Recently, the number of people using social network sites(SNS) is increasing rapidly as the penetration rate of smartphones has increased.
Although SNS was used to build and maintain relationships in the beginning, but it has recently been used to exp...
Recently, the number of people using social network sites(SNS) is increasing rapidly as the penetration rate of smartphones has increased.
Although SNS was used to build and maintain relationships in the beginning, but it has recently been used to express personal interests directly and indirectly, such as posting personal thoughts or favorite photos.
Therefore, analyzing SNS is good data to extract personal tendencies and interests, and these analysis results can be used in many ways, including marketing and recommendation system.
In this thesis, contents of SNS are summarized using algorithm of Text Rank, and classify the contents using CNN(Convolutional Neural Network) to extract individual interests.
While Conventional CNN(Convolutional Neural Network) - based classification techniques show an average of 70.4% accuracy, the proposed method has significantly improved accuracy to about 82.8% and appears to be more efficient in extracting individual interests.