• Get source text from internet or from user's local disk. if the source text is HTML or XML, strip off all the tags
• Tokenize the text based on user's choice. For example, if user wants all the word pairs, the token is word. If user wants word pairs not spanning sentences, the token will be sentence, then word
• Get all the word pairs, then filter all the pairs not needed by user.
• Sort and count the word pairs based on user requirement.
• Generate output format

Way of Using

• Enter a valid URL in the URL field or enter a local path to upload text
• Select one of the word pairs you want to get and then select/enter the corresponding text/selection if necessary
• Select sorting criterion
• Select output format
• If you want the results displayed in the same window with taporware interface, uncheck the check box - "Open results in new window"
• Finally, click the "Submit" button

Word Pair Finder

Description

History

Pseudocode

1. Find the top X high frequency words. Exclude stop words.
2. Find high frequency pairs that include the original list. Thus it would find high frequency collocating pairs.
   1. To do this it would find the collocates immediately to the left or right Y places
3. It would generate a list of the pairings sorted first by the key high-freq word and then by the frequency of the collocations

To Do

• Talk to Lian to see if it can be done.

-- GeoffreyRockwell - 26 Jun 2006