Table 6.

Evaluation results of embedding using different embedding models for TKG (adapted from [30])

Triple type Model Dataset MRR MR Hits@10 Hits@3 Hits@1
ALL TransE ALL 0.354 5040.016 0.486 0.381 0.286
No-CPE 0.052 13 909.149 0.096 0.077 0.019
No-CAPEC 0.355 4943.392 0.491 0.381 0.287

DistMult ALL 0.205 7770.024 0.334 0.222 0.138
No-CPE 0.104 6837.032 0.231 0.097 0.047
No-CAPEC 0.219 7 520 358 0.355 0.230 0.153

CVE-CWE TransE ALL 0.077 11 284.297 0.196 0.075 0.025
No-CPE 0.004 16 158.898 0.009 0.003 0.002
No-CAPEC 0.075 11 495.046 0.190 0.069 0.024

DistMult ALL 0.101 16 724.084 0.225 0.102 0.044
No-CPE 0.078 7910.668 0.185 0.069 0.028
No-CAPEC 0.104 15 791.059 0.228 0.098 0.051

CWE internal TransE ALL 0.305 3411.406 0.530 0.457 0.120
No-CPE 0.315 1614.365 0.596 0.519 0.079
No-CAPEC 0.289 4616.500 0.457 0.393 0.156

DistMult ALL 0.128 2376.068 0.241 0.126 0.068
No-CPE 0.258 1413.712 0.474 0.252 0.165
No-CAPEC 0.137 4238.609 0.214 0.130 0.090

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.