Add note on reference coding into data README.

1 files changed, 20 insertions, 1 deletions

diff --git a/data/readme.md b/data/readme.md
index 9fcc4779..f1580e7e 100644
--- a/data/readme.md
+++ b/data/readme.md
@@ -12,6 +12,25 @@ find a certificate ID (to analyze why that happened and whether we could fix tha
 ID (to analyze why that happened and whether we could fix that). These files are used by the `cert_id_eval.ipynb`
 Jupyter notebook which evaluates a dataset with respect to the manually labeled ground truth in them.
 
+
+### Reference coding
+
+#### Component used
+The referenced certificate is for a component used in the product (e.g., IC used by a smartcard).
+
+#### Re-certification of
+The referenced certificate is of a previous version of the product that is now being re-certified.
+
+#### Evaluation reused
+The evaluation results of the referenced certificate were used for evaluation of the product, likely due to a shared component.
+
+#### Basis for
+The current certificate is a basis for the referenced certificate, likely during a simultaneous certification.
+
+#### Previous version
+The referenced certificate is for a previous version of the product, but re-certification is not explicitly mentioned.
+
+
 ## CPE Matching
 
 The directory `./cpe_eval` contains digests of 100 randomly sampled certificates, together with predicted and ground-truth labels.
@@ -19,4 +38,4 @@ The file `random.csv` summarizes the data above, while `manual_cpe_labels.json`
 
 These files can be utilized from [cpe_eval notebook](../notebooks/cc/cpe_eval.ipynb) to see the performance of the classifier.
 
-Folder `./old_manual_cpe_labels` contains some old incomplete labeling that was obtained highly unoptimized classifier.
\ No newline at end of file
+Folder `./old_manual_cpe_labels` contains some old incomplete labeling that was obtained highly unoptimized classifier.