Google Hacks 2Nd Edition [Electronic resources] نسخه متنی

Hack 21. Like a Version

Gather a list of what Google thinks are
synonyms for a keyword you provide .

The Google ~ synonym operator ["Special
Syntax" in Chapter 1] widens
your search criteria to include not only the specific keywords in
your search, but also words Google has found to be
synonyms of, or at least in some way
related to, your query words. So while, for example, food
facts may only match a handful of pages of interest to you,
~food ~facts seeks out nutrition information,
cooking trivia, and more. And finding these synonyms is an
entertaining and potentially useful exercise in and of itself.
Here's one way...

Let's say we're looking for all the
synonyms for the word "car." First,
we search Google for ~car to find all the pages
that contain a synonym for "car" In
its search results, Google highlights synonyms in bold , just as it highlights regular keyword
matches. Scanning the results (the second page is shown in Figure 2-1) for ~car finds car, cars,
motor, auto, BMW, and other synonyms in boldface.

Now let's focus on the synonyms rather than our
original keyword, "car."
We'll do so by excluding the word
"car" from our query, like so:
~car -car. This saves us from having to wade
through page after page of matches for the word
"car." Once again, we scan the search results for new synonyms. (I ran
across automotive, racing, vehicle, and motor.) Make a note of any new bolded synonyms and subtract them from the
query (e.g., ~car -car -automotive
-racing -vehicle -motor) until you hit
Google's 10-word limit ["The
10-Word Limit" in Chapter 1], after which Google
starts ignoring any additional words that you tack on.

In the end, you'll have compiled a goodly list of
synonyms, some of which you'd not have found in your
typical thesaurus thanks to Google's algorithmic
approach to synonyms.

2.3.1. The Code

If you think this all sounds a little tedious and more in the job
description of a computer program, you'd be right.
Here's a short Python script to do all the iteration
for you. It takes in a starting word and spits out a list of synonyms
that it accrues along the way.

You'll need the PyGoogle [Hack #98] library to provide an interface to the Google API.

#!/usr/bin/python
# Available at http://www.aaronsw.com/2002/synonyms.py
import re
import google # get at http://pygoogle.sourceforge.net/
sb = re.compile('<b>(.*?)</b>', re.DOTALL)
def stripBolds(text, syns):
for t in sb.findall(text):
t = t.lower( ).encode('utf-8')
if t != '...' and t not in syns: syns.append(t)
return syns
def findSynonyms(q):
if ' ' in q: raise ValueError, "query must be one word"
query = "~" + q
syns = []
while (len(query.split(' ')) <= 10):
for result in google.doGoogleSearch(query).results:
syns = stripBolds(result.snippet, syns)
added = False
for syn in syns:
if syn in query: continue
query += " -" + syn
added = True
break
if not added: break # nothing left
return syns
if __name__ == "__main_ _":
import sys
if len(sys.argv) != 2:
print "Usage: python " + sys.argv[0] + " query"
else:
print findSynonyms(sys.argv[1]) Save the code as synonyms.py.

2.3.2. Running the Hack

Call the script on the command line ["How to Run the
Hacks" in the Preface], passing it a starting word
to get it going, like so:

% python synonyms.py car

2.3.3. The Results

You'll get back a list of synonyms like these:

['auto', 'cars', 'car', 'vehicle', 'automotive', 'bmw', 'motor', 'racing', 'van',
'toyota'] Aaron Swartz