Add dataset card (README)

README.md

@@ -1,3 +1,37 @@
 ---
 license: mit
 ---
+
+# PoC (Patents with One Citation) dataset
+
+This dataset is useful for training or evaluating models that predict patent-to-patent similarity, such as those used for patent searching.
+
+It was developed and used for the training of an ML model that powers the [PQAI](https://search.projectpq.ai/) search engine.
+
+
+## Details
+
+The dataset contains 90,013 samples.
+
+Each sample contains:
+- a subject patent (`sp`)
+- its only citation (`cit`)
+- its CPC code (`cpc`)
+- a list of 10 patents (`sims`) that are similar to `sp` (in that they share the CPC code) and published before `sp`
+
+Every line of the dataset is a JSON parsable string (`.jsonl` format), which upon parsing given an array of this format:
+
+```
+[pn, cit, cpc, [...sims]]
+```
+
+## Task
+
+Given the subject patent `sp` the task is to assign a similarity score to each patent `[cit, ...sims]`. Ideally, the score should be maximum for `cit`.
+
+## Metrics
+
+It's a ranking task, so the following metrics make the most sense:
+
+- DCG/NDCG
+- Accuracy