I wrote a python script which reads the Lojban thesaurus and, using a given lujvo as a template, writes all lujvo that have a seltau belonging to the same category the the seltau of the lujvo belongs to, and a tertau belonging to the same category that the tertau of the lujvo belongs to. For example, for the lujvo "tamsmi", the output is all the lujvo with a seltau in the category "3.5 Forms" (where tarmi is found) and a tertau in the category "1.5 Comparison" (where simsa is found). The lujvo can then be defined in English and Lojban. Most of the lujvo are useless, and many are more or less synonyms, but it still gives some interesting new ideas. This was done as an experiment of exploring Lojban lujvo in a manner independent of English.
Here is the current list of lujvo that I have defined which are generated from the script. I also added a few lujvo which occured to me while defining the others. If you want to try playing around with the python script itself see here.