Dataset research report
Newsroom research report
A reproducible data report with schema notes, generated chart evidence, suggested follow-up questions, and export-ready Helix queries.
Executive Summary
NEWSROOM is a large dataset for training and evaluating summarization systems. It contains 1.3 million articles and summaries written by authors and editors in the newsrooms of 38 major publications. Dataset features includes: - text: Input news text. - summary: Summary for the news. And additional features: - title: news title. - url: url of the news. - date: date of the article. - density: extractive density. - coverage: extractive coverage. - compression: compression ratio. - density_bin: low, medium, high. - coverage_bin: extractive, abstractive. - compression_bin: low, medium, high. This dataset can be downloaded upon requests. Unzip all the contents "train.jsonl, dev.josnl, test.jