Péter's Digital Reference Shelf
December 2003

Amazon Search Inside the Book

URL: http://www.amazon.com
Publisher: Amazon.com
Cost: Free
Tested: Nov. 16-19, 2003

It was about 10 years ago that Jeff Bezos sat down to write his first business plan for selling books on the Web. In 1995, he and his company, Amazon.com, became the prime example for user-friendly electronic retail. It was making no profit at that time, but Bezos made it clear in his plan that it would take years to make it profitable. He did not promise those magic ROI figures that many Web ventures promised, but kept working on building and improving the online shopping experience.

After the phenomenally successful initial years (in terms of establishing brand name and gaining market share), Amazon extended its business line to apply its software and know-how to a variety of goods, from kitchen appliances to baby toys. It devoted less resources to its book catalog, even though I believe that Bezos could have expanded Amazon to become the customized online public access catalog in tens of thousands libraries, customizing the software and the content to reflect the holdings of those libraries and the specific procedures (acquisition, reserving, renewal, lending) that logistically are very similar to the Amazon ordering and buying process. I wrote a column five years ago discussing these issues.

Luckily, a few years ago, Amazon returned to its roots by offering the Look Inside The Book (LIB) feature. It allowed users to look at millions of pages of 250,000 books in all genres, ranging from the cover to the table of contents, the back of the book to the index, as well as excerpts. This was a very important and appealing feature and in many cases the excerpts covered 60-80 pages from a book. This fall, Amazon launched yet another new service, which was tentatively called Look Inside The Book II. Luckily, at the last minute the name was changed to Search Inside the Book (SIB).

Content

According to the open letter to customers, Amazon scanned 33-million pages of 120,000 books as approved by publishers (not necessarily by authors, some of whom challenged their publishers for giving approval). Being able to search the content of the books rather than just their metadata (title, author, publisher, genre, year), mightily expands the users' capabilities to judge the relevance of the book from their perspective. Not only the topical relevance, but also the relevance of the treatment level, which is quite apparent after reading a few pages from a book.

Although there are no details available as to how the books were made searchable; how the various genres are represented; or what the distribution of the searchable books is by publication year; my tests suggest that the scope was quite wide in all regards.

The Quasi Bookstore/Library Experience

At first, there may be nothing unusual when you get to the Amazon site. It looks as it did before, but the results of your first query will reveal an important difference, if not on the first page then on the second or third. The query "southeast Oahu" yields new results with some extra information.

In addition to the familiar informative bibliographic data (author, title, publication year), you can now see an excerpt with a hot-linked page number where the query term occurs. In addition, there is another link at the end of the record that leads to a page showing all of the excerpts from the book that match the query term. These excerpts can tell you if the book is promising and worth the clicking to fetch and display the full page.

In the case of "southeast Oahu," there are 13 excerpts from the first book, "Hidden Hawaii." They can be quite informative in and by themselves, or enigmatic, truncating the sentence just before the most interesting section. But then comes the best part — jumping to the specific page to read the top part of the page then scrolling down for the rest of the page. You can turn two pages forward and two pages backward to read the text and see a good map of the region. The next page is a continuation of the narrative. If you click on the next page link, you get a warning that you may not display more than two pages in this direction.

This five-page limit was introduced to ease the controversy after some authors and their guilds lodged protests and complained that the SIB feature would reduce book sales. A savvy searcher can often overcome this limitation (see the software section of the review), but even without the additional shrewd move, reading five pages of a book may give enough of an impression to decide if it is appropriate for the intended purposes.

Of course, for a better comparison, you may also look at the matching pages of some of the other books in your result list, or just quickly scan the excerpts. In this case, a quick look at the list of six excerpts from the Frommer guidebook may suggest that the first book, "Hidden Hawaii" is a better choice for the topic. (It also shows that the quotation marks do not necessarily mean exact-phrase searching).

On the other hand, you should not forget that the Frommer book may prefer to use the broader term Windward Oahu. Indeed, it has 40 more promising excerpts, while the other book does not appear on the result list, and other travel guides move to the top of the result list. The query "windward coast" changes the top-ranked books' order once again, so you have to be careful with your query formulation.

In spite of the quirks of the software, you can now do much the same thing, as you can in a large bookstore when browsing books: not judge them just by their covers. It is true that in a bookstore you can browse all the books on the shelves, except those that are shrink-wrapped (like some encyclopedias and dictionaries), whereas in Amazon's book store you can browse only some of the pages of 120,000 books in the multi-million item collection. Additionally, you can view the millions of displayable, but non-searchable, pages of the quarter-million books partially digitized earlier for the LIB feature.

However, by using this new search feature you can quickly zero in on the books that include target words and phrases. With a little savvy, and the help of the Power Search mode, you can limit your search to the sub-domain of the 120,000 books. This is not a small collection, especially in comparison with the inventory of bookstores and libraries.

The Digital Ready Reference Perspective

Most importantly, from the perspective of the digital ready reference, SIB can offer new and free alternatives using any book to find factual information quickly or to answer a reference question.

For example, if your patron wants to know if Elvis ever stayed in Kailua, you can easily check it out on Amazon and find 74 hits. You can see from the excerpt of the first hit that he did, according to the best-known biographer of Elvis. This is a much more credible source than reports of sightings by John Doe on the Web who saw Elvis on Kailua beach in 1959. This is even so if in his other book, which is the next hit from Amazon, Guralnick relocates Kailua to the west side of the island (but confirms the fact).

