WebService::GoogleHack - Perl package that ties together all GoogleHack modules (Webservice::GoogleHack::Search, Webservice::GoogleHack::Spelling, Webservice::GoogleHack::Rate, and Webservice::GoogleHack::Text) to implement Natural Language Processing techniques that use the World Wide Web as a source of information. Use this package to access all the functionality of GoogleHack.
use WebService::GoogleHack; my $google = new WebService::GoogleHack; #Initializing the object to the contents of the configuration file # API Key, GoogleSearch.wsdl file location. $google->initConfig("initconfig.txt"); #Printing the contents of the configuration file $google->printConfig(); #Measure the semantic relatedness between the words "white house" and #"president". $measure=$google->measureSemanticRelatedness1("white house","president"); print "\nRelatedness measure between white house and president is: "; print $measure."\n"; #Going to search for words that are related to "toyota" and "ford" my @terms=(); push(@terms,"toyota"); push(@terms,"ford"); #The parameters are the search terms, number of web page results to look #at, the number of iterations,output file and the "true" indicates that the #diagnostic data should be stored in the file "results.txt" $results=$google->Algorithm1(\@terms,10,25,1,"results.txt","true"); print $results;
WebService::GoogleHack is a PERL package that interacts with the Google API, and implements basic functions that allow the user to interact with Google and retrieve results in an easy to use format. GoogleHack also implements and extends a number of Natural Language Processing by using the World Wide Web as a source of information.
Some of the features are:
* Issue queries to Google (WebService::GoogleHack, WebService::GoogleHack::Search) * Retrieve Spelling suggestions from Google (WebService::GoogleHack, WebService::GoogleHack::Spelling) * Find the Pointwise Mututal Information (PMI) measure between two words (WebService::GoogleHack,WebService::GoogleHack::Rate) * Given a paragraph find if the paragraph has a positive or negative semantic orientation.(WebService::GoogleHack,WebService::GoogleHack::Rate) * Given a set of words along with a positively oriented word such as "excellent" and a negatively oriented word such as "poor", find if the word has a positive or negative semantic orientation.(WebService::GoogleHack,WebService::GoogleHack::Rate) * Given a set of phrases along with a positively oriented word such as "excellent" and a negatively oriented word such as "poor", predict if the given phrases are positive or negative in sentiment.(WebService::GoogleHack,WebService::GoogleHack::Rate) * Given two or more words finds a set of related words. (WebService::GoogleHack)
WebService::GoogleHack::Text - GoogleHack uses this module to manipulate text retrieved from the web (Get n-word sentences, words,and parse HTML etc).
WebService::GoogleHack::Search - GoogleHack uses this module to query Google.
WebService::GoogleHack::Rate - GoogleHack users this module to implement some of the Sentiment Classification algorithms.
WebService::GoogleHack::Spelling - GoogleHack uses this module to query Google for spelling suggestions.
1) Google API (http://www.google.com/apis/)
2) Brill Tagger (If using Sentiment Classification stuff)
Installation file and instructions @ : http://www.cs.jhu.edu/~brill/RBT1_14.tar.Z Instructions also available in GoogleHack INSTALL file.
3) Required PERL Modules
SOAP::Lite; Set::Scalar; Text::English; LWP::Simple; URI::URL; LWP::UserAgent; HTML::LinkExtor; Data::Dumper;
These are the GoogleHack functions that are common to all sort of operations. These functions are used to create and initialize GoogleHack objects.
Purpose: This function creates an object of type GoogleHack and returns a blessed reference.
returns: A blessed reference to a GoogleHack object.
Purpose: This function is used to read the configuration file containing information such as the Google-API key, the base directory path, and the path to the Brill Tagger. The configuration file is in the WebService/GoogleHack/Datafiles directory.
This function must be called in order to initialize the GoogleHack object.
Valid arguments are :
configLocation
string. Location of the configuration file.
returns : Returns an object which contains the parsed information.
Purpose: This function is used to print the information read from the configuration file
No arguments.
This set of functions deal with the problem of finding sets of related words by using the World Wide Web as a source of information.
