With so much content these days, TL;DR (too long; didn’t read) has become an ongoing theme in our lives. Skimcast can make your reading list just a little shorter with their machine-learning software that skims over your preferred text and gives you the gist of it. Exciting, right?
Founder/CEO and current University of Georgia professor Bill Hollingsworth came up with the idea after having a hard time reading notes during graduate school as a result of a rare medical condition. After obtaining a scholarship to the University of Cambridge, Hollingsworth studied linguistics and created the first Skimcast prototype. Now, Skimcast has grown up into a machine-learning online tool and Chrome extension to help you get through articles and books faster, especially if you’re doing research for school, work, or just want to keep up with the latest news.
Here, the Skimcast team gives Hypepotamus the latest news on their new chrome extension, future product development plans, and what makes their technology life-changing to readers.
How did you get the idea for Skimcast?
I don’t see well enough to read standard text in textbooks. So I was originally a math major at the University of Georgia and as an undergraduate, I was getting all my books on tape. When I got to graduate school, I wasn’t reading textbooks anymore, I was reading notes. They didn’t come on tape and I started experimenting with OCR — computer programs that would scan the document and turn it into text and then let me listen to it on the computer.
I got out of math and into computer science with the idea of making software to help me more. I started looking at linguistics, language and speech and trying to improve the quality of it. I wrote my own speech synthesizer and in doing that,I got a Gates scholarship to go to the University of Cambridge.
I figured out in doing my thesis that when you’re doing a search for information, when you’re doing research, you have to be able to answer 2 questions quickly every time you pick up a document.
Number 1 is, is this document relevant to my search? If it’s not, you don’t read it, especially if it’s a book. You’re not going to read it cover to cover if it’s not relevant. The 2nd question, if it is relevant, where’s the relevant information? If it’s just one passage in one chapter, again you’re not going to read the whole thing, you’re going to want to be able to find it quickly. I couldn’t answer either of those questions without listening to the entire document from beginning to end. I proposed this idea of simulating the skimming process of a skilled researcher on a computer. Then, I wrote my first prototype of Skimcast, which I used to do my own literature review.
How does Skimcast work?
You upload a document: it could be an URL, the web document, or it could be a PDF. Whatever it is, once it gets the text, Skimcast reads it linguistically and pulls out linguistic concepts, semantic concepts or themes that are most important in characterizing the content of the text.
You just released your new Chrome extension, what are its features?
Currently, we don’t have a way for people to skim material behind subscriber walls, such as online course material and we want people to be able to do that. With the browser extension, you’ll be able to get a summary of whatever is on your screen. We’re still figuring out whether that’s something that we can transfer to an app on your phone.
Who are the ideal users for Skimcast?
To start off with — college students. The basic formula for who the customer is, it’s anyone who has more to read than they have time. Anyone who’s doing that kind of research. I think that college students are a natural first step because they certainly have more than they want to read. A lot of them have more than they can read. Definitely in terms of developing a thesis or a dissertation, it’s great to know, when you’re going through a literature review, all of the documents you could potentially need.
What are the various ways to use Skimcast?
There are different uses for Skimcast. That’s one of the things that’s exciting for me. It has the intelligence to do so many different things, depending on what you need. The obvious thing it does is it summarizes. A researcher doesn’t always want a summary of everything. A researcher wants to understand what he or she is reading better. The “View in Context” button does that. I get the intelligence of Skimcast telling me what’s important, but it’s also showing me what’s important in the context of the document. I still don’t have to read the whole thing, but when I read what Skimcast says is important, it’s easy for me to look up and down and see where I am. It’s the summary overlaid on the full text of the document, the best of both worlds.
Are there any plans to monetize this idea or are you just offering it up as a free service?
It’s a free service now and really all high school and college students are a core, initial audience that we’re rolling it out to and once we get to a certain level of users, we’ll introduce ads. There will always be a number of skims you can do each month for free, but we’ll get to a premium model where perhaps it’ll be, you’ll be able to get 5 skims a month for free and for unlimited skims and unlimited utilities it’ll be $4.99 a month or something like that. I think over time we’ll see how many total users we have obviously, but we’ll also be able to see how many heavy users we have.
Are there any product development plans?
We’re pricing out app development right now, because we know that people, particularly this audience, they interact first and foremost with their phones. Because this is research and homework oriented, they might use their laptops a little more because it’s more academic-oriented, but we’re working on an app now.
What about other languages?
Even if the vocabularies are completely different, Skimcast doesn’t have to have databases of vocabulary for every subject. It just looks at how usage works. What I’m learning from these test cases is that the language of technical terms works very similar in other languages.
Even though Skimcast doesn’t have any understanding of German or Spanish, it’s kind of picking up what the themes should be. I’ve a plan, once we get past the app and the extensions, to start properly adding some linguistic knowledge to the system for doing other languages like Spanish, so it does an even better job. We want reading to be easy and skimming and studying and research to be easy no matter where users are and no matter what device they use.
As an academic, are you encouraging your students to use Skimcast?
They know about it. I’m not actually pushing it at school. I want to see if other people push it for me. They know about it and they’re excited about it. They ask me, they come to my office and ask me how it works and they want to get involved, but in class we don’t talk about it.
A lot of them are interested in pursuing natural language processing on their own, to try to come up with their own apps. What I tell them every time is that all you need to come up with the next interesting app in natural language processing is an idea and some data. There’s so much low hanging fruit with what you can do, maybe not technology-wise but certainly when you’re merging with a task or with data, because there’s so much data now and so many tasks. It’s an exciting world to be living in and thinking about.
What are the goals for the next 6-12 months?
Over the next 6 months, we would all like to see meaningful usage develop. It’s hard to say what exactly that means. We’re living in a strange marketing world where passive or viral marketing is the most effective kind of marketing. Almost by definition you can’t contrive to create it.
We hope that we’re going to strike enough little matches here and there, that enough little pockets of usage will catch on. Hopefully in 6 months, there’s meaningful usage, and a decent percentage of high school and college students have heard of Skimcast and are using it and we have an app up and going.
I think I would say 12 months from now, it certainly would be good to have some active usage in the 6 figures, but we’ll see. The hardest part of understanding that answer is knowing what success looks like for this sort of thing. We want a lot of users, but what I really want to see is regular users, and heavy users. Then we know that this is actually helping people on a regular basis. It’s making an impact on the way that they study if they’re regular. Repeat users.