jbovlaste has mappings between places of Lojban brivla and word-sense pairs in English, and jbofihe has internal data files that encode similar info:
$ echo "le crida cu batci le cribe" | jbofihe -x
[([batci1 (bite-r(s)):] le /the/ crida /fairy(ies)/) cu /is/does/ <<batci /bit-ing/>> ([batci2 (bitten thing(s)):] le /the/ cribe /bear(s)/)]
Back when I was active, I dreamed of making a really excellent machine translator using the jbovlaste data, because you can do it in the Lojban -> English direction without needing tons of ambiguity handling like you do in other languages.