Today, we're thrilled to announce the public launch of Primer Labs.
Labs is a free, public facing platform where we’ll showcase the evolving machine learning technology that powers Primer. We designed Labs to be accessible and intuitive, emphasizing interactive experiments that will appeal to everyone from the most seasoned data scientists to those who are just curious about artificial intelligence and want to experience it for the first time.
We’re launching with three experiments:
Imagine having your own AI analyst who quickly reads your most critical documents and can answer your questions on the spot. That’s what Ask the Text is all about. This is our most forward looking experiment, and it pushes the boundaries of what’s currently possible with natural language processing. Test it against documents you know well or pop in something you’ve never read before to learn something new.
Have you ever found yourself in a position where you need to understand the gist of a long document but simply do not have enough time to read it? That’s where our summarization technology comes in. We’re able to sift through a document in seconds and present the key information, generating an executive summary for you. In this experiment we utilize abstractive summarization technology that, much like a human, is capable of pulling in vocabulary and tacit knowledge beyond what’s written in the text.
Named Entity Recognition (NER) is the starting point for most of the machine learning pipelines at Primer. It’s where we extract the people, organizations, and locations from streams of text. We are going two steps further here to extract and attribute quotations and to highlight key relationships between entities to offer lightweight but effective insights. We’re also doing coreference resolution to recognize when the text uses "he" or "she" to refer to an individual, rather than calling them by name. Knowing the "who, what, where" of a text is a powerful starting point for a human reader.
You’ll be able to play with all three of these experiments across a wide range of text. We have pre-loaded Labs with a wealth of sources ranging from news to financial documents, government publications, think tank pieces, scientific journals, and more. Our algorithms work on all sorts of text. But don’t just take our word for it. The real magic happens when you drop in your own documents -- give it a try. We’ll be able to summarize, extract key entities, and even let you ask questions of any document you serve up.
Transparency is one of our core product principles at Primer. We aspire to build artificial intelligence offerings that are entirely explainable instead of hiding behind a black box. We took this to heart as we built Labs, and established a design language revolving around transparency and focus. Upon visiting the site, you’ll notice elements of layered glass, with nuanced animations that shift the focus. When the AI answers a question, it strives to also show you the location in the text where it gleaned the answer, giving you full context. This design is part and parcel to Primer's mission to slice through the noise to help reveal the signal.
Stay tuned to Labs as we continue to add new experiments over the coming weeks and months. We encourage play! Over time, as we continue to release new experiments, the site will become a living history of natural language processing where you’ll be able to take part in the evolution of this science. So, join our community and come along for the ride -- we can't wait to see where Primer Labs takes us next.