Evaluating Word Representation Features in Biomedical Named Entity Recognition Tasks

<table class="table-group" id="tab4"><tr><td><table class="table"><tr><td class="thead-hr" colspan="1"><hr/></td></tr><tr><td align="left">the: 0.067476 −0.017934 0.036855 0.348073 0.063362 −0.138005 −0.144527 −0.014324 0.161269 0.152643 …</td></tr><tr><td align="left">of: 0.067905 −0.074922 0.012121 0.050542 0.327945 0.098191 −0.087244 0.194758 0.218592 −0.115941 …</td></tr><tr><td align="left">gene: −0.254542 0.100417 −0.124032 0.084818 −0.279409 0.081752 −0.378949 −0.068434 −0.050847 0.142284 …</td></tr><tr><td align="left">transcript: −0.157966 −0.303626 0.010010 −0.081133 −0.111763 −0.088829 −0.160671 0.185505 0.097515 −0.014036 …</td></tr><tr class="table-tr"><td colspan="1"><hr class="tbody-hr"/></td></tr></table></td></tr></table>

Word embeddings of 4 words in the JNLPBA corpus. Each number denotes the feature value in a latent semantic/syntactic space.

BioMed Research International

tab4

Table 4

Table 4: Evaluating Word Representation Features in Biomedical Named Entity Recognition Tasks 