing the cases themselves, rather than classifying and routing them.
Zarifian said: "The way AI usually works is, the more data you feed, the better it gets over time. I imagine at the beginning, we will be working closely, monitoring the outcomes and adjusting it. But with time and additional data, it should be a high rate of accuracy."
A proof of concept was completed last year to help categorize employment appeals cases. The OpenText system used AI to route files to about 12 different work paths, where analysts would then complete the process.
The department is also looking to create a benefits bot that employees can reach through the department website.
Editor's note: This story has been updated to correct the title for Razoon Zarifian.
