| 摘要 |
Disclosed are a system for classifying emotion and a method for classifying emotion strengthened against an orthographical error in a sentence when analyzing the emotion of the sentence. The system for classifying the emotion according to the present invention includes: a dictionary generating part for extracting a term for classifying the emotion of a predicate and a substantive from a learning sentence based on a word phrase pattern and initial/medial phoneme pattern to generate a dictionary having the extracted term; a term extracting part for extracting at least one term by searching the dictionary with respect to an inputted sentence; a term selecting part for searching a term database (DB) with respect to the extracted individual term to select the term successfully searched therefrom; a weighted value querying part for querying a weighted value DB with respect to the selected individual term to request a weighted value; and an emotion classifying part for inputting the weighted value of the each term into a learning model based on a probability distribution algorithm of dichotomized classification, and receiving a result of positiveness or negativeness to classify as an emotion. According to the present invention, the result of the emotion classification, which is effective with respect to a short sentence such as a sentence on an internet having the orthographical error, is provided. |
