Research Article

A Deep-Learning Framework for Analysing Students’ Review in Higher Education

Table 6

Precision (P), recall (R), and F1-score (F1) for models developed for the aspect-term identification model.

Aspect categoriesLSTMGRUBi-GRUBi-LSTM
PRF1PRF1PRF1PRF1

Assessment0.930.820.870.930.820.870.880.880.880.940.880.91
Content0.670.330.441.000.420.591.000.420.591.000.420.59
Delivery approach0.750.810.780.700.830.760.670.810.740.670.790.72
General0.730.850.790.790.850.810.580.850.690.730.850.79
Lecture hours0.920.73810.831.000.911.000.730.850.920.800.86
Management0.710.500.590.560.500.530.750.600.670.540.700.61
Practical hours0.820.950.880.860.950.900.860.950.900.900.950.92
Tutorials1.000.830.911.000.760.861.000.830.911.000.690.82
Experiential learning1.000.910.950.921.000.961.001.001.000.921.000.96
Professional competencies0.560.560.560.750.330.460.670.220.330.500.220.31
Teaching and learning environment0.531.000.690.570.800.670.530.900.670.621.000.77
Teaching and learning resources0.700.780.740.700.780.740.820.780.800.700.780.74
Accuracy0.780.790.780.77
Macroaverage0.780.760.750.800.750.750.810.750.750.790.760.75
Weighted average0.800.780.780.810.790.780.810.780.780.800.770.77