Skip to main content

Table 4 Binary versus term frequency features using SVMLight and SVM-perf on unigrams and bigrams

From: Feature engineering for MEDLINE citation categorization with MeSH

  

Binary

  

TF

 
 

Precision

Recall

F-measure

Precision

Recall

F-measure

SVMLight Unigram

0.678

0.302

0.418*

0.694

0.269

0.387

SVMLight Bigram

0.711

0.284

0.406

0.708

0.273

0.394

SVMLight TIAB unigram

0.678

0.302

0.418*

0.700

0.263

0.383

SVMLight TIAB bigram

0.730

0.294

0.420*

0.715

0.268

0.389

SVM-perf Unigram

0.395

0.654

0.492

0.390

0.686

0.497

SVM-perf Bigram

0.414

0.675

0.513

0.442

0.594

0.507

SVM-perf TIAB unigram

0.398

0.659

0.496*

0.401

0.609

0.483

SVM-perf TIAB bigram

0.408

0.685

0.512

0.428

0.611

0.503

  1. For each row, significantly better results (p >0.05) are indicated with *.
Back to article page

BMC Bioinformatics

ISSN: 1471-2105

Contact us