Apparatus and method for detecting sentence boundaries

An apparatus for detecting sentence boundaries, comprising:

• a sentence boundary candidate extracting unit extracting a sentence boundary candidate from an input document;
• a document context analyzing unit extracting features from information on preceding and following contexts of the sentence boundary candidate, the features being used in a statistical algorithm;
• a sentence boundary candidate classifying unit classifying whether the sentence boundary candidate is a sentence boundary or not, using the features and the statistical algorithm;
• an error post-processing unit correcting an error based on a predetermined rule with respect to the classified sentence boundary candidate;
• and
• a sentence generating unit extracting sentence units from the document based on a result of the classification of whether the sentence boundary candidate is a sentence boundary or not corrected by the error post-processing unit.

An apparatus for detecting sentence boundaries, comprising:

• a sentence boundary candidate extracting unit extracting a sentence boundary candidate from an input document;
• a document context analyzing unit extracting features from information on preceding and following contexts of the sentence boundary candidate, the features being used in a statistical algorithm;
• a sentence boundary candidate classifying unit classifying whether the sentence boundary candidate is a sentence boundary or not, using the features and the statistical algorithm;
• and
• a sentence generating unit extracting sentence units from the document based on a result of the classification of whether the sentence boundary candidate is a sentence boundary or not,
• wherein the sentence boundary candidate comprises at least one of punctuation marks used in sentence termination, sentence-ending elements of predicates used in the sentence termination, and other sentence-terminating syllables extracted from a list of all syllables used in the sentence termination and pre-extracted from learning data tagged with sentence termination symbols.

A method for detecting sentence boundaries, comprising:

• extracting a sentence boundary candidate from an input document;
• extracting features used in two or more statistical algorithms from information on preceding and following contexts of the sentence boundary candidate;
• classifying whether the sentence boundary candidate is a sentence boundary or not, using the features and the two or more statistical algorithms;
• and
• extracting sentence units from the document based on a result of the classification of whether the sentence boundary candidate is a sentence boundary or not.

A method for detecting sentence boundaries, comprising:

• extracting a sentence boundary candidate from an input document;
• extracting features from information on preceding and following contexts of the sentence boundary candidate, the features being used in a statistical algorithm;
• classifying whether the sentence boundary candidate is a sentence boundary or not, using the features and the statistical algorithm;
• correcting an error based on a predetermined rule with respect to the classified sentence boundary candidate;
• and
• extracting sentence units from the document based on a result of the classification of whether the sentence boundary candidate is a sentence boundary or not corrected in the correcting of an error.