발간년도 : [2021]
| 논문정보 |
|
| 논문명(한글) |
[Vol.16, No.2] Accuracy Improvement of Movie Recommender System Using Word2Vec and Two-Channel Convolutional Neural Networks |
|
| 논문투고자 |
Boo Sik Kang |
|
| 논문내용 |
Improving the prediction accuracy of product recommender systems is one of the main topics in the field of recommender systems. Recent research on improving prediction accuracy using Word2Vec, which has been widely utilized in text analysis, has been presented. Furthermore, research on improving prediction accuracy using various neural networks has been presented. This study proposes a method that can improve the prediction accuracy of a movie recommender system combining Word2Vec and two=channel convolutional neural networks. Using Word2Vec, we express inter-user associations as multi-dimensional vector space. It also expresses inter-movie associations as multi-dimensional vector space. Furthermore, to infer user rating propensity, we use user-specific mean rating information to find a second user-specific multidimensional vector space. Similarly, we find a second movie-specific multidimensional vector space. User vector and movie vector are organized into two channels and learned through two-channel convolutional neural networks. Convolutional neural networks are composed of user convolutional models and movie convolutional models, respectively, and are connected in a fully connected layer. We experiment with filmtrust data to evaluate the predictive accuracy performance of the proposed model. Experiments show that the proposed technique improves the prediction accuracy compared to the existing prior studies. |
|
| 첨부논문 |
|
|
|
|
|
자료실 > 논문지 
