Algorithm 1. w-LDA Generation Process

Input:K, U, T, $\mathit{\alpha }, \mathit{\beta }$

Output: ${\mathit{\phi }}_{\mathit{z}}$ , ${\mathit{\vartheta }}_{\mathit{p}}$, uProfileSet

//Step 1: User profiles data generation

For each author u = 1…U

Traverse all the Weibo posts t = 1…T;

Aggregate the posts generated by author u into users’ profile s uProfileSet(p)

Return uProfileSet

//Step 2: Topics generation

For each topic z = 1…K

Draw ${\mathit{\phi }}_{\mathit{z}}$ ~ Dirichlet ($\mathit{\beta }$)// sample mix components for topic-word

//Step 3: Topic words generation

For each user profile p = 1…P

Draw ${\mathit{\vartheta }}_{\mathit{p}}$ ~ Dirichlet ($\mathit{\alpha }$)// sample mix components for user profile-topic

For each word in generated user profile p, i = 1…Np

Draw $z_i$ ~ Multinomial (${\mathit{\vartheta }}_{\mathit{p}}$) // Sample topic $z_i$ for user profile p

Draw $w_i$~ Multinomial (${\mathit{\phi }}_{{\mathit{z}}_{\mathit{i}}}$) //Sample word $w_i$ for topic $z_i$

Return ${\mathit{\phi }}_{\mathit{z}}$ and ${\vartheta }_p$