Skip to main content

Table 6 Retrieval performance comparison between expansion terms and combination terms with different similarity thresholds

From: Using NLP in openEHR archetypes retrieval to promote interoperability: a feasibility study in China

Methods AP P@3 P@5
Low level set Original search terms   0.050 0.050 0.050
  Expansion terms Top3 0.963 0.950 0.975
   Top5 0.975 0.950 1.000
   Top10 0.963 0.950 0.975
   Mean 0.967 0.950 0.983
  Combination terms Top3 0.512 0.500 0.525
   Top5 0.575 0.550 0.600
   Top10 0.6125 0.600 0.625
   Mean 0.567 0.550 0.583
Medium level set Original search terms   0.137 0.125 0.15
  Expansion terms Top3 0.888 0.850 0.925
   Top5 0.888 0.850 0.925
   Top10 0.875 0.850 0.900
   Mean 0.883 0.850 0.917
  Combination terms Top3 0.200 0.200 0.200
   Top5 0.200 0.200 0.200
   Top10 0.250 0.250 0.250
   Mean 0.217 0.217 0.217
High level set Original search terms   0.150 0.150 0.150
  Expansion terms Top3 0.637 0.525 0.750
   Top5 0.612 0.550 0.675
   Top10 0.575 0.500 0.650
   Mean 0.608 0.525 0.692
  Combination terms Top3 0.150 0.150 0.150
   Top5 0.175 0.175 0.175
   Top10 0.175 0.175 0.175
   Mean 0.167 0.167 0.167
  1. Different similarity thresholds: in the process of synonym expansion, the first 3, 5, and 10 terms of similar results are used as expansion terms and then composed as combination terms
  2. AP: average precision, P@3: precision at 3, P@5: precision at 5