Skip to main content

IBM Research - Haifa

Workforce Management

Customer

Our customer is the IBM Global Business Services (GBS), a worldwide service organization with roughly 150,000 service professionals.

Challenge

Matching highly skilled professionals to open positions is a high-stakes task that requires careful consideration by experienced resource managers. Poor decisions can result in understaffing, under-qualification, or over-qualification of assigned personnel, and high turnover of poorly matched workers. While the importance of quality matching is clear, promptly dealing with pools of hundreds of open positions and resources in a dynamic market is a serious challenge. Careful deployment and development of individuals is essential for boosting productivity in today's marketplace.

Solution

By applying constraint programming, a practical scientific discipline on the border of artificial intelligence and optimization, Optimatch successfully provides prioritization lists and near-optimal assignments that take into account all resources and positions in the pool, as well as the complex constraints defining a good match. This output helps resource deployment managers (RDM's) implement better assignments. Optimatch has been successfully piloted since 2005 and is now widely deployed by IBM's Global Business Services (GBS) in all geographies.

Optimatch's main inputs are lists of available professionals and open positions, and a set of rules and prioritizations defined by the user (see figure below). The more prevalent rules are provided explicitly in the user interface as seen in the figure below. In addition, users can define their own custom rules using a straightforward rules-definition language.



Given a list of professionals, open positions, matching rules, and prioritization rules, Optimatch provides two outputs:

  1. Prioritization list – provides prioritized lists of feasible matches for both open positions and professionals.
  2. Complete assignment – an explicit near-optimal assignment, attempting to staff as many positions as possible, while enforcing all relevant constraints, including constraints on the availability, qualities, and preferences of the professional.

Internally, Optimatch builds a constraint satisfaction problem using the following considerations:

  1. The user's matching rules on job role, skill level, geographical location, language, and more.
  2. A score for each possible match provided by a text-matching tool. The text matching tool checks the quality of each possible match by analyzing the professional's CV and the open-position description.
  3. The user's prioritization rules that describe, in a general way, what constitutes a best match between professionals and open positions.
  4. The professionals' availabilities and the open position start and end dates.

Optimatch then uses GEC to produce a near-optimal solution to this problem.

Achievements

Optimatch helps IBM's services organizations in the following ways:

  • Provides professionals with a list of open positions relevant to their qualifications.
  • Reduces the overall time required for the assignment process, while improving quality and quantity of assignments.
  • Helps deployment managers focus on possible matches for their open positions/professionals, with minimal manual search in databases. Optimatch has proven successful in finding matches that were missed using manual search techniques.
  • Further benefits are available to services organizations that wish to run 'what if' analyses for special training programs or new service models.

Quantitative numerical analysis has shown that by applying the Optimatch assignment scheme, up to 15 percent more matches can be found compared to a semi-automated first-come-first-serve scheme. In addition, Optimatch results show an increase in the average quality of the matches.

Optimatch has been successfully deployed worldwide for more than two years within GBS. Optimatch reports are generated daily and sent to the professionals and to deployment managers. By using Optimatch listings, many previously unmet assignments are now being filled by quality professionals. Feedback from professionals and deployment managers alike has been highly positive.

Optimatch is a joint work between the Constraint Satisfaction team and the Mathematical Sciences department at Watson Research Center. Optimatch has been recognized as an IBM Research Accomplishment for 2008, and an IBM Outstanding Research Accomplishment for 2010.

Contact: Yael Ben-Haim (yaelbh@il.ibm.com)