In the past, Forum Engineering’s consultants manually matched engineers to jobs by evaluating open positions and interviewing candidates, but this labor-intensive process was limiting the company’s growth.
The company introduced a solution to collect and analyze massive amounts of structured and unstructured data from both internal and external sources. By using this data to automate and optimize the matching process, it enables faster, more accurate job placement services to its clients.
83% increasein matching efficiency, greatly reducing the number of failed recruitment and placement attempts
Strengthensreputation and credibility in the staffing industry, attracting new customers
Acceleratesthe placement process, helping to outperform competitors and win market share
Business challenge story
Seeking an efficient and accurate matching process
Forum Engineering provides human resources and recruitment services for engineering-related industries such as automotive, industrial machinery, electrical and electronic products, precision equipment, communications and IT. Its main focus is to match skilled personnel to open positions at Japanese manufacturing companies.
Human resources has traditionally been a labor-intensive industry. Forum Engineering fills 14,000 engineering positions every year. But the matchmaking process was slow, subjective work, and their manual personnel matching processes were limiting the company’s growth, due to the heavy workload they created for their recruitment consultants.
The staffing specialists had to evaluate candidates by searching through large numbers of résumés, interview notes, customer feedback, and other materials. In addition, Japanese laws forbid companies from obtaining certain confidential information about potential contractors, so Forum Engineering spent a lot of time clarifying requirements and trying to find the best fit—not only in terms of skills and experience, but also in regard to personality and performance under different conditions.
However, there is only so much information a human being can synthesize in a short period of time. Despite their best efforts, staffing specialists sometimes missed the mark, resulting in early contract termination, dissatisfied customers, and openings for competitors to steal the business. The company needed to increase the speed and accuracy of its placement services, keep up with high demand for engineers in the manufacturing industry, and establish its leadership in the technical staffing sector.
Harnessing big data to automate and optimize matching
To solve these problems, Forum Engineering worked with IBM to create a united data strategy, and has developed two new initiatives—“Insight Matching” and “Cognitive Staffing”—to fundamentally change the way they run their business. The IBM Watson® powered solution collects and analyzes enormous amounts of data from both internal and external data sources, and enables a more intelligent matching process based on predictive analysis, reasoning, and mathematical optimization.
In addition to structured data such as length of service, qualifications and technical proficiencies, the solution also draws on unstructured data such as résumés, job history, activity reports, interview notes, and feedback from customers concerning individual engineers.
The solution uses advanced search and natural language processing technologies, which allow consultants to identify optimal candidates by simply typing questions or keywords into the system. For each candidate, the solution can also provide a matching score, which assesses how closely the engineer matches the requirements of the job.
Major applications within the solution are the "IM Dictionary" which helps to visualize corporate personnel requirements and engineers’ skills; "SOEFIA", which assesses the market value of engineers; "SIDUS", which analyzes engineers’ personalities; "COMPAS", which matches the engineers to job positions; and "COMMUNICATION" which handles the natural language interface.
On a technical level, IBM® Watson Explorer Analytical Components is used for morphological and other analysis of natural language, while IBM SPSS® Modeler and IBM SPSS Collaboration and Deployment Services are used for scoring. The solution also uses the IBM ILOG® CPLEX® mathematical optimization engine to power a "negotiation conversation" mechanism, which proposes optimal solutions to maximize revenues when the company needs to make a trade-off between pricing and utilization rates.
The IM Dictionary is one of the most important applications within the Insight Matching solution, because it provides the basic information that is used in the evaluation and matching processes. It uses a tree structure to visualize the relationships between words that relate to engineering concepts such as products/parts, technologies, jobs and skills. By clicking on a product in the tree, for example, a user can see what kinds of technologies, jobs and academic knowledge are needed to develop that product.
To keep the IM Dictionary relevant and useful, Forum Engineering must continue to grow the dictionary by adding new words—and this can now be achieved automatically. For the initial implementation of the dictionary, the basic keywords were extracted and placed manually in the tree structure. Now, however, the company is using IBM Watson Studio (formerly IBM Data Science Experience on Cloud), an analytics platform that makes full use of open source technology, to automate three of the key processes involved in growing the dictionary.
First, the team is using Watson Studio to extract knowledge by automatically identifying new words that are not in the dictionary, and vectorizing them using the Word2vec algorithm. Second, the team uses a deep neural network to apply probabilistic techniques to the word vectors and identify the correct position of each new word in the dictionary tree. Finally, the team uses GraphFrames to create a graph of the relationships between the new word and the existing words.
Boosting matching accuracy and driving further business growth
By adopting the AI-powered staffing solution, Forum Engineering saw an 83% increase in their matching efficiency—more than 58% of their contracts are now fulfilled through one single matching attempt. In addition, the average time needed from initial recruitment to job placement has been reduced from two weeks to one week.
These results were achieved only one month after the new system has been up and running. Since then, Forum Engineering has seen a dramatic improvement in customer satisfaction and worker fulfillment. The technology boosted the company’s reputation and credibility in the staffing industry, attracting new customers from around the world. It also positions the staffing agency to take advantage of a shortage of engineers in the manufacturing industry, filling positions more quickly than its competitors and earning greater market share.
"We’re starting to use data to our advantage more," says Mr. Masahiro Takeuchi, Chief Information Officer of Forum Engineering. “In the beginning, we didn’t have any data to work with. The key to success is to collect useful data that will respond well to analysis techniques such as machine learning.”
As a result of significant improvements in business efficiency, the company’s staffing specialists can now be more in touch with, and focus on the human element of their jobs—spending more time strengthening their relationships with customers and delivering higher-value interactions. The company aims to create new business by redirecting surplus capacity in the team to work on growth areas and new business development. Further growth is expected by transforming the company’s traditional labor-intensive talent management business into a completely knowledge-based, data-driven industry.
About Forum Engineering Inc.
Forum Engineering Inc. is a leading company that provides personnel services, specializing in the placement of technical and engineering contractors. The company employs approximately 5,000 engineers, and dispatches them to work on more than 14,000 projects every year.