Figures & data
Figure 2. The workflow of the proposed approach. The methodology is composed of three steps: CNN design and training, word sensitivity test and duplication test.
![Figure 2. The workflow of the proposed approach. The methodology is composed of three steps: CNN design and training, word sensitivity test and duplication test.](/cms/asset/5d236204-c480-4498-9f05-3ce67cc9a316/tjde_a_1523956_f0002_oc.jpg)
Table 1. Parameter settings.
Figure 4. Training accuracy and loss of designed CNN architecture on flooding and non-flooding dataset supported by GTX 1050 GPU and CUDA.
![Figure 4. Training accuracy and loss of designed CNN architecture on flooding and non-flooding dataset supported by GTX 1050 GPU and CUDA.](/cms/asset/3e72e109-3e1f-4e6b-aa8d-d2c0851da4e4/tjde_a_1523956_f0004_oc.jpg)
Figure 5. CNN performance with different : (a) number of pictures classified as positive (flooding), verified as positive (TP) and verified as negative (FP) given different
and (b) trend of precision and recall given different
.
![Figure 5. CNN performance with different Tf: (a) number of pictures classified as positive (flooding), verified as positive (TP) and verified as negative (FP) given different Tf and (b) trend of precision and recall given different Tf.](/cms/asset/d79d04e3-9b76-437f-8ba1-7bbe6db0b918/tjde_a_1523956_f0005_oc.jpg)
Table 2. CNN labeling result.
Figure 6. Temporal distribution of selected flood-sensitive words (). The X-axis and Y-axis in each subfigure denote time (from 15th August to 15th September in 2017) and number of tweets (count) respectively. Number of tweets for each word has been scaled for better visualization and comparison.
![Figure 6. Temporal distribution of selected flood-sensitive words (Sf>15). The X-axis and Y-axis in each subfigure denote time (from 15th August to 15th September in 2017) and number of tweets (count) respectively. Number of tweets for each word has been scaled for better visualization and comparison.](/cms/asset/0dd34ed6-c010-4360-8589-3ed9669c310d/tjde_a_1523956_f0006_oc.jpg)