Fig 1.
Overview of deep learning model for prediction of future growth potential.
Fig 2.
Construction of training and prediction datasets.
Fig 3.
Motif distribution examples detected in various fields of network [23].
Fig 4.
Motifs used to embed citation network.
Fig 5.
Optimal deep learning structure for prediction of future growth potential of technology clusters.
Table 1.
Performance of prediction models.
Fig 6.
Similarity-based mapping of all technology clusters.
Fig 7.
Abstract-based mapping of all technology clusters.
Fig 8.
Technology clusters with high growth potential (purple nodes).
Fig 9.
Five fields’ technology clusters: Ratio of number of field’s clusters to total clusters (red color); Ratio of number of field’s clusters to clusters predicted to have high growth potential (blue color).
Fig 10.
Keyword mapping of technology clusters with high growth potential.
Fig 11.
Comparison of growth rate distribution of technology clusters with and without high growth potential over all periods.
Fig 12.
Comparison of growth rate distribution of technology clusters with and without high growth potential over last 5 years.
Table 2.
Performance of deep learning-based prediction model.