Information extraction and annotation systems and methods for documents

A method for annotating documents using a context and contiguity analysis module having a processor and a memory for storing logic, the processor executing the logic to perform the method, comprising:

• receiving training documents, the training documents having annotated words;
• identifying a predetermined number of characters preceding and following each annotated word for each of the training documents to determine a context for each of the annotated words;
• dividing a target document into n-gram strings;
• identifying overlapping sequences of characters between the n-gram strings of the target document and the training documents;
• assigning an annotation to each word in the target document when the overlapping sequences between the n-gram strings of the target document and the training documents meet or exceed a threshold;
• for each annotated word in the training documents, determining a distance between the annotated word and one or more document features;
• calculating an averaged location for each annotated word in the training documents, the averaged location defining an averaged placement of each annotated word within the training documents;
• identifying matches between a distance and averaged location of words of the target document and distances and the averaged location of at least one of the annotated words of the training documents;
• and
• based at least in part on the identified matches, assigning an annotation to a word of the target document that matches the at least one of the annotated words of the training documents.

A context and contiguity analysis module, comprising:

• a processor;
• and
• a memory for storing logic, the processor executing the logic to:
• receive training documents, the training documents having annotated words;
• identify a predetermined number of characters preceding and following each annotated word for each of the training documents to determine a context for each of the annotated words;
• perform an alignment of an annotated word and its context with characters in a target document, by dividing the target document into n-gram strings and identifying overlapping sequences of characters between the n-gram strings of the target document and the training documents,
• wherein the target document comprises a document that has been processed using optical character recognition (OCR), the document comprising OCR errors that distort at least a portion of the characters of the target document;
• for each annotated word in the training documents, determine a distance between the annotated word and one or more document features;
• calculate an averaged location for each annotated word in the training documents, the averaged location defining an averaged placement of each annotated word within the training documents;
• identify matches between a distance and averaged location of words of the target document and distances and the averaged location of at least one of the annotated words of the training documents;
• and
• in response to performing the alignment, assign an annotation to a word in the target document when a common sequence between the target document and the training documents is found, the annotation comprising an annotation of an annotated word of the training documents included in the common sequence.