Update README.md

README.md

@@ -32,80 +32,90 @@ This dataset contains the 20 trending repositories of each type: models, dataset
 
 ### Languages
 
-[More Information Needed]
+Not relevant.
 
 ## Dataset Structure
 
 ### Data Instances
 
-[More Information Needed]
+The dataset contains three configurations:
+
+**models**: the history of trending models on Hugging Face
+**datasets**: the history of trending datasets on Hugging Face
+**spaces**: the history of trending spaces on Hugging Face
 
 ### Data Fields
 
-[More Information Needed]
+- date (string): the date of the lookup to the trending repositories
+- author (string): id of the repository owner. It can be null.
+- id (string): id of the repository
+- rank (int64): rank in the trending repositories of its kind (model, dataset, or space). Starts at 1.
+- recent_likes (int64): number of likes received lately (last week)
+- likes (int64): total number of likes
+- month_downloads (int64): number of downloads in the last month
 
 ### Data Splits
 
-[More Information Needed]
+Each configuration only has one split: `train` that contains all the rows.
 
 ## Dataset Creation
 
 ### Curation Rationale
 
-[More Information Needed]
+The dataset is updated daily through a cron job that calls the `https://huggingface.co/api/trending?type=${repoType}&limit=20` endpoint for each repository type (model, dataset, space). The script runs in an [Observable](https://observablehq.com/@huggingface) notebook, and the files are uploaded using the [huggingface.js](https://github.com/huggingface/huggingface.js) library. 
 
 ### Source Data
 
 #### Initial Data Collection and Normalization
 
-[More Information Needed]
+Not relevant.
 
 #### Who are the source language producers?
 
-[More Information Needed]
+Not relevant.
 
 ### Annotations
 
 #### Annotation process
 
-[More Information Needed]
+Not relevant.
 
 #### Who are the annotators?
 
-[More Information Needed]
+Not relevant.
 
 ### Personal and Sensitive Information
 
-[More Information Needed]
+Only public repositories are included in the trending repositories.
 
 ## Considerations for Using the Data
 
 ### Social Impact of Dataset
 
-[More Information Needed]
+Not relevant.
 
 ### Discussion of Biases
 
-[More Information Needed]
+The trending repositories reflect the likes given by Hugging Face users in the last week. Any bias that applies to the users can be reflected in this dataset. As a vanity metric, some users might also be tempted to generate fake likes.
 
 ### Other Known Limitations
 
-[More Information Needed]
+Not relevant.
 
 ## Additional Information
 
 ### Dataset Curators
 
-[More Information Needed]
+Sylvain Lesage, Hugging Face
 
 ### Licensing Information
 
-[More Information Needed]
+Apache License 2.0
 
 ### Citation Information
 
-[More Information Needed]
+Not relevant.
 
 ### Contributions
 
-[More Information Needed]
+Not relevant.