How research supports collaboration in the new Office

By now, you’ve probably seen that the new Office launched globally just a couple of days ago on Sept. 22 — we’re thrilled to see how it is already offering new ways for people to work together.

Harry Shum
Harry Shum

In my role leading Microsoft’s Technology and Research (TnR) group, a big part of my job is to bring long-term research discoveries to the product teams, a process that involves a lot of close collaboration. In today’s workplace, information is everywhere, with fewer and fewer barriers to access. But in a world where information flows faster than ever before, just transferring information doesn’t necessarily mean better collaboration. We believe that collaboration is made better when intelligence, inference, context and intent are applied.

Today, I want to talk about how some of the collaboration between research and Office teams has resulted in new features that can empower people and their teams to do and achieve more. Let me share just a few examples of many in the new Office….

  • Clutter is a feature in Office 365 designed to help people focus on the most important messages in their inboxes by moving lower priority messages into a new Clutter folder. Clutter came about from a longstanding collaboration between Microsoft Research Cambridge and the Exchange team. The work arose from the creation of a new machine learning framework called Infer.NET, which uses a technique called probabilistic programming previously only used in academic contexts. In fact, this is its first use in a large-scale production system—from a research perspective, that ability to reach millions of people is a very exciting first.
  • The Smart Lookup feature, available in Word, Outlook, PowerPoint and Excel, was built in close collaboration across three teams at Microsoft — Office, Bing and Microsoft Research. In this case, our research team was actually embedded within the product teams to build the technology, and we contributed system architecture as well as machine learning models. How does it work? If you’re creating a document and need more information, you can select a word or phrase to prompt an “Insights” pane — giving you more contextual information from sources like Wikipedia, Bing and the Web. This feature lets you do what we call contextual search — it translates your selection into a search query in the context of your document. For example, if you’re reading an article about Paul Simon the politician, your search results will reflect that context, not Paul Simon the musician. This is smart machine intelligence at work.
  • The Microsoft Translator cloud-based automatic translation service, available in Office, was built on much of the work of researchers in Microsoft Research who developed new techniques for automatic translation. One example of this is the application of deep neural networks, which has brought groundbreaking advances in speech recognition. This technology is being applied in text translation as well as speech recognition.
    For a more personalized translation experience in Office, we offer the Translator App for Office in the Office Store. This app allows Office users to use a custom translation engine, optimized to you, trained and customized using the Microsoft Translator Hub — a secure, personalized machine learning portal that can create custom translation systems based on previously translated documents and your own preferences.

Clearly, the role that collaboration plays is huge. In this case, it meant that the efforts of long-term research combined with the great work of our product teams to deliver the new Office that empowers individuals and teams to do and achieve more. We in Technology and Research are really excited about what the new Office can do to help reinvent productivity and our role and contributions in helping the Office team deliver on that promise. Check out the new Office today and find out for yourself how it can take the work out of working together.

Harry Shum is the executive vice president Microsoft’s Technology and Research group.