Table 8 Evaluation of accuracy using various feature representations
From: Web crawling based context aware recommender system using optimized deep recurrent neural network
Size of training data | Feature representation | DNN | DAE-SR | DAE | DRNN |
|---|---|---|---|---|---|
Complete (100%) | TF–IDF | 92.5 | 93.2 | 94.6 | 95.2 |
TF–IDF + CBOW | 95.2 | 96.8 | 97.5 | 98.3 | |
TF–IDF + CBOW + Skip-gram | 96 | 97.3 | 98.4 | 99.5 | |
80% | TF–IDF | 91.3 | 92.7 | 93 | 94.3 |
TF–IDF + CBOW | 92.8 | 94 | 95.8 | 97.6 | |
TF–IDF + CBOW + Skip-gram | 95.2 | 96.2 | 97.2 | 98.2 | |
60% | TF–IDF | 90.2 | 91.3 | 92.1 | 93.2 |
TF–IDF + CBOW | 92.3 | 93 | 94.2 | 95.8 | |
TF–IDF + CBOW + Skip-gram | 93 | 94.3 | 95.4 | 97.2 | |
40% | TF–IDF | 90.2 | 90.8 | 91.4 | 92 |
TF–IDF + CBOW | 91.3 | 92.2 | 93.6 | 94.8 | |
TF–IDF + CBOW + Skip-gram | 93.5 | 94.6 | 95.8 | 96 |