Tutorial: ABE new search facilities

This discusses new search features at the largest used/OP/rare
bookselling site: http://www.abebooks.com


From: Jayson (head of systems) on the ABE discussion forum:

With the new search engine in operation, some of the new features are
available through the web site and more will be coming as we phase in
the complete user interface for search. In the meantime, so that you
can take advantage of new search features, here are some of the
concepts supported by search on AbeBooks.

Full Text Search

The KEYWORD field is now searches title, author, publisher, ISBN,
subject and the FULL TEXT of the description. Previously full text
searches were not available. Wants matching will be updated early
next week to support this as well.

Advanced Boolean Searches

Advanced search now sports a new option to turn on boolean searching,
which you have to do if you use the old style boolean search (~ means
NOT) or the new boolean searches that use AND, OR and NOT keywords to
indicate comparisons between different search terms. There is also a
new help page for boolean searches that gives you the basics, but here
are a few of the concepts in more detail...

BOOLEAN SEARCH OPTION A new option on the advanced search pages
allows you to turn boolean search on or off. If you use AND, OR and
NOT within a normal search, they are just ignored unless they are
within double quotes in which case they are treated like any other
more significant word. Turning on boolean search adds meaning to
these words and uses them to compare different parts of the search
term as described in the boolean search help and in this tutorial.

TOP LEVEL NEGATION A fancy word for saying that you cannot use a
search term that only consists of something that doesn't exist. For
example, doing a search with only the author field and that field
containing "not King" will not work. Searching for "stephen not king"
on the otherhand would work. You always need a positive term that
will find something before you can use a negative term to subtract
something else. In many cases you can get away with using top level
negation in a field as long as you have at least one other text field
containing something that is not a top level negation. For example,
searching for Author "not king" while also searching for Title
"christine" would work in boolean mode.

PARENTHESIS These handy pieces of punctuation help you to group your
boolean searches so that you can do more complex queries. For
example, searching in the title field for "new flower not (york or
jersey)" will return all occurances of the two words "new" and
"flower" within the title field but not if they contain either "york"
or "jersey." The parenthesis are important so that the search engine
knows what you are talking about and in what order to do the
comparisons. It is similar to math, where the expression 5+8*2 is not
as clear as 5+(8*2) especially if you really meant (5+8)*2. The
search engine is smart enough to figure out that "new flower not new
york" means "(new flower) not (new york)" but if you want to be safe,
then group your words together to indicate your meaning.

Phrase Searches

If you want to search for a phrase of text, rather than individual
words, you can place them within double quotes to perform that match.
So searching on the two words, "big" and "lake" (excluding the quotes)
would return 162 results all containing those two words in any order,
and the phrase quoted "big lake" (including the quotes) would return
15 more accurate results with the words appearing exactly in that
order.

Picking Your Search Terms

We see many searches where searchers are using the entire book title
including punctuation and every word in the search. It is quicker and
will most likely return better results if you instead search for the
main significant words in the term. The small conjunctions and small
words like "if, is, an, ..." are ignored anyway if not placed in
quotes. Also, if you pick a term like "mountain" the search engine
would match that against "mountains" as well, but the opposite is not
true and "mountains" will miss the shorter non-plural version of
"mountain". Like the old search help says, less is usually better. I
agree, as long as the "less" contains the significant terms that
identify what you are looking for and filter out those that you are
not.

Advanced Use of the Keyword Field

The keyword field has a few hidden features that are not commonly used
and are somewhat experimental, but highly useful. One of those is
called "restrict operators." Since this field can search any of the
other text fields, you can do complex boolean queries all in one field
and get the search engine to do more of what you want it to do.
Restrict operators limit the search to look at only one field rather
than the entire list that keyword normally can scan (title, author,
publisher, subject, isbn, and full text).

So in the keyword field I could say...

text:reprint

to search for the word "reprint" in only the book description. (which
also will find "re-print" by the way). The full list of restrict
operators is:

* title
* author
* publisher
* subject
* isbn
* text

and a shorter version for most of these as

* tn (title)
* an (author)
* pn (publisher)
* sn (subject)

To use a restrict operator, just place a colon after it and then the
term you want to search. If you have more than one term to use with
that operator, you can either use a restrict operator before each
term, or put the list of terms in parenthesis. For example, looking
for Stephen King I could do either of the following:

* author:stephen author:king
* an:stephen an:king
* anstephen king)
* authorstephen king)

You can also place full boolean expressions inside the parenthesis as
well and also between restrict operators. Here are a few other
examples...

Search for all Stephen King books that do not have exlib or bce in any
field, the following would go into the keyword field:

anstephen king) not (exlib or bce)

Oops, want to get rid of those with edge wear, change that to:

anstephen king) not (exlib or bce) not text"edge wear" or
edgewear)

Note that the use of the keyword field and restrict operators is the
only way to explicitly search the subject field. For example:

anstephen king) sn:horror

And for a more complex answer, and homework, determine what this last
query does and have fun creating your own in the future...

In the keyword field, us:

anstephen king) pn:1983 sn:horror not tn:christine not text:wear

By the way, if you want to skip prefix operators for most of this
query, and do the same query, you could do the following:

Author field: stephen king
Title field: not christine
Publisher field: 1983
Keyword field: sn:horror not text:wear

Have fun searching! And please share query ideas on the forums for
others to use.

-- Jayson Minard
VP Systems
Abebooks