MetaGen

Files

• metagen.pl (binary)
• metagen.html (this file)
• filter.txt (common words to filter out)
• states.txt (state abreviations and names)
• acronyms.txt (acronyms and their meanings)

Requirements

• Linux, UNIX, or Windows
• PERL 5.003_07
• Bundle::LWP - libwww PERL module
• HTML::TokeParser - Alternative HTML::Parser interface PERL module
• Optional:
• Local Search - MetaGen will use pertinent results to enhance weighting.

Description

Setup

• Copy all of the metagen files to /cgi-bin/daboo/metagen/ directory.
• Create tmp directory to following location: /cgi-bin/daboo/metagen/tmp/
• CHMOD everything to 755

Usage

• http://yourdomain.com/cgi-bin/daboo/metagen/metagen.pl from browser
• Enter name of page you want to create the tags for.
• Remove any non-pertinent words or phrases from the Title, Keywords, and Description until the text will fit without scrolling. This guarantees that there will be a maximum of 250 characters in the Title, a maximum of 1000 characters in the Keywords tag, and a maximum of 1000 characters in the Description tag. Upon pasting into your HTML be sure to remove any line feeds that may carry over. The Title should be it's own line, as should the Keywords and Description MetaTags.
• You may wish to add and remove entries from filter.txt to tailor the MetaGen results to be more to your liking. Just enter one word per line and the next time you run MetaGen, the word will be filtered out.
• You may also create an exclude file for a specific domain by creating a text document in the metagen directory using the same name that you enter in the "Replace 1st word of Title with:" Example: I use Dienhart, I create a document called Dienhart_omit.txt and add all words and phrases I wish to exclude from only my site there. Place one word or phrase on each line, it is always a good idea to add an extra line feed at the end of the file.
• You may also create a text file using the "Replace 1st word of Title with:" as the name of the file, with extension.txt containing manually weighted words and Phrases. If these words or phrases are found in you document they will receive more weight than those not included in the list. To use, place the file in the same directory with the script. For Dienhart, the name of my file is Dienhart.txt, this way you may have a separate weight file for each domain. Place one word or phrase on each line, it is always a good idea to add an extra line feed at the end of the file.

Notes

History

• Initial Release

• fixed bug that filtered out phrases that ended with a number.

• Updated Internal prefix and extension filters to weed out more useless information so I don't have to.
• Minor improvements in filtering. Aggressively updated filter.txt file.

• Modified weight algorithm to return more accurately weighted word and phrase results.

• Removed text links from results, removed graphic text links from results. Added text link text to results. Improved performance.

• Rewrote weighting routine for Keywords tag, much improved. Added manually weighted words file that will add weight to words found in the document that tags are being generated for. Added WordSpy referencing and weighting scheme.

• Removed WordSpy referencing and weighting scheme. Although it works perfectly, It takes to long to return the results, I attribute this to the fact that I am using an ASCII database.

• Refined the Title and Description Tag Generators
• Re-Added the WordSpy routine, This is not publically enabled in this release, as I do not want to support it. This will be included in the Delphi release.
• Added US State abbreviation to US State name and vice versa for all of the tags.
• Added acronym to meaning and vice versa for all of the tags (I am not releasing this at this time).

• Totally disabled Wordspy routine from the source and replaced it with the localanalysis localsearch.txt log file. This should provide more site specific information, rather than the broad range collected from the search engines. Made a few refinements to improve performance and accuracy.

• Removed Word Spy routine from source code. Removed Document compare routine from source code. Improved Title Results. Concentrating on improving the algorithms for automatic generation of the tags, and improving efficiency.