Experiential parser

Abstract

Experiential parsing (EP) is a technique for natural language parsing that falls into the category of dependency parsing. EP supports applications that derive meaning from chat language. An experiential language model parses chat data, and uses documented experiences with language without using automatic natural language processing (NLP) methods. A descriptive grammar is built at word level rather than a prescriptive grammar at phrase level. The experiential model is designed to understand that word “A” associates with word “B” by function “C”. The experiential model understands the relationship between words, independent of whether or not the overall phrase structure is grammatical. A high accuracy of producing the syntactic roles (such as main verb, direct object, etc.) is attained even when confronted with a variety of agrammatical inputs.

What is claimed: 
     
         1 . A system comprising:
 a parser that receives an input comprising a plurality of words, generates a plurality of collocate links among the plurality of words, generates a plurality of environment links among the plurality of words, removes at least one of the plurality of the collocate links or the environment links, and provides a processed language output; and   a memory.   
     
     
         2 . The system of  claim 1 , wherein the memory comprises a misspelling database that comprises a plurality of words and associated correct spellings of the plurality of words and alternate spellings of the plurality of words. 
     
     
         3 . The system of  claim 2 , wherein the memory further comprises a word form database comprising metadata associated with a plurality of word forms of the plurality of words, and a usage data database comprising a plurality of collocates for the plurality of words and a plurality of syntax environments for the plurality of words. 
     
     
         4 . The system of  claim 3 , wherein the memory further comprises a usage data database comprising a plurality of collocates for the plurality of words and a plurality of syntax environments for the plurality of words. 
     
     
         5 . The system of  claim 1 , wherein the input comprises one of chat data, a keyword search, a non-conversational phrase, or a tokenized agrammatical phrase. 
     
     
         6 . The system of  claim 1 , wherein the processed language output comprises at least one of an action of the input, an object of the input, an actor of the input, an indirect object of the input, or one or more modifying details of the action, the object, the actor, or the indirect object. 
     
     
         7 . The system of  claim 1 , wherein the parser is configured to provide the processed language output to a chatbot. 
     
     
         8 . The system of  claim 1 , wherein the parser is configured to perform a repeating elimination process on the at least one of the plurality of the collocate links or the environment links. 
     
     
         9 . The system of  claim 8 , wherein a first pass of the repeating elimination process processes the collocate links and the environment links separately. 
     
     
         10 . The system of  claim 1 , further comprising a lemmatizer that receives a plurality of word possibilities and checks the plurality of word possibilities against a word form database to adjust the plurality of word possibilities, and provides the adjusted plurality of word possibilities to the parser as the plurality of words. 
     
     
         11 . A method comprising:
 receiving, at a parser, an input comprising a plurality of words;   generating, at the parser, a plurality of collocate links among the plurality of words;   generating, at the parser, a plurality of environment links among the plurality of words;   removing, at the parser, at least one of the plurality of the collocate links or the environment links; and   providing, by the parser, a processed language output.   
     
     
         12 . The method of  claim 11 , wherein the input comprises one of chat data, a keyword search, a non-conversational phrase, or a tokenized agrammatical phrase. 
     
     
         13 . The method of  claim 11 , wherein the processed language output comprises at least one of an action of the input, an object of the input, an actor of the input, an indirect object of the input, or one or more modifying details of the action, the object, the actor, or the indirect object. 
     
     
         14 . The method of  claim 11 , wherein providing the processed language output comprises providing the processed language output to a chatbot. 
     
     
         15 . The method of  claim 11 , further comprising performing a repeating elimination process on the at least one of the plurality of the collocate links or the environment links. 
     
     
         16 . The method of  claim 15 , wherein a first pass of the repeating elimination process processes the collocate links and the environment links separately. 
     
     
         17 . The method of  claim 11 , further comprising receiving, at a lemmatizer, a plurality of word possibilities and checking the plurality of word possibilities against a word form database to adjust the plurality of word possibilities, and providing the adjusted plurality of word possibilities to the parser as the plurality of words. 
     
     
         18 . A system comprising:
 at least one processor; and   a memory storing instructions that when executed by the at least one processor cause the at least one processor to:
 receive an input comprising a plurality of words; 
 generate a plurality of collocate links among the plurality of words; 
 generate a plurality of environment links among the plurality of words; 
 remove at least one of the plurality of the collocate links or the environment links; and 
 provide a processed language output. 
   
     
     
         19 . The system of  claim 18 , wherein the instructions further comprise instructions that when executed by the at least one processor cause the at least one processor to:
 performing a repeating elimination process on the at least one of the plurality of the collocate links or the environment links.   
     
     
         20 . The system of  claim 18 , wherein the instructions further comprise instructions that when executed by the at least one processor cause the at least one processor to:
 receive a plurality of word possibilities and check the plurality of word possibilities against a word form database to adjust the plurality of word possibilities, and provide the adjusted plurality of word possibilities as the plurality of words.