http://racerealty.com/

Man sets up book-search program based on writing styles

Posted: Thursday, April 03, 2008

Aaron Stanton, founder of the Web site CanGoogleHearMe.com, was tired but happy this Monday morning.

Courtesy Of Booklamp.org
Courtesy Of Booklamp.org

He and his team of programmers had been up 26 hours, broken only by a 4 a.m. trip to the store for more Red Bull energy drink, putting the final touches on a secret he's been keeping for more than a year. Now he has finally unveiled the secret on www.BookLamp.org.

You may remember Stanton. He walked into Google headquarters in California in February 2007 to pitch an idea to the mammoth Internet search company.

Stanton, 26, was turned away but didn't give up. He set up CanGoogleHearMe.com, his personal blog on the challenge of trying to get an audience with Google. The site attracted thousands of visitors, including someone inside Google. He soon got his meeting. He left California with an agreement with Google to not disclose his idea.

"California was kind of a Hail Mary," he said. "I either had to let the idea go or do something to try and make it happen." And his idea? A software program that allows people to search for books in a database, based not on reviews but on the writing style of a particular book.

Over the last year he's been assembling a team of programmers working for free in the hope that the idea pays off.

The idea came in 2003 while Stanton was a student majoring in industrial organizational psychology at the University of Idaho. He was doing a lot of writing and wanted to find out if there was a way to find similarities in writing. So he scanned in some books and started noticing that all writing had similar structures that could be examined to help determine the style of writing of particular books. He tried to get the idea off the ground then and formed a company called Novel Projects Inc. But his efforts faltered until now.

Say you like Stephen King novels. BookLamp can search a database of books and find books written in a similar style. The BookLamp program picks up on things like sentence length, the use of adjectives and adverbs, placement of quotations and commas and the use of "I" to determine if a book is in first person or third person.

Stanton believed that by analyzing the types of words used, the program can make educated guesses about the content of a scene in a book without a human actually reading it.

For example, Stanton says high verb use, high dialogue use and short sentences tend to indicate fast-paced action, while large words, low dialogue and high descriptive adjectives tend to indicate slower and denser scenes.

BookLamp would analyze King's writing style and match his style up with other books in the database.

"We look at the stylistic elements first, and that's not being done by any other system," he said.

Search engines for books now on the market can offer a reader books based on genre and user ratings, which can help determine plot, story line and characters. But Stanton says if a reader doesn't like the writing style, a book isn't going to be read. To help a reader further, BookLamp also includes any online reviews of the book that may be available.

The Web site launched earlier this month lets people use the program and submit their feedback. The database of books is small - about 180 books. It's made up mostly of science fiction. Stanton and his team created the database to test the program, but said to work most effectively it would need a database of more than one million books.

That's why he's been talking to Google, Amazon and Yahoo. He's got no response from Microsoft. He still has a nondisclosure agreement with Google, so he can't talk about what he's been discussing with the company. But he said talks with Amazon have gone well. He has a patent pending.

Stanton approached Google because of its efforts to create an online database of books. He felt his program could offer a solution to some of the issues the company was facing. Google hadn't been very successful with fiction books because the only search it was using was a keyword search, and that didn't pick up styles of writing. And there were copyright issues with scanning books.

Stanton says his program gets around copyright problems because it often doesn't have to scan a whole book - just enough so the program can pick up style cues. And the program provides only a short preview of the book to a site visitor.

Stanton said the goal of booklamp.org is to find out from people what they think should be the company's next step. It could be partnering with a larger company or going alone to start building a book database. It could mean opening up the BookLamp software codes so others can use it.

Stanton said the support of thousands of e-mails he's received since starting CanGoogleHearMe.com kept this project going, so he felt it was important to get feedback before deciding his next steps.

"Honestly, I'm trying to keep my hopes in check," he said. "We want to put it out there and want to see the response. We don't know what the next step is." Stanton said effort hasn't been about making money.

"We talked about two ways to run a company," he said. "It could be money driven, but when you run out of money it's over. Or it could be fun driven, and you never run out of fun."



CONTACT US

  • Switchboard: 907-586-3740
  • Circulation and Delivery: 907-523-2295
  • Newsroom Fax: 907-586-3028
  • Business Fax: 907-586-9097
  • Accounts Receivable: 907-523-2270
  • View the Staff Directory
  • or Send feedback

ADVERTISING

SUBSCRIBER SERVICES

SOCIAL NETWORKING