Skip to main content
. Author manuscript; available in PMC: 2024 Jan 11.
Published in final edited form as: Artif Intell Med. 2023 Feb 28;139:102523. doi: 10.1016/j.artmed.2023.102523

Figure 8. Comparison of scenario-based accuracy of the similarity-measuring algorithms.

Figure 8.

The conventional Resnik method is more aligned with experts’ assessments only in two similar scenarios: where the reference term is rare and is compared with only one hierarchically related common candidate (S8: RCAC¬A, S9: RCAR). However, Resnik fails on four other scenarios where such relationships do not exist (S1: CRDRD, S3: CR¬DR¬D, S4: CCDCD, S5: CC¬DC¬D), two of which with 0% accuracy (S1, S4). Table 4 provides details outlining each scenario.