Purpose: This function is used to measure the relatedness between two words.
Formula used: log(hits(w1)) + log(hits(w2)) - 2 * log(hits(w1w2))
searchString1
string. The search string which can be a phrase or word
searchString2
Returns: Returns the object containing the relatedness measure.
Formula used: log(w1w2/(w1+w2))
Formula used: log( hits(w1w2) / (hits(w1) * hits(w2)))
Purpose:Given two or more words, this function tries to find a set of related words. This is the Google-Hack baseline algorithm 1. For example, given the two words gun and pistol, an example of an expanded set of related words would be,
{laser,paintball, case,bullet, machine gun, rifle} etc.
The features of Initial Approach (Algorithm 1) is given below - Frequency Based - Accepts only 2 terms - Results also contain only unigrams - A frequency cutoff is used - Stop words and web stop words are removed.
searchTerms
string. The array of search terms (Can only be a word). =item *
N
number. The number of web pages results to be looked at.
C
number. Words occuring less than the frequencyCutoff would be excluded from results.
I
number. The number of iterations that you want the function to search and build cluster on.
trace.
string. The location where the file containing the retreived information has to be stored.
html.
bool. If set to "true" then the results returned by the algorithm is in HTML format. If not "true", the results are in plain text format.
returns : Returns an html or text version of the results.
Purpose:Given two or more words, this function tries to find a set of related words. This is the Google-Hack algorithm 2.
The features of Second Approach (Algorithm 2) is given below - Accepts more than 2 terms - Has 3 relatedness scores - Accepts unigrams and 2-word collocation as input - Results also contain 2-word collocations - A score cutoff is also included along with frequency cutoff - A bigram cutoff is also included. - Stop words and web stop words are removed. - Stop phrases and web stop phrases are removed.
F
BC
number. Bigrams occuring less than the bigramCutoff would be excluded from results.
S
number. Takes on the values 1,2 or 3 indicating the relatedness measure to be used.
SC
number. Words and Bigrams with relatedness score greater than the scoreCutOff would be excluded from results.
Purpose:Given a set of search terms, this function will retreive the resulting URLs from Google, it will then follow those links, and retrieve the text from there. Once all the text is collected, the function finds the intersecting or co-occurring words between the top N results. This function is basically used by the function Algorithm1.
string. An array which contains each search term (It can only be a word not phrase).
number. The number of web pages to be looked at.
number. The current iteration number.
NT
number. The number of search terms in the initial set.
BI
number. The bigram cutoff.Set to 0 to exclude bigrams.
string. The location of the trace file.
returns : Returns nothing.
This set of functions deal with sentiment classification. The functions include the PMI-IR, and some other similar functions that try to classify if a given word or phrase is positively or negatively oriented in its sentiment.
Purpose: This function tries to predict the semantic orientation of a paragraph of text. The semantic orientation of a paragraph is calculated according to the paper "Thumbs Up or Thumbs Down? Semantic Orientation Applied to Unsupervised Classification of Reviews" By Peter Turney. The difference between Peter Turneys implementation of the PMI-IR algorithm and the implementation of the PMI-IR algorithm in Google Hack is small, but very important.
In Peter Turneys implementation, the PMI-IR algorithm uses the search engine Alta Vista. However, in Google-Hack, we are using Google as our search engine. More importantly, AltaVista provides a "near" operator which the original PMI=IRuses, however, Google does not. Hence, we are using the "AND" operator.
rfile
string. The location of the review file
posInf.
string. Positive inference such as excellent
negInf.
string. Negative inference such a poor
string. The location of the trace file. If a file_name is given, the results are stored in this file
Returns : the PMI measure and the prediction which is 0 or 1.
Purpose:Given an file containing text, this function tries to find the positive and negative words. The formula used to calculate the sentiment of a word is based on the PMI-IR formula given in Peter Turneys paper.
(hits(word AND "excellent") hits (poor)) log2 ---------------------------------------- (hits(word AND "poor") hits (excellent))
For more information refer the paper, "Thumbs Up or Thumbs Down? Semantic Orientation Applied to Unsupervised Classification of Reviews" By Peter Turney.
infile
string. The input file
posInf
string. A positive word such as "Excellent"
string. A negative word such as "Bad"
string. Set to "true" if you want the results to be HTML formatted
string. Set to a file if you want the results to be written to the given filename.
Purpose:Given an file containing text, this function tries to find the positive and negative phrases. The formula used to calculate the sentiment of a phrase is based on the PMI-IR formula given in Peter Turneys paper.
(hits(phrase AND "excellent") hits (poor)) log2 ------------------------------------------ (hits(phrase AND "poor") hits (excellent))
Purpose: This is function is used to retrieve a spelling suggestion from Google
searchString
string. Need to pass the search string, which can be a single word
Returns: Returns suggested spelling if there is one, otherwise returns "No Spelling Suggested":
Use this function to issue queries to Google.
Purpose: This function is used to query googles
string. Need to pass the search string, which can be a single word or phrase, maximum ten words
num_results
integer. The number of results you wast to retrieve, default is 10. Maximum is 1000.
Returns: Returns a GoogleHack object containing the search results.
This set of functions deal with retrieving text from the World Wide Web. Basically, the user can use these functions to retrieve sentences, words, or phrases that occur in web pages (In snippets, cached web pages, links etc.
Purpose: Given a search word, this function tries to retreive the text surrounding the search word in the retrieved snippets.
string. The search string which can be a word or phrase
numResults
string. The number of results to be processed from google.
trace_file.
proximity
string. The number of words surrounding the searchString (Not Implemented) yet
returns : Returns an object which contains the parsed information
Purpose: Given a search word, this function tries to retreive the text surrounding the search word in the retrieved CACHED Web pages. It basically does the search and passes the search results to the WebService::GoogleHack::Text::getCachedSurroundingWords function.
returns : Returns a hash with the keys being the words and the values being the frequency of occurence.
Purpose: Given a search word, this function tries to retreive the sentences in the snippet.It basically does the search and passes the search results to the WebService::GoogleHack::Text::getSnippetSentences function
returns : Returns an array of strings.
Purpose: Given a search word, this function tries to retreive the sentences in the cached web page.
returns : Returns a hash which contains the parsed sentences as values and the key being the web URL.
Purpose:Given two search words, this function tries to retreive the common text/words surrounding the search strings in the retrieved snippets.
stemmer.
bool. Porter Stemmer on or off.
returns : Returns a hash which contains the intersecting words.
Purpose:Given a search string, this function retreive the top frequency words , and does a search on those words, and builds a list of words that can be regarded as a cluster of related words.
*=item *
iterations
returns : Returns a set of words as a hash.
Purpose:Given two search strings, this function retreive the snippets for each string, and then finds the intersection of words, and then repeats the search with the intersection of words.
returns : Returns a hash which contains the intersecting words as keys and the values being the frequency of occurence.
Purpose:Given a search string, this function will retreive the resulting URLs from Google, follow those links, and retrieve the text from there. The function will then clean up the text and store it in a file along with the URL, Date and time of retrieval.The file will be stored under the name of the search string.
string. The search string which can be a word or phrase.
path_to_data_directory.
string. The location where the file containing the retrived information has to be stored.
Pratheepan Raveendranathan, <rave0029@d.umn.edu>
Ted Pedersen, <tpederse@d.umn.edu>
WebService::GoogleHack home page - http://google-hack.sourceforge.net
Pratheepan Raveendranathan - http://www.d.umn.edu/~rave0029/research
Ted Pedersen - www.d.umn.edu./~tpederse
Google-Hack Maling List <google-hack-users@lists.sourceforge.net>
Copyright (c) 2005 by Pratheepan Raveendranathan, Ted Pedersen
This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version.
This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with this program; if not, write to
The Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
1 POD Error
The following errors were encountered while parsing the POD:
Expected '=item *'
To install WebService::GoogleHack, copy and paste the appropriate command in to your terminal.
cpanm
cpanm WebService::GoogleHack
CPAN shell
perl -MCPAN -e shell install WebService::GoogleHack
For more information on module installation, please visit the detailed CPAN module installation guide.