Google for Books

Another fascinating piece about Amazon's new book search facility. I have been checking this out over the weekend, and it really is interesting. One tip: go to the index first. Then select the topics which interest you and run a search. Eg: I was looking at something about meat eating and evolution, so I tried 'brain'. Fantastic. I suspect that this will help to sell books in most cases, since the same argument applies to the book as applies to software: you don't know what you're missing till you have a try! But there is that other little sub-divide: those who read directly online and those who run printouts. My feeling is that the on-screen readers are still a very very small minority, so we get to be the 'free riders' in what will otherwise probably be a very worthwhile commercial experiment. My book purchasing habits will definitely change: check this space a year from now to find out how.

Call it a Google for books: Amazon's latest feature allows readers to search millions of pages online to browse before they buy. The question now is whether they will buy after they browse.

The nimble search engine unveiled by mammoth online retailer Amazon.com makes 120,000 of its books -- or 33 million pages -- fully searchable for free. When the reader types in a word or phrase, the ``Search Inside the Book'' technology will call up every reference in each book, along with the page numbers. Readers can also call up the two pages before and after, if they sign in and provide a credit card number, which is not charged unless they buy. The feature is already drawing rave reviews from librarians and researchers. "This is a really great feature for the public,'' said Mary McGrath, librarian at the Redwood City Public Library. ``It crosses over into being a real reference tool. Definitely a value add.'' For the reader, at least. Many bricks-and-mortar book stores are mixed about the effect on their sales.

Amazon said it has gained permission from more than 190 publishers to scan their books for use in the search engine. Some publishers are concerned that users will gain too much access to copyrighted text, and it will discourage sales, especially in non-fiction. "If you're only looking for hotels in Beijing that serve hamburgers, and you can find that information by browsing the books on Amazon, then you don't need to buy the book,'' said Karen Pennington, retail director at Menlo Park's Kepler's Books and Magazines. With fiction, Pennington notes, "Online is not the best place to read a novel.'' Calling up several pages from a book can give you a sense of the writer's style, but the electronic format and the limited number of pages make it quite distinct from having the whole book in your hands, she said.

Some publishers are letting Amazon include a limited number of books in the search program. The University of California Press provides 2,000 titles, out of a possible 4,000, to Amazon for the search. The university press denied Amazon the right to access many reference books, as well as books of poetry. "Often an entire poem is on one page,'' said Laura Driussi, assistant marketing director for the University of California Press. "Making that available would deprive the poet of the royalties for the poem.'' However, the feature could also work to UC's benefit, Driussi said. Amazon helps bring attention to books that otherwise don't receive much publicity, she said, and so can only help sales of such books. Readers may now find key words or phrases hidden deep inside books they'd never thought of looking at before. Among some safeguards to encourage buying, Amazon agreed to shut down a customer's access to a book after they scan more than 20 percent of a book in any given month, Driussi said. Also, reference pages cannot be copied or downloaded, although they can be read on-screen and printed.


Amazon is being cagey about its formula for ranking books. Steve Kessel, vice president of Amazon's media group, said only that Amazon would take into account whether a search term is contained in the title or author of a book. The ranking would also take into consideration the number of times a search term appears, and how close the words appear together to each other in proximity, Kessel said. Kessel would not comment on whether price or popularity played a role in rankings. In one test of the feature, the Mercury News searched for ``King James Bible.'' The first four results were King James versions of the bible, with the most expensive one -- for $24 -- coming first. Date or sales ranking didn't seem to play a big role. The technology is the brainchild of Udi Manber, Amazon's search technology guru. His title is chief algorithm officer, and his role is to set the mathematical formula for how results are ranked. "That's our secret sauce, that's the crown jewels,'' he said.

Manber said the company has been busy scanning book texts all summer, but that now that the product has been launched, his technological challenge is over. He said the ``Search Inside the Book'' feature will be expanded in coming months, but did not specify how. He will now focus on launching another Amazon search engine product, this one more similar to the much-hyped search engine Google. The product will be run by a separate company -- owned by Amazon -- called A9 and based in Palo Alto. However, Manber said it is still too early to say exactly what the product's mission will be. He said he is not ruling out competing with Google, but stressed that such competition is not his ``purpose'' for now.
Source: Silicon Valley.Com
LINK