Answering reference questions about Elvis sightings, of course, is not a big deal. Finding credible and appropriate sources is. Books have been traditionally the most widely accepted reference sources for reliable facts. Encyclopedias, almanacs, fact books and dictionaries have made it to the open access Web at an impressive pace, but there is a big lacunae in certain categories, such as the social sciences, when it comes to traditional ready reference sources.

Digitizing and making full-text searchable tens of thousands of non-fiction books in all disciplines — memoirs, travel guides and traditional ready-reference books — not available online, especially without subscription, is very relevant for digital ready-reference.

It makes many genres of books ready-reference tools that were never considered as such because you could not instantly locate tidbits or passages of information about concepts, events, people, locations, procedures, treatments, customs and practices, even if the books had informative titles and good indexes.

The traditional ready-reference sources digitized by Amazon are particularly relevant if you take the time to explore which encyclopedias and dictionaries are available as searchable books and not available free of charge on the Web.

One of the words that I use in testing dictionaries and encyclopedias is umra, the small pilgrimage. Many of these digital ready-reference resources do not provide even a short definition, let alone an article. A quick search in Amazon Power search mode yielded 100 results (some of them were different editions of the same work) and most of them were searchable books. A few of them were irrelevant (because umra turned out to be someone's first name, the name of a geographic location or an acronym), but most of them were useful and informative. Often, the excerpts were sufficient, but if they were not then the passage on the page was only a click away. The good news is that after some experiencing and learning about the best sources in a discipline, it is possible to limit the search to specific books or genres, although such operations are not always perfect and Amazon does not endorse, let alone advocate, it.

Software

The interface is appealing and there is no need for a help file for navigation and simple searching (nevertheless, there is a good help file). You should be able to do exact phrase searching by using quotation marks around your words, such as "beach park," but it does not always guarantee that the first word will be followed by the second word on the target page. Sometimes the words were apart and not in the specified order. Space between words is treated like an AND operator, i.e. "beach AND park," yielding more results that are not necessarily relevant, although Amazon does a pretty good job in ranking results by relevance.

Advanced searching is offered via a structured, field-specific template. One oddity was that the software could not find anything for my search with Budapest in the title and Dorling-Kindersley as the publisher (either with or without the hyphen), even though there are two matching records, one for the out-of-print, and another for the upcoming edition of the Eyewitness Travel Guide.

Power Mode

The Power Search mode was enhanced, and the most important feature is the keywords prefix. Why? Because if you want to limit your search to an encyclopedia or a dictionary when looking for a good article or definition from a well-known, widely respected traditional ready-reference source, you can enter a query and specify your term with the keywords prefix along with the word "encyclopedia" or "dictionary" in the title, such as "title: encyclopedia AND kw: psychosurgery."

Along with 11 other titles, this query will list the Concise Encyclopedia of Psychology of John Wiley. It does not hurt that it won kudos from Library Journal, and — at least for my fellow information professionals in Hawaii — that the editor is a professor at the University of Hawaii. Never mind the "review" of a single reader (just take his/her comment with a grain of salt).

When you are inside the book and use the small query cell to enter the target word "psychosurgery," it will list nine excerpts, including the article about psychosurgery and a good illustration.

If you just search for "psychosurgery" in the basic search cell of Amazon, it brings up 431 books, but almost none of them have the SIB feature. The first does have the LIB logo and you can see a few pages selected by Amazon or the publisher, but you can't search inside the book. A large number of the following 40-50 books are out of print and have only minimal bibliographic information. Searching for "encyclopedia" in the title and "psychosurgery" as the subject in the advanced mode yields zero result, so your best choice, indeed, is the Power Search.

Once you learn that some of the best encyclopedias and dictionaries are not (yet) available as a database, and are unlikely ever to be offered online free of charge, you will appreciate the SIB option of Amazon and go directly to an SIB search of the encyclopedias, which you may bookmark. In many cases, the relevant excerpts show up directly through the link at the bottom of the record in the result list. In some cases these worked, in others they did not.

Overcoming The Two-Page Limit

Reacting to complaints by authors soon after the release of the SIB function, Amazon limited the number of pages that can be displayed after the page that includes the query terms. You can display the previous two and the following two pages. This is sufficient for a reference question, even in case of a long article. It can be a hindrance when you read an interesting chapter from a novel or a travel guide. It was not a problem in the southern Oahu search example at the beginning, as the 2nd, 3rd and 4th excerpts are on the 114th, 118th and 121st pages — more than dove-tailing for your reading pleasure. However, the 4th excerpt is on page 154, so you can't read the range between page the 124th and 151st pages, just when your appetite was whetted by the map on page 123, showing all those magnificent beach locations.

On the one hand, you can't complain, you had the chance to read about 40 pages of the book in the comfort of your home: you can just go ahead and order the book if you liked what you have read. On the other hand, if you would like to plan out your itinerary right then and there, you would like to look at those other pages as well in the 124-151 range.

There is a solution. On the map you have seen various beach names that you are interested in, so just scroll to the top of the page, and enter one of the names, like Heeia, and you get a list of excerpts with links to pages, describing that area, including p. 130.

Doing a few more directed searches would provide more than enough pages for your planning and to feel safe to spend the $13.27 + S/H for the book. Amazon more than deserves its cut on the sale.

In spite of some software glitches and limitations, the new Amazon enhanced by the SIB function offers impressive alternatives for digital ready-reference.