System and method for searching electronic documents created with optical character recognition
OwnerWilliam Petermaninventor
Life of this patent
Estimating the term…
Abstract
A method, system and computer product for processing search requests in order to compensate for characters and character strings misread during OCR scanning is disclosed. After an alphanumeric search request is entered, the system determines variant words associated with the entered alphanumeric search request according to a predefined table of possible OCR errors, the OCR errors' probability of occurrence and a predefined threshold of probability of occurrences. When the preprocessing is complete, a search engine then uses the variant words to search a database containing OCR scanned documents.
Claims21
A computer based method for associating variant words with an alphanumeric search request utilizing a database of optical character recognition documents, the method comprising:
- obtaining an alphanumeric search request;
- generating at least one variant word associated with the alphanumeric search request according to predefined optical character recognition substitution data, the optical character recognition substitution data including a set of optical character recognition substitutions and a probability of occurrence corresponding to each optical character recognition substitution, wherein the variant word is different from the alphanumeric search request;
- and
- identifying any electronic documents in the database having the at least one generated variant word.
A program storage device readable by a machine, tangibly embodying a program of instructions executable by the machine to perform method steps for searching a document database, the document database comprising optical character recognition documents, the method steps comprising:
- obtaining a search request comprising a target alphanumeric word;
- obtaining a plurality of optical character recognition substitutions corresponding to the target alphanumeric word;
- generating at least one variant word corresponding to the target alphanumeric word, wherein the at least one variant word is different from the target alphanumeric word and is based upon the plurality of optical character recognition substitutions corresponding to the target alphanumeric word;
- and
- identifying the documents in the document database having the at least one generated variant word.
A computer based method for searching a database of electronic documents created by the use of optical character recognition, the method comprising:
- receiving an alphanumeric search request;
- determining at least one variant word associated with the alphanumeric search request according to predefined optical character recognition substitution data, the optical character recognition substitution data including a set of optical character recognition substitutions and a probability of occurrence corresponding to each optical character recognition substitution, wherein the variant word is different from the alphanumeric search request;
- creating a tree data structure having a plurality of substitution nodes, each substitution node corresponding to an element of an optical character recognition substitution corresponding to the alphanumeric search request, the tree data structure having a root node and at least one leaf node, each leaf node linked to the root node by a path corresponding to the leaf node;
- and
- combining the optical character recognition substitutions corresponding to each substitution node on the path corresponding to one of said leaf nodes to create the variant word;
- and
- searching the database of electronic documents according to said at least one determined variant word.
Dependencies are inferred from the claim text.