add dataset files

README.md

@@ -1,3 +1,41 @@
----
-license: cc-by-4.0
----
+# SciClaimEval Shared Task: Dataset Information
+
+## Subtask 1: Claim Label Prediction Task
+Each sample includes the following information:
+
+- paper_id: the ID of the paper; it can be an arXiv ID or a PeerJ ID
+- claim_id: the ID of the claim
+- claim: the claim for which the label needs to be predicted
+- label: there are two labels in our dataset: Supported and Refuted
+- caption: the caption of the evidence file
+- context: the preceding sentences from the same paragraph, provided as a short contextual field for each claim sentence
+- domain: three domains, ML, NLP, and PeerJ (medical domain)
+- use_context: No (the claim is understandable without context), Yes (short context is needed; information is taken from the context field), or Other sources (the full paper is needed to understand the claim)
+- operation: how the evidence is modified to obtain the modified evidence that pairs with the same claim to create a refuted sample
+- paper_path: the path to the paper
+- detail_others: if the operation is Other, a description is provided here
+- claim_id_pair: one claim is paired with two pieces of evidence, creating two labels: Supported and Refuted
+
+Please refer to the file [here](https://github.com/SciClaimEval/sciclaimeval-shared-task/blob/main/examples/task1_ground_truth.json) for an example.
+
+
+Please prepare your prediction file following the format in [this file](https://github.com/SciClaimEval/sciclaimeval-shared-task/blob/main/examples/task1_pred_format.json).
+
+
+### Information about the Test Set:
+
+- You will receive the input for the test set, but the gold labels are not available.
+The example format is the same as in the file [task1_ground_truth.json](https://github.com/SciClaimEval/sciclaimeval-shared-task/blob/main/examples/task1_ground_truth.json), except that the following keys are missing: label, operation, detail_others, and claim_id_pair.
+
+- This is also the case for the second subtask.
+
+
+## Subtask 2: Claim Evidence Prediction Task
+
+- Please refer to the file [here](https://github.com/SciClaimEval/sciclaimeval-shared-task/blob/main/examples/task2_ground_truth.json) for an example.
+
+
+- Please prepare your prediction file following the format in [this file](https://github.com/SciClaimEval/sciclaimeval-shared-task/blob/main/examples/task2_pred_format.json).
+
+
+

data/dev_task1_release.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d18c91bfc2336858e9f5ce070e4607b1690aff1308450454d80179486fb5ed49
+size 970566

data/dev_task2_release.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f822b4419cb92e8a1fba29f2e09fa504c5e21d6b5babd5497f41acab6de3ea08
+size 511009