Character recognition apparatus using a keyword

A character recognition apparatus comprising:

• a character recognition unit for scanning a document and outputting character candidates corresponding to the characters in a scanned line of characters of the document;
• a correct character string selection unit for selecting a character string from character candidates output from said character recognition unit to provide a selected character string including plural word candidates, said string selection unit selecting a character string such that word candidates within the string are consistent with a vocabulary dictionary and a grammar dictionary;
• a keyword extraction unit for extracting keywords from a character string selected by said correct character selection unit based on occurrence of a certain word in the subject matter of the document scanned by said character recognition unit;
• a BUNSETSU extraction unit for selecting BUNSETSU candidates from among a character string selected by said correct character string selection unit, and
• a BUNSETSU evaluation unit for evaluating BUNSETSU selected by said BUNSETSU extraction unit from within a character string selected by said selection unit, and wherein
• when the document contains Japanese language characters,
• said BUNSETSU evaluation unit utilizes a keyword output from said keyword extraction unit in evaluating a BUNSETSU candidate selected by said BUNSETSU extraction unit.

A character recognition apparatus comprising:

• a character recognition unit for scanning a document containing Japanese language characters and outputting character candidates corresponding to the characters in a scanned line of characters in the document;
• a word extraction unit for determining a set of word candidates from said character candidates which word candidates are consistent with a word dictionary;
• a BUNSETSU extraction unit for determining BUNSETSU candidates from said set of word candidates which BUNSETSU candidates are consistent with a grammar dictionary;
• a BUNSETSU evaluation unit for evaluating BUNSETSU candidates determined by said BUNSETSU extraction unit based on grammatical and contextual correctness of said BUNSETSU candidates and outputting evaluated BUNSETSU candidates;
• a BUNSETSU selection unit for selecting from among evaluated BUNSETSU candidates output from said BUNSETSU evaluation unit and outputting at least one corrected character string;
• a keyword extraction unit for extracting keywords from a corrected character string output from said BUNSETSU selection unit, said keyword extraction unit extracting keywords based on the subject matter content of the document scanned by said character recognition unit;
• a partial keyword accord unit for providing additional keywords which are partially accordant with said keywords extracted by said keyword extraction unit;
• and
• a word candidate adding unit for adding partially accordant words provided by said partial keyword accord unit to said keywords extracted by said keyword extraction unit.

A character recognition apparatus comprising:

• a character recognition unit for scanning a document containing Japanese characters and outputting character candidates corresponding to the characters in a scanned line of characters of the document;
• a word extraction unit for determining a set of word candidates form said character candidates which word candidates are consistent with a word dictionary;
• a phrase extraction unit for determining phrase candidates from said set of word candidates which phrase candidates are consistent with a grammar dictionary;
• a phrase evaluation unit for evaluating said phrase candidates based on said grammatical and contextual correctness of said phrase candidates and outputting evaluated phrase candidates;
• a phrase selection unit for selecting among evaluated phrase candidates output from said phrase evaluation unit and outputting at least one corrected character string;
• a keyword extraction unit for extracting keywords from corrected character string output from said phrase selection unit, said keyword extraction unit extracting keywords based on the subject matter of the document scanned by said character recognition unit;
• a partial keyword accord unit for providing additional keywords which are partially accordant with said keywords extracted by said keyword extraction unit;
• and
• a word candidate adding unit for adding said partially accordant words provided by said partial keyword accord unit to said keywords extracted by said keyword extraction unit.