diff options
| author | Adam Janovsky | 2021-11-04 08:58:42 +0100 |
|---|---|---|
| committer | Adam Janovsky | 2021-11-04 08:58:42 +0100 |
| commit | 4b1f47a7243c1a0d4118107386440ffef53affb9 (patch) | |
| tree | e60091eba67291f45a84e960cd62007413c38b9b /data/validation_test_split | |
| parent | ff8094b06c9438cdbaeb777f1d5728b19ed773ec (diff) | |
| download | sec-certs-4b1f47a7243c1a0d4118107386440ffef53affb9.tar.gz sec-certs-4b1f47a7243c1a0d4118107386440ffef53affb9.tar.zst sec-certs-4b1f47a7243c1a0d4118107386440ffef53affb9.zip | |
move data readme
Diffstat (limited to 'data/validation_test_split')
| -rw-r--r-- | data/validation_test_split/readme.md | 9 |
1 files changed, 0 insertions, 9 deletions
diff --git a/data/validation_test_split/readme.md b/data/validation_test_split/readme.md deleted file mode 100644 index 623eb9af..00000000 --- a/data/validation_test_split/readme.md +++ /dev/null @@ -1,9 +0,0 @@ -## Dataset split into validation / test set - -To evaluate quality of our automatic matching of CPE names to certificates (which in turn allows us to assign CVEs to certificates), we manually evaluated candidate CPE records for all of the certificates both in FIPS and CC framework. - -Not to cause data leakage, prior to developing automatic method of CPE matching, we split the dataset into validation a test parts in 50:50 fashion. The automatic CPE matching was finetuned using the validation set and only then evaluated using the test set. - -This directory contains digests (primary keys of certificates) of the formed validation and test sets. The output present herein was produced with [validation_test_split.ipnyb](https://github.com/crocs-muni/sec-certs/tree/dev/notebooks/validation_test_split.ipynb). - -To see how the evaluation was performed, take a look at [cc_cpe_labeling.py](https://github.com/crocs-muni/sec-certs/blob/dev/examples/cc_cpe_labeling.py) and [fips_cpe_labeling.py](https://github.com/crocs-muni/sec-certs/blob/dev/examples/fips_cpe_labeling.py).
\ No newline at end of file |