Skip to main content
. Author manuscript; available in PMC: 2017 Jul 26.
Published in final edited form as: Nat Lang Process Inf Syst. 2013 Jun;7934:176–188. doi: 10.1007/978-3-642-38824-8_15

Procedure.

Apply-Context (H, β, 𝒞, ℳ[ ][ ])

1: for all candidate terms t in H do
2:   Set co-occurrence score F = 0
3:   for each context term c in 𝒞 do
4:     F = F + ℳ[c][t]
5:   if F/|𝒞| < β then
6:     H.delete(t) {F/|𝒞| is the average co-occurrence}
7: return H