The founder of a leading developer of text analytic solutions is confident that machine learning hasn’t displaced the learning done by human beings.
Computers and human beings generate digital data at the rate of 1,000,000,000,000 bytes every hour. That’s more bytes than stars in the average galaxy – and there are more of them every hour of every day.
The Big Bang of data in all aspects of our professional and personal lives is sometimes overwhelming. Where it comes to the drug development pipeline, when time matters to patients with cancer and rare diseases, researchers must struggle to keep current with the multiplying volume of information from peer-reviewed scientific research, patent filings, clinical trials data, news, competitive briefs, and much, much more.
Machine learning algorithms go a long way to managing the burden, of course. Yet the founder of a leading developer of text analytic solutions is confident that machine learning hasn’t displaced the learning done by human beings.
Lee Harland earned a Ph.D. in genetics from King’s College, London, and worked for more than 15 years in life sciences, leading semantic web, data integration, and text mining efforts. In 2013, he founded SciBite, developers of semantic solutions for the life sciences and pharmaceutical industries. Harland is an enthusiastic advocate of what he calls “human curation” even as his firm wins awards for its technology including Best of Show at Bio IT 2017.
“The quality of what you get out is directly related to the quality of what you put in,” he tells CCC’s Chris Kenneally. “If you go a little bit further and pretreat your data so that it’s a bit more structured, a bit more organized, and then feed that to these algorithms, we’ve seen time and time again with our customers that these algorithms start performing much better.
“Of course, this isn’t a battle of machine learning vs. non-machine learning. These things are not polar opposites. This is an ecosystem of technology where applying the right technology at the right time is what you want to do. So you can use machine learning to help build these really deep layers of disambiguation and rules for applying ontologies. Then, of course, you can apply the ontologies to extract really cool data, and then you can push that back to machine learning to do some nice analytics.
“The two combined give you an incredible power that up until a few years ago was just not available.”