Major publisher uses machine learning to drastically improve text categorization efficiencyBy Innodata Inc. on December 14, 2017
Our client is a major publisher and information services provider. With more than a dozen subsidiaries, and customers spanning academic, K-12, medical, law, enterprise and government, the company maintains massive libraries and archives which are constantly growing. The need for quality text categorization is obvious. Busy professionals across the customer base need easy access to databases that provide them accurate, up-to-date results.
Maintaining a large team of subject matter experts to read individual entries and categorize them is timeconsuming, mind-numbing and, of course, expensive. At the other end of the spectrum, publishers and information providers often look to automation to solve their categorization problem. But fully-automated solutions without human intervention, while cheaper, are far, far less accurate. So, manual categorization is wildly expensive, full automation is wildly inaccurate. What to do?