Datasets:
Update README.md
Browse files
README.md
CHANGED
@@ -13,10 +13,10 @@ LePaRD is a massive collection of U.S. federal judicial citations to precedent i
|
|
13 |
The distribution of passage citation frequency is long tailed, a small number of passages appear thousands of times in the data while many are cited just once of twice. As a result, the passage retrieval task becomes harder as we consider more data.
|
14 |
|
15 |
We provide four versions of LePaRD:
|
16 |
-
* top_10000_data.csv.gz: Contains the data corresponding to the 10,000 most cited passages
|
17 |
-
* top_20000_data.csv.gz: Contains the data corresponding to the 20,000 most cited passages
|
18 |
-
* top_50000_data.csv.gz: Contains the data corresponding to the 50,000 most cited passages
|
19 |
-
* all_data.csv.gz: Contains data associated with all passages.
|
20 |
|
21 |
Each row of LePaRD contains the following features:
|
22 |
* passage_id: A unique identifier for each passage
|
@@ -25,7 +25,7 @@ Each row of LePaRD contains the following features:
|
|
25 |
* court: The court from which the passage originated
|
26 |
* date: The date when the opinion from which the passage originated was published
|
27 |
|
28 |
-
Passage_ids are mapped to the passage text in passage_dict.json.
|
29 |
Note that multiple slightly different quotes can map to the same passage as judges will sometimes cite different parts of the same sentence.
|
30 |
|
31 |
In the vocabulary of information retrieval, the destination_context can be seen as a query, and the predicted passage_id (or the actual text of a passage in passage_dict.json) can be seen as the targets.
|
|
|
13 |
The distribution of passage citation frequency is long tailed, a small number of passages appear thousands of times in the data while many are cited just once of twice. As a result, the passage retrieval task becomes harder as we consider more data.
|
14 |
|
15 |
We provide four versions of LePaRD:
|
16 |
+
* [top_10000_data.csv.gz](https://huggingface.co/datasets/rmahari/LePaRD/blob/main/top_10000_data.csv.gz): Contains the data corresponding to the 10,000 most cited passages
|
17 |
+
* [top_20000_data.csv.gz](https://huggingface.co/datasets/rmahari/LePaRD/blob/main/top_20000_data.csv.gz): Contains the data corresponding to the 20,000 most cited passages
|
18 |
+
* [top_50000_data.csv.gz](https://huggingface.co/datasets/rmahari/LePaRD/blob/main/top_50000_data.csv.gz): Contains the data corresponding to the 50,000 most cited passages
|
19 |
+
* [all_data.csv.gz](https://huggingface.co/datasets/rmahari/LePaRD/blob/main/all_data.csv.gz): Contains data associated with all passages.
|
20 |
|
21 |
Each row of LePaRD contains the following features:
|
22 |
* passage_id: A unique identifier for each passage
|
|
|
25 |
* court: The court from which the passage originated
|
26 |
* date: The date when the opinion from which the passage originated was published
|
27 |
|
28 |
+
Passage_ids are mapped to the passage text in [passage_dict.json](https://huggingface.co/datasets/rmahari/LePaRD/blob/main/passage_dict.json).
|
29 |
Note that multiple slightly different quotes can map to the same passage as judges will sometimes cite different parts of the same sentence.
|
30 |
|
31 |
In the vocabulary of information retrieval, the destination_context can be seen as a query, and the predicted passage_id (or the actual text of a passage in passage_dict.json) can be seen as the targets.
|