Table 1 Rule set to extract causal relations from tweets
From: Extracting health-related causality from twitter messages using natural language processing
# | Causal relation types | Dependency rules | Examples |
|---|---|---|---|
1 | A (noun) caused B | {} = subj < subj ({+ Causal verb +} = target >dobj {} = cause) | Stress causes insomnia |
2 | A (verb-ing) caused B | {} = subj < csubj ({+ Causal verb +} = target >dobj {} = cause) | Over thinking can increase anxiety and cause insomnia. |
3 | B was caused by A | {} = ncsubjpass<nsubjpass({+ Causal verb +} = target >/nmod:agent/{} = cause) | My insomnia was caused by stress. |
4 | A is a reason of B | Causal noun + < nsubj ({} = target > /nmod:of/ {} = cause) | Stress is a reason of my insomnia |
5 | B was caused by A (verb-ing) | {} = nsubj< nsubjpass ({} = target >/advcl:by/ + Causal noun) | Insomnia was caused by overthinking |
6 | A results “in/to/from” B | Causal verb + < [nc] subj ({} = target> /nmod:(to|in|from)/{} = cause) | Stress results to insomnia. |