Data Science in Action

Working with Seattle-based AI start-up, Seasalt.ai, students from UBC’s Master of Data Science in Computational Linguistics program created a universal NER (Named Entity Recognition) system that applied transfer learning from high-resource language datasets to low-resource languages. This allowed crucial information to be extracted from previously underrepresented languages, like Indonesian, Javanese, Malay, Vietnamese, Tagalog, Croatian, and Czech, for use across a variety of Natural Language Processing tasks.

As a result of a question posed by UBC Linguistics and French, Hispanic, and Italian Studies faculty, a group of UBC MDS Computational Linguistics students embarked on the first phase of a project that could improve graded readers for any language. The capstone project focused on A1, A2, and B level documents for the Spanish language, but it laid the foundation for the creation of a universal reader.

A group of UBC MDS Vancouver students were asked by a global, private investment firm to answer a single question for their capstone project—could they detect deception in earnings calls using data science? After establishing a meaningful proxy for deception, engineering linguistics features to map deception, evaluating their worth, and identifying anomalies, the team supplied the firm with a reproducible report to build upon.

Together with the BC Centre on Substance Use (BCCSU), students from UBC’s MDS Vancouver program developed an interactive DrugSense dashboard interface for the BCCSU’s website. The goal of the project was to give those who use drug-checking services a way to visualize and explore key trends and statistics related to specific drugs and an ever-changing and unregulated drug market, while providing opportunities to learn more.

Ilana Zimmerman, MDS Computational Linguistics Alumna (Class of 2020) was tasked to create a search engine based on two transformer models in her role as a Natural Language Processing Engineer with ALEX - Alternative Experts. ALEX is an ISO 9001:2015-certified solutions provider to Government, Defense, and Commercial contracts, based in Washington, DC.

Agri-Food Canada enlisted the help of a team of students from the University of British Columbia’s Master of Data Science Okanagan program to help AAFC improve the overall efficiency of the tree fruit breeding process by providing unbiased and precise data on sweet cherry traits in an expedited fashion, while avoiding cost arising from unproductive procedures and human errors.

Working with the Vancouver Whitecaps, Vancouver’s professional Major League Soccer team, a group of UBC MDS Vancouver students analyzed players physical performance to build a model for player fatigue and fitness.  The students developed a data pipeline which can help the Whitecaps analyze and understand training loads for each session and closely examine the players' physical outputs during training sessions and matches.

A team of UBC MDS Vancouver students worked with Biba, a smart playground company, to build several learning models in order to estimate the number of monthly sessions at particular playgrounds within a given month. Data was collected from 2506 playgrounds across the US to help understand how playgrounds are being used in order for park managers to make meaningful decisions regarding the management of existing playgrounds and the planning of new playgrounds.

In partnership with Statistics Canada, students of UBC’s Master of Data Science Okanagan program used clustering methods, principal component analysis and principal component regression to provide greater insight into the spread of COVID-19 among of long-term care homes. The project showcased how leveraging various open data sources can produce comprehensive and meaningful results.

Students from UBC’s MDS in Computational Linguistics partnered with Minerva Intelligence, an AI company that provides knowledge in earth science domains like mining and natural hazards, to extract information from MINFILE, a British Columbia Government mineral occurrences database.  The team developed a way to extract details from these reports in order to help Minerva’s knowledge base and make their AI system more robust.