Wei Li, PhD, Linguistics
Dr. Wei Li is teaching bots and crawlers to not just read language, but to understand it.
Li is the Chief Scientist at Netbase Solutions, a Silicon Valley company that mines social media data in order to assess public opinions through a cutting-edge technology that Li designed. Li completed a PhD in Computational Linguistics at SFU in 2001.
Unlike most other data mining companies, Netbase's parsing technology is one of the first in the field to apply an approach to interpreting text data--everything from tweets to Facebook updates--based on how people actually understand and use language.
“The mainstream approach used in this field is machine learning based on a bag of words, where keywords are counted and categorized as good, bad or neutral. But that doesn’t take into account the complexities of linguistic context and structures,” he says.
Li explains that his approach to data mining leads to more detailed and practical understanding of public sentiments--for example, going beyond identifying whether popular opinion on a product is positive or negative to explain the cause of those trends. These richer and more insightful results can then actually be applied to remedy weaknesses or misconceptions.
Li's technology has been hugely successful, winning his company Fortune 500 clients ranging from music industry professionals to soft drink giants. As a result, he explains, the field as whole has begun shifting.
“My field has been dominated by statistical models based on keywords rather than linguistic structures. Linguistics is often critiqued as not a science. This is unhealthy as both approaches have their respective pros and cons.” says Li.
Li is continuing to push the field of computational linguistics forward by leading a team of linguists to apply his parsing technology beyond its base in English and Chinese. His team has developed deep parsers for eight languages and is now expanding to 12 more new languages, including almost all major European and Asian languages.
“As a linguist I’m very excited to see if I can come up with a recipe that is general enough to adapt to all languages with very limited adjustment,” he says.
Li credits his graduate training for helping him gain the skills and insight that has led him to where he is now.
“I had the industry experience when I came to SFU but my thinking was too narrow. SFU helped broaden my academic horizons. Thanks to my Ph.D. advisors, Dr. Paul McFetridge and Dr. Fred Popowich, I learned how to communicate with both linguists and engineers and learn from both sides,” he says.
Author: Jackie Amsden