Sports Fans Enjoy Power of Leibniz Entity Recognition
With the World Cup final between Germany and Argentina set for July 13 at Estadio do Maracana in Rio de Janeiro, the powerful appeal of sporting events to a global audience becomes apparent once again. What began with a 3-1 victory for host Brazil over Croatia nearly a month ago has reached the point of soccer-fueled hysteria.
Simply stated, people love their sports.
Give the people what they want—that’s the goal of the team of Microsoft researchers and developers that have put together the Leibniz contributions to the Bing Sports app, which debuted on the Windows 8.1 desktop on Feb. 23 with Associated Press feeds about the National Basketball Association. Article-reader impressions in the app spiked immediately.
“Leibniz has been instrumental in enabling the first of a suite of entity-centric scenarios for the Bing Sports app,” says Dhyanesh Narayanan, a program manager with Microsoft Research Silicon Valley, who helped shepherd the Leibniz sports technology to market. “These scenarios are intended to drive up user engagement with the app and were viewed by the business as an important priority for its release.”
A feed for Major League Baseball—in addition to the National Hockey League and the National Football League—followed on March 14, and the baseball offering, too, experienced an ensuing spike in interest, even though, at that point, the sports was only preparing for its season.
“Initial telemetry from the launch shows a doubling of article-reader impressions,” says researcher Bo Zhao, “which is promising and suggests that users are being served more relevant content in the Sports app that keeps them engaged, thanks to algorithmic tagging of news articles by Leibniz.”
The Bing Sports app makes it easy to keep tabs on the teams and sports you follow, with headlines, scores, schedules, standings, and statistics—the lingua franca of the sports-obsessed. The experience, also available on Windows Phone 8.1, is personalized with Live Tile updates on your favorite teams. Highly relevant content proves particularly beneficial to Windows Phone users, who can access the information in which they are interested without additional taps and swipes.
Leibniz is an entity-resolution system that uses web data as a proxy that can be applied to an entity. Leibniz applies knowledge technologies based on named-entity recognition, linking, and salience to enable the sorts of entity-centric scenarios appropriate for the Bing Sports app. The result is an enhanced, in-app experience that delivers recent, relevant news about a user’s favorite sports teams. The more popular a story gets, the higher it is elevated in the app. This appeal tends to keep users’ eyes glued to the constantly updated content.
“Sports fans care deeply about entities—from players to teams to leagues, says Shamik Basu, engineering manager for the project. “The Leibniz technology from Microsoft Research allows us to bring together best-of-breed content from diverse sources in high-value user experiences pivoted around these entities.
“This is a shining example of Microsoft Research providing a product team with a differentiator that the competition will be hard-pressed to match. We are also very excited about the potential to scale this beyond the sports domain to news, finance, and elsewhere, as well as to markets around the world.”
An interesting angle to the development of the Leibniz sports system is that the Bing Sports app contribution resulted from the work of three small teams from three countries on two continents.
The project was driven by Narayanan, with strategic guidance from team lead Ashok Chandra, along with colleagues Aitao Chen and Zhao. Jedidja Bourgeois, a contract developer deeply engaged in transferring the technology into the Bing Sports app, is part of the Leibniz developmental team, based in the Canadian capital of Ottawa. And the Information and Connected Experiences sports team—including Basu, Manu Pushpendran, Kanwaljeet Singla, Arunachalam Thenappan, and Pramod Kumar Chandoria—works out of the Microsoft India Development Center, located in Hyderabad.
“The ingested news enabled by Leibniz on our team pages is resonating well with our users,” Pushpendran reports. “Having ingested content show up on team pages also enables us to build intelligent systems to select and rank articles. We look forward to a continuing partnership as we onboard additional leagues, teams, and players in international markets through the Leibniz pipeline.”
The appreciation, Narayanan says, is mutual.
“Collaboration across organizational and geographic boundaries is an interesting challenge,” he says, “and requires the teams to be in resonance relative to planning and execution. This release is the culmination of a strong cross-organizational, cross-geography partnership across the Bing Sports app and Microsoft Research’s Leibniz team, with a focused virtual team across India, Silicon Valley, and Ottawa, driving passionately toward creating delightful experiences in Microsoft offerings.
“Over a period of several weeks, we established the foundations of mutual trust across the product and research stakeholders through conference calls—mostly at odd times, considering the time-zone differences—and charted out a plan for the scenarios and deliverables. During the execution phase, we had daily standup calls and email status updates to track progress and make tactical adjustments to features, the schedule, and priorities, which enabled us to land our deliverables on time with high quality.”
The Bing Sports app release further underscores Leibniz’s capability to apply its general-purpose matching technology to disparate domains. In 2013, for example, appeared a story about the use of Leibniz for the Windows 8 Travel app.
As Narayanan explains, the travel and sports apps use differing techniques to serve the needs of their corresponding verticals. With the travel app, the team used “conflation,” which pertains to matching structured data with structured data. In the Bing Sports app, the team applied “extraction,” which matches unstructured data with a structured type ontology.
You could ask why Microsoft researchers would be getting involved in such pursuits at all.
“The reason Microsoft Research is working on this project has to do with the mission of Microsoft Research and our role in creating differentiating experiences for Microsoft products,” Narayanan says. “As [Microsoft Research founder] Rick Rashid used to say: ‘Our researchers are here to push ahead the state of the art in computer science. When we have great ideas that work, we strive to move those ideas and technologies into Microsoft products as rapidly as possible.’
“What Leibniz has done for the Bing Sports app is exactly this: take bleeding-edge technology around algorithmic processing of content and leverage that to power a concrete product scenario that has a measurably positive impact on end-user experience.”