Skip to main content

Table 9 Result by skip-gram model and multi-scale word clustering.

From: CHEMDNER system with mixed conditional random fields and multi-scale word clustering

System

CDI

CEM

 

Precision

Recall

F-score

Precision

Recall

F-score

A

0.87207

0.89328

0.88255

0.89243

0.85085

0.87115

B

0.86934

0.89861

0.88373

0.88985

0.85910

0.87420

C

0.86879

0.90256

0.88535

0.88840

0.86699

0.87756

D

0.86965

0.90351

0.88625

0.89017

0.86671

0.87828

E

0.87322

0.90007

0.88644

0.89263

0.86218

0.87714

F

0.86721

0.90782

0.88705

0.88725

0.87413

0.88064

G

0.86847

0.90387

0.88582

0.88845

0.86427

0.87619

H

0.86298

0.90818

0.88501

0.88373

0.87401

0.87884

  1. System configuration: each configuration includes clustering method, the corpus size (S) and cluster number (N).
  2. A: Brown clustering, S = 74 K, N = 1 K; B: Skip-gram model, S = 74 K, N = 1 K;C: Skip-gram model, S = 100 k, N = 1 K; D: Skip-gram model, S = 400 k, N = 1 K; E: Skip-gram model, S = 400 k, N = 10 k; F: Skip-gram model, Multi-level clustering, S = 400 k, N = 10 k,1 k,50; G: Skip-gram model, S = 2.4 M, N = 10 k; H: Skip-gram model, Multi-scale clustering, S = 2.4 M, N = 10 k,1 k,50.