Current behaviour on importing a corpus from another exported corpus:

1. create any corpus on any instance from any source (use few docs to test faster)
2. make some changes in Terms (remove or group ngrams, etc.)
3. look at and remember the total number of Terms
4. then export Documents and Terms (JSON or CSV)
5. create a new corpus node on another instance (OR on the same instance, but here we want to be sure we are working with non-existent documents)
6. on the new corpus node, import the exported Documents (so it will generate all children nodes: Docs, Terms and Graph)
7. then, import the previously exported Terms
8. look at the total number of Terms: the total number is greater than the expected total.

We can observe that imported terms does not replace the previous ones.

Here is a sample set of few exported Documents with Terms to test:

• tardigrades_and_co2_-_GarganText_NgramsList-251.json
• tardigrades_and_co2_-_GarganText_DocsList-nodeId-250.json