Progress in AI research and applications is exploding, and that explosion extends to our own team working on academic services. Continuing our work supercharging Bing and Cortana, we are also applying new technologies to Microsoft Academic, which serves the research community. If you’re not familiar with Microsoft Academic, this online destination helps researchers connect with the papers, conferences, people, and ideas that are most relevant, using bots that read, understand, and deliver the scientific news and papers researchers need to further their work.
Designed by and for researchers like myself, the site puts the broadest and deepest set of scientific information at your fingertips, with the ability to go beyond keywords to the contextual meaning of the content. Recently, we further enhanced the analytic content so users can see the latest research, news, and people, ranked by importance and credibility. Users can even drill down on the people, events, and institutions they care most about.
Behind the scenes, we are taking advantage of the fact that machines do not require time to sleep or eat, and have superior memory to humans. We have trained our AI robots to read, classify, and tag every document published to the web in real time. The result is a massive collection of academic knowledge we call the Microsoft Academic Graph (MAG), which is growing at roughly 1 million articles per week. While one set of robots is busy gathering knowledge from the web, another set of robots is dedicated to analyzing citation behaviors and computing the relative importance of each node in the MAG so that users are always presented with information they need and want.
Microsoft Academic is based on the work our team developed for Microsoft Cognitive Services, including open APIs that give developers AI-based semantic search tools and entity-linking capabilities. We’re also applying AI semantic search—which is contextual and conversational—to Cortana, Bing, and more.
As a research organization, we understand the pivotal role that open communication plays in advancing science. As such, we’re making the back-end dataset and algorithms available to all through Cognitive Services. There, everyone can access and conduct research on the massive and growing dataset through the cloud-based APIs. This means you don’t have to worry about the logistics of transmitting the massive dataset over the Internet, or manage a cluster of computers just to host and analyze the data. We are particularly excited that the research community has taken advantage of these cloud resources and already is collaborating on a common data and benchmarks platform to advance the state of the art. Earlier this year, we saw 81 teams participate in the WSDM Cup 2016 to develop new methods to rank papers, including newly published ones that have yet to receive any citations. An ongoing challenge is the KDD Cup 2016, which is focused on finding a better way to rank the importance of research institutions. The results of the first two stages of the contest have already been published, and I cannot wait to see the final outcomes and learn what new insights and technologies the 500 participating teams have developed when results are announced in August at KDD 2016 in San Francisco!
I encourage you to start experiencing the breadth and depth of what Microsoft Academic currently has to offer and to continue this journey with us in our mission to empower every academic and every academic institution on the planet to achieve more.