Algorithm 1 . Identifying related posts.

Input: Set of posts D

Input: Set of attributes Δ

Input: K value

Input: Sentence-BERT model

1foreach discussion $d_i\in$ D do

2    $D\prime \leftarrow D\prime +preprocessing(d_i)$;

3 end

4 $/*similarityChecker*/$

5 foreach discussion $d_i\in D\prime$ do

6   foreach discussion $d_j,(d_j\ne d_i)\in D\prime$ do

7    $maste{r}_i\leftarrow d_i$;

8    $targe{t}_j\leftarrow d_j$;

9    $similarity\mathrm{\_}valu{e}_{i\mathrm{\_}j}\leftarrow similarity(maste{r}_i,targe{t}_j)$ / ∗ Using model ∗ /;

10     $tuple\leftarrow (maste{r}_i,targe{t}_j,similarity\mathrm{\_}valu{e}_{i\mathrm{\_}j})$;

11     $save(tuple,similarity\mathrm{\_}values\mathrm{\_}file)$;

12   end

13 end

14 $/\ast DescriptiveStatisticcalculationandThresholddefinition\ast /$

15 $S\leftarrow \{\}$

16 foreach discussion $d_i\in D\prime$ do

17   $maste{r}_i\leftarrow d_i$;

18   $S=S+topK\mathrm{\_}sim\mathrm{\_}values(maste{r}_i,K)$;

19 end

20 $Q1\leftarrow 25th\mathrm{\_}percentile(S)$;

21 $Q2\leftarrow 50th\mathrm{\_}percentile(S)$;

22 $Q3\leftarrow 75th\mathrm{\_}percentile(S)$;

23 $IQR\leftarrow Q3-Q1$;

24 $T_{related}=Q3+(1.5\ast IQR)$;

25 $/\ast Selectionofrelatedpostcandidates\ast /$

26 $R\leftarrow \{\}$;

27 foreach $maste{r}_i,targe{t}_j,similarity\mathrm{\_}valu{e}_{i\mathrm{\_}j}\in similarity\mathrm{\_}values\mathrm{\_}file$ do

28  if $(similarity\mathrm{\_}valu{e}_{i\mathrm{\_}j}\ge T_{related})$ then $R\leftarrow R+(maste{r}_i,targe{t}_j)$;

29 end

30 return R;