Back to top

How Search and Big Data Analytics Make Recruiting Faster and Easier

Using Document-to-Document Match for a 360 Degree View of Your Candidates

Mark David
Mark David
Functional & Industry Analytics Manager

Going beyond simply scanning resume keywords, combining big data and document-to-document match enables recruiters to find qualified candidates faster and easier. 

How does this recruiting analytics technique work? Watch our architect's video story from the trenches.



Racing against your competitors to win the best candidates is rewarding but can sometimes be very stressful for recruiters. You’re not alone: 95% of nearly 1,500 recruiters surveyed by Jobvite anticipated the job market to remain or get more competitive. The rise of social media and external candidate database services may help recruiters reach far and wide, but reaching a large base of job seekers may not be as difficult for recruiters as being able to zero-in on the right candidates quickly when demand comes up. 

We’ve learned from our recruiting and staffing customers that, due to its competitive nature, the industry really understands the value of good search. Why? Because candidate search accuracy and timeliness contributes directly to the recruiting firm’s bottom line. Fill rate increase and job advertising cost reduction are among the most obvious metrics showing the connection between good search and higher revenue. And who doesn’t like a quick and easy search experience after getting too accustomed to Google search?

But search is not easy. As a recruiter, you may recall having to use Boolean search or spending time entering keywords only to get no qualified candidates from search results. As our Chief Architect, Paul Nelson, summed it up in a previous post:

Even text search relevancy is an open problem. After all, relevancy ranking – "Is the document a good match?" – touches on meaning which depends on the data structure, the usage of the terms, and the infinite variety of how people express themselves, both as writers and searchers.

Take a java developer’s resume for example. Is it java the coffee or java the programming language? All of these things are very difficult to figure out unless the entire document context, not just individual keywords, is taken into account. Doing this properly eventually will make a tremendous impact on the recruiter’s performance and bottom line. 


Beyond Keywords – How Search and Analytics Help Recruiters “See” Real-Life Candidates

In the last several years, big data analytics in recruiting has gone a long way from just scanning for resume or CV keywords to solving a more interesting problem: automatically matching people to jobs. 

Document-to-document match and big data work for the recruiting industry in a number of ways:

  1. Building new models of how a match between one document to another should be scored (how good is the match between this candidate and a given job description?). The idea behind automated document-to-document search and match is to compare the 360o view of a candidate with the recruiter’s job description, including skills, experience, geography, salary, social media profiles, past hirings, etc.
  2. Generating phrase statistics ahead of time across entire sets of candidate and job data 
  3. Parsing resumes and job descriptions to produce a wide range of relevant phrases for dictionaries and acronym dictionaries 
  4. Using big data to come up with dictionaries of useful jargon. Generating one single dictionary to enable the search engine to understand the differences across multiple industries is difficult, so we leverage a big data cluster to parse all the documents the recruiting company has into different slices according to industries and then figure out the most relevant or popular phrases in each industry. 

This process can save recruiters significant time and make the company more competitive, as long as the search engine is configured so that resumes are parsed and analyzed semantically. Just like how Google constantly refreshes and optimizes their ranking algorithms for Internet search, we rely on search engine scoring, machine learning, and multilingual text analytics in some instances, to continuously improve search and match algorithms for big data recruiting applications. 


Outlook: Search and Match for the Legal, Pharmaceutical, and Real Estate Sectors

The same search and match technique can also be applied in data intensive business functions such as legal precedent search, pharmaceutical search, and real estate search. The value it brings – time savings, cost reduction, and scalability – will be enormously beneficial to these sectors. For now, we’re excited about taking big data analytics and scoring algorithms further to improve search and match accuracy, getting customers where they need to go quicker.

-- Mark