Add dataset derivation info to dataset card.

README.md

@@ -14,8 +14,11 @@ license: unknown
 
 ### Dataset Summary
 
-The CLIMATE-FEVER dataset modified to a format that SentenceBERT training scripts can use as an NLI-style dataset.
-(details to follow)
+The CLIMATE-FEVER dataset modified to supply NLI-style (cf-nli) features or STSb-style (cf-stsb) features that SentenceBERT training scripts can use as drop-in replacements for AllNLI and/or STSb datasets.
+
+There are two cf-nli datasets: one derived from only SUPPORTS and REFUTES evidence (cf-nli), and one that also derived data from NOT_ENOUGH_INFO evidence based on the annotator votes (cf-nli-nei).
+
+The feature style is specified as a named configuration when loading the dataset: cf-nli, cf-nli-nei, or cf-stsb. See usage notes below for `load_dataset` examples.
 
 ### Supported Tasks and Leaderboards
 
@@ -43,23 +46,52 @@ The CLIMATE-FEVER dataset modified to a format that SentenceBERT training script
 
 ### Curation Rationale
 
-[More Information Needed]
+SentenceBERT models are designed for 'Domain Adaptation' and/or 'Fine-tuning' using labeled data in the downstream task domain. As a bi-encoder, the primary objective function is real-valued similarity scoring. Typical training datasets use NLI-style features as input, and STSb-style features as model evaluation during training, and to measure post-hoc, _intrinsic_ STSb performance. Classification tasks typically use a classifier network that accepts SentenceBERT encodings as input, and are trained on class-labeled datasets.
+
+So, to fine-tune a SentenceBERT model in a climate-change domain, a labeled climate change dataset would be ideal. Much like the authors of the CLIMATE-FEVER dataset, we knew of no other _labeled_ datasets specific to climate change. And while CLIMATE-FEVER is suitably labeled for classification tasks, it is not ready for similarity tuning in the style of SentenceBERT.
+
+This modified CLIMATE-FEVER dataset attempts to fill that gap by deriving NLI-style features typically used in pre-training and fine-tuning a SentenceBERT model - the 'train' split. SentenceBERT also uses STSb-style features to evaluate model performance both during training - the 'dev' or 'validation' split - and post-hoc to gauge _intrinsic_ model performance on STSb - the 'test' split.
 
 ### Source Data
 
 #### Initial Data Collection and Normalization
 
-[More Information Needed]
+see CLIMATE-FEVER
 
 #### Who are the source language producers?
 
-[More Information Needed]
+see CLIMATE-FEVER
 
 ### Annotations
 
 #### Annotation process
 
-[More Information Needed]
+##### NLI Derivation
+
+**cf-nli**
+For each Claim that has both SUPPORTS evidence and REFUTES evidence, creates the following labeled pairs in the style of NLI dataset:
+
+_NLI Fields_
+| split | dataset | sentence1 | sentence2 | label |
+|---|---|---|---|---|
+| {'train', 'test'} | 'climate-fever' | claim | evidence | evidence_label SUPPORTS -> 'entailment', REFUTES -> 'contradiction' |
+
+> Note that by defintion, only claims classified as DISPUTED include both SUPPORTS and REFUTES evidence, so this dataset is limited to a small subset of CLIMATE-FEVER
+
+**cf-nli-nei**
+This dataset uses the annotator 'votes' list to cast a NOT_ENOUGH_INFO (NEI) evidence to a SUPPORTS or REFUTES evidence. By doing so, Claims in the SUPPORTS, REFUTES, and NEI classes can be used to generate additional sentence pairs.
+
+_Casting NEI Evidence to SUPPORTS or REFUTES_
+| votes | effective evidence_label |
+|---|---|
+| SUPPORTS > REFUTES | _SUPPORTS_ |
+| SUPPORTS < REFUTES | _REFUTES_ |
+
+Any Claims that have *_at least one_* SUPPORTS or REFUTES evidence, and NEI evidences that can be cast to effective _SUPPORTS_ or _REFUTES_ are then included in the datasset.
+
+##### STSb Derivation
+
+TBD
 
 #### Who are the annotators?