Determining “Owner” of Text Edited by Multiple Users

User 38193 owns 42% (922 / 2171) of the final post
User 2635 owns 28% (625 / 2171) of the final post
User 116 owns 24% (529 / 2171) of the final post
User 745 owns 3% (76 / 2171) of the final post
User 13005 owns 0% (11 / 2171) of the final post
User 18941 owns 0% (5 / 2171) of the final post
User 8562 owns 0% (3 / 2171) of the final post

53 ms

User 38193 owns 42% (922 / 2171) of the final post
User 2635 owns 28% (625 / 2171) of the final post
User 116 owns 24% (529 / 2171) of the final post
User 745 owns 3% (76 / 2171) of the final post
User 13005 owns 0% (11 / 2171) of the final post
User 18941 owns 0% (5 / 2171) of the final post
User 8562 owns 0% (3 / 2171) of the final post

53 ms

#!/usr/bin/env python

import collections
import difflib
import logging
import pprint
import urllib2
import re

class OwnageDeterminer(object):

    add_coefficient = 1
    remove_coefficient = .5

    def __init__(self, edits):
        self.edits = edits
        self.counts_by_username = {}

    def __call__(self):
        edits, counts_by_username = self.edits, self.counts_by_username
        for i, edit in enumerate(edits):
            username = edit['username']
            unique_counts = {'added': 0, 'removed': 0}
            existing_text = edits[i-1]['text'] if i > 0 else ''
            new_text = edits[i]['text']
            for char_diff in difflib.ndiff(existing_text, new_text):
                if char_diff.startswith('+'):
                    unique_counts['added'] += 1
                elif char_diff.startswith('-'):
                    unique_counts['removed'] += 1
            user_counts = counts_by_username.get(username, collections.defaultdict(int))
            user_counts['removed'] += self.remove_coefficient * unique_counts['removed']
            user_counts['added'] += self.add_coefficient * unique_counts['added']
            counts_by_username[username] = user_counts
        winner = None
        winning_score = 0
        score_by_username = {}
        for username, counts in counts_by_username.iteritems():
            score = counts['removed'] + counts['added']
            if score > winning_score:
                winner = username
                winning_score = score
            score_by_username[username] = score
        logging.debug('Scores: %s', pprint.pformat(score_by_username))
        return winner


if __name__ == '__main__':
    logging.basicConfig(level=logging.DEBUG)
    site = urllib2.urlopen('http://stackoverflow.com/revisions/327973/list')
    contents = site.read()
    regex = re.compile(r'(/revisions/viewmarkup/d+).*?/users/d+/([w-]+)',
                       re.MULTILINE|re.DOTALL)
    revisions = regex.findall(contents)
    print revisions
    edits = []
    for reluri, username in sorted(revisions, key=lambda t: t[0]):
        text = urllib2.urlopen('http://stackoverflow.com{0}'.format(reluri)).read()
        edit = {'username': username, 'text': text}
        edits.append(edit)
    od = OwnageDeterminer(edits)
    print od()

DEBUG:root:Scores: {'blorgbeard': 0.5,
 'dave-markle': 0.5,
 'dbr': 1172.0,
 'gatekiller': 69.5,
 'joseph-ferris': 0.0,
 'lkessler': 0.0,
 'mark-harrison': 592.0,
 'mdb': 3.0,
 'onebyone-livejournal-com': 0.0,
 'paul-oyster': 482.0,
 'rob-wells': 0.0,
 'simucal': 1070.5,
 'skiphoppy': 0.0,
 'thesoftwarejedi': 701.0}
dbr

Break latest revision into a set of sentences.
//Sentence is any text fragment surrounded by punctuation

For each Sentence
    Find which user created that sentence.
    Add 1 to the user who created the sentence
    Add 1 to the number of sentences

For Each user
    % ownership = Count for that user / Number of sentences.

Let n = total revisions
Let m = the revision number of a poster
Let post[it] = array with text of post at revision 'it'
Let votes[it] = votes that revision 'it' received (also add bonus for comments/answers)
value = 0
for (it = m; it < n; ++it) {
  value += (Levenshtein(post[it-1], post[m]) / average_length_post) * (votes[it])
}