Parsing system
View Patent ↗A system of parsing unstructured or partially structured data is provided, specifically including name and address data in any language or script. The system processes at least portions of the data in an incremental manner, preferably using multiple parsing steps, with each parsing step performed by consulting an inference engine which uses an inference strategy. A knowledge base is used with the inference engine to analyze data at one or more levels of analysis. The knowledge base can optionally include a lexico-grammatical analysis level, and orthographic level, a semantic level, a contextual level.
1. A system for parsing unstructured, or partially structured, name and address data in any language and script so as to produce structured name and address data; said system comprising:
a. Knowledge base which analyses said unstructured or partially structured, name and address data at a number of levels, including at a lexicogrammatical level and an orthographic level and a semantic level and a contextual level, wherein linguistic theory enables the complete representation of all possible forms of said data;
b. Computer readable medium, wherein said knowledge base is stored upon said computer readable medium;
c. Knowledge base using a knowledge representation language wherein said knowledge base can build a complete representation of all possible forms of said data provided in any language and script, from any country;
d. Knowledge base containing a knowledge representation layer, a knowledge base management layer, a language programming interface layer and a language inference layer containing an inference engine;
e. System processing at least portions of said unstructured or partially structured name and address data in an incremental manner, involving multiple parsing steps, whereby at each parsing step said inference engine utilizes an inference strategy, represented within a number of inference modules that reason about the language knowledge based on knowledge instances maintained in the knowledge base management layer, to build a body of executable knowledge about patterns contained in said data, thus producing more structured data that may be submitted to further parsing steps until fully structured data is produced.