Skip to main content

Table 4 Coherence scores over different number of topics for \({G}'_P\), \({G}'_S\) and \(G'\) in the cutting model

From: Improving topic modeling through homophily for legal documents

  \({G}'_P\), \(w_n \ge 5\) \({G}'_S\), \(w_n \ge 10\) \(G'\), (\(w_j^P \ge 0.50\), \(w_j^S \ge 0.25\))
LDA \(|T| = 200\) 0.102 0.102 0.102
LDA \(|T| = 100\) 0.113 0.113 0.113
LDA \(|T| = 50\) 0.125 0.125 0.125
LDA \(|T| = 10\) 0.151 0.151 0.151
RTM(\(w \rightarrow \infty\)) \(|T| = 200\) 0.159 0.159 0.159
RTM(\(w \rightarrow \infty\)) \(|T| = 100\) 0.165 0.165 0.165
RTM(\(w \rightarrow \infty\)) \(|T| = 50\) 0.169 0.169 0.169
RTM(\(w \rightarrow \infty\)) \(|T| = 10\) 0.174 0.174 0.174
RTM \(|T| = 200\) 0.167 0.167 0.167
RTM \(|T| = 100\) 0.166 0.174 0.173
RTM \(|T| = 50\) 0.171 0.179 0.167
RTM \(|T| = 10\) 0.159 0.182 0.180