Dear Valdis,
You understand the lempos attribute correctly. It is a combination of lemma and abbreviation of part of speech. We do not have any in-built tool for computing this attribute.
You can easily create it with a simple python script where you map each part of speech a to single-letter abreviation (for undefined/unspecified parts of speech you can provide the default value "x", e.g. usually used for interjection or particles, but depending on you). Then you use a for loop for reprocessing your vertical file and replacing the lemma column with the lempos values created by your script. The lemma attribute is finally created as a DYNAMIC attribute from the lempos attribute (see more at
Otherwise, the warning on missing lempos can be ignored. It is only recommendation for better use of a corpus. The presence of the lempos attribute within the corpus allows corpus creator to add a part-of-speech selector available on the Advnaced tab of the word list, concordance or n-grams feature. This can implemented by adding the LPOSLIST line into your corpus configuration file, please check our documentation
https://www.sketchengine.eu/documentation/fine-tune-your-corpus/#toggle-id-10
Also you can then compile word sketches on the lempos attribute to allow users select a particular part of speech when generate word sketch results in the interface.