Algorithm 1. Algorithm for the proposed AICDR.

Data: Text vectorization matrix $XϵR_{+}^{m\times n}$, the number of topics (K)

1 define $ave\mathrm{\_}dis$ list (L)

2 for each $kϵ\left\{2,\dots ,K\right\}$ do

3    text classification

4    define diameter list (L1)

5       for each $k_iϵ\left\{1,\dots ,k\right\}$ do

6          calculate $D_{k_i}$ by Eq. (3)

7          L1 append $D_{k_i}$

8       end

9   define diameter list (L2)

10  for each $D_{k_p}$ $ϵ\left\{L{1}_1,\dots ,L{1}_{m-1}\right\}$ do

11     for each $D_{k_q}$ $ϵ\left\{L{1}_2,\dots ,L{1}_{m-2}\right\}$ do

12        calculate $D_{p,q}$ by Eq. (3)

13        calculate $D\left(p,q\right)$ by Eq. (5)

14        L2 append $D\left(p,q\right)$

15     end

16  end

17  calculate $ave\mathrm{\_}dis\left(k\right)$ by Eq. (6)

18  L append $ave\mathrm{\_}dis\left(k\right)$

19 end

20 calculate $AICDR\left(k\right)$ by Eq. (7)