SCROLL DOWN TO READ THE POST
Google’s new Talk to Books: Semantic search for book and idea discovery
I am truly excited about sharing this new approach to search! Imagine if you had the power to ask authors across time and disciplines your most burning questions or for their best advice. Now you can.
This week TED curator Chris Anderson and futurist Ray Kurzweil introduced Talk to Books. The feature was developed by a Google Research team headed by Kurzweil, a Google director of engineering, who is also well known for his work on optical character recognition, text-to-speech synthesis, speech recognition, transhumanism, and robotics.
What is Talk to Books?
Rather than using keywords and phrases, ask any question and Talk to Books will use semantic search strategies, machine learning that attempts to understand the meaning behind your natural language query. You can read more on Google’s new Semantic Experiences page.
Talk to Books results are retrieved from Google Books’ collection of 100,000 titles or 600 million sentences. Responses appear as a list of titles and book covers with the appropriate quoted text. Links lead users directly to the passages in Google Books.
In his April 13th post on the Google Research blog, Kurzweil introduced Google’s latest semantic search research projects:
Today, we are proud to share Semantic Experiences, a website showing two examples of how these new capabilities can drive applications that weren’t possible before. Talk to Books is an entirely new way to explore books by starting at the sentence level, rather than the author or topic level. Semantris is a word association game powered by machine learning, where you type out words associated with a given prompt. We have also published “Universal Sentence Encoder”, which describes the models used for these examples in more detail. Lastly, we’ve provided a pretrained semantic TensorFlow module for the community to experiment with their own sentence and phrase encoding.
an entirely new way to explore books. You make a statement or ask a question, and the tool finds sentences in books that respond, with no dependence on keyword matching. In a sense you are talking to the books, getting responses which can help you determine if you’re interested in reading them or not.The models driving this experience were trained on a billion conversation-like pairs of sentences, learning to identify what a good response might look like. Once you ask your question (or make a statement), the tools searches all the sentences in over 100,000 books to find the ones that respond to your input based on semantic meaning at the sentence level; there are no predefined rules bounding the relationship between what you put in and the results you get.This capability is unique and can help you find interesting books that a keyword search might not surface, but there’s still room for improvement . . . However, one benefit of this is that the tool may help people discover unexpected authors and titles, and surface books in a way that is fresh and innovative.
For a background of the research involved in the development of Talk to Books, read the Xiv paper, Universal Sentence Encoder.
My results were a little uneven but Talk to Books is so promising!
If you ask a question that is likely to be answered in a nonfiction book, you are likely to successfully develop context as well as an impressive range of answers to questions. I can see this as a valuable strategy for helping students gain background knowledge in a new area of inquiry and for promoting the discovery of new books and authors relevant to their areas of interest. I can see using it to discover varying opinions on a thorny issue, inspiration relating to philosophical dilemmas, and for gathering quotes.
For book lovers, it is simply fun.
Here’s what it looks like:
Here is my pretty darn good test search for a typical student question: How did Elizabethan’s dress?
Here is one example of a search shared on the home page: Why is a free press important?
Other sample searches include:
And for a look under the hood:
As you introduce a machine learning tool like Talk to Books, you may want to lead students in exploring underneath the hood with the pair of word association games gathered in Semantris. The games explore how Google’s artificial intelligence predicts semantically-related words and phrases. The background material on the About page will guide you.
Thanks to Mary-Catherine in my 530 class for the lead!
About Joyce Valenza
Joyce is an Assistant Professor of Teaching at Rutgers University School of Information and Communication, a technology writer, speaker, blogger and learner. Follow her on Twitter: @joycevalenza
SLJ Blog Network