System and method for detecting the occureances of irrelevant and/or low-score strings in community based or user generated content

A system for posting user generated content and detecting profane/objectionable content, comprising:

• one or more microprocessors and memory;
• a computer server for user generated content, servers executing on the one or more microprocessors and memory;
• and
• a profanity filter executing on the computer server which includes:
• a disallowed words dictionary which identifies a plurality of disallowed words;
• a disallowed variants dictionary which identifies a plurality of variants of the plurality of disallowed words in the disallowed words dictionary wherein each disallowed variant corresponds to a modified version of one of the plurality of disallowed words;
• and
• a misuse table which specifies morphological variants of characters and morphological variants of character pairs;
• wherein the computer server is configured to generate, during a preprocessing phase, the disallowed variants dictionary from the disallowed words dictionary, using the misuse table, by, for each of the plurality of words of the disallowed words dictionary, iteratively, for each character of each disallowed word of the plurality of disallowed words, performing the steps of:
• utilizing the misuse table to identify character variants for said each character;
• generating a first set of disallowed variants by substituting the character of the disallowed word with each of the one or more character variants identified with the misuse table;
• adding each disallowed variant of the first set of disallowed variants to the disallowed variants dictionary on said computer server;
• utilizing the misuse table to identify character pair variants for a character pair including said each character in combination with an immediately preceding character of the disallowed word;
• generating a second set of disallowed variants by substituting the character pair with character pair variants identified with the misuse table;
• and
• adding each disallowed variant of the second set of disallowed variants to the disallowed variants dictionary on said computer server;
• wherein the computer server receives, during runtime, a plurality of messages from users;
• and
• wherein, for each of a plurality of words in each of the plurality of messages received during runtime, the computer server checks said each word against the disallowed words dictionary and disallowed variants dictionary by directly looking up said each word in the disallowed words dictionary and disallowed variants dictionary, and determines that said each word is disallowed when there is a match;
• wherein the computer server calculates a profanity score for each message of the plurality of messages based on disallowed words determined in said each message;
• and
• wherein the computer server one of allows or disallows each message of the plurality of messages to be posted on the computer server based on the profanity score for said each message.

A computer-implemented method for posting user generated content and detecting profane/objectionable content, comprising:

• providing, at a computer server for user generated content that includes one or more microprocessors and memory, a profanity filter executing on the computer server which includes:
• a disallowed words dictionary which identifies a plurality of disallowed words;
• a disallowed variants dictionary which identifies a plurality of variants of the plurality of disallowed words in the disallowed words dictionary wherein each disallowed variant corresponds to a modified version of one of the plurality of disallowed words;
• and
• a misuse table which specifies morphological variants of characters and morphological variants of character pairs;
• generating, during a preprocessing phase, the disallowed variants dictionary from the disallowed words dictionary using the misuse table, by, for each of the plurality of words of the disallowed words dictionary, iteratively, for each character of each disallowed word of the plurality of disallowed words, performing the steps of:
• utilizing the misuse table to identify character variants for said each character;
• generating a first set of disallowed variants by substituting the character of the disallowed word with each of the one or more character variants identified with the misuse table;
• adding each disallowed variant of the first set of disallowed variants to the disallowed variants dictionary on said computer server;
• utilizing the misuse table to identify character pair variants for a character pair including said each character in combination with an immediately preceding character of the disallowed word;
• generating a second set of disallowed variants by substituting the character pair with character pair variants identified with the misuse table;
• and
• adding each disallowed variant of the second set of disallowed variants to the disallowed variants dictionary on said computer server;
• receiving, by the computer server during runtime, a plurality of messages from users;
• and
• for each of a plurality of words in each of the plurality of messages received during runtime, performing the following steps in order:
• removing stop words from the message;
• checking each remaining word against the disallowed words dictionary by directly looking up said each remaining word in the disallowed words dictionary and determining that said each remaining word is disallowed when there is a match;
• checking said each remaining word not matched in the disallowed word dictionary against an allowed words dictionary by directly looking up said each word in the allowed words dictionary and determining that said each remaining word is allowed when there is a match;
• and
• checking said each remaining word not matched in the disallowed word dictionary and allowed words dictionary against the disallowed variants dictionary by directly looking up said each remaining word in the disallowed variants dictionary and determining that said each remaining word is disallowed when there is a match;
• calculating a profanity score for each message of the plurality of messages based on allowed words and disallowed words determined in said each message;
• and
• posting or not posting each message of the plurality of messages on the computer server based on the profanity score calculated for said each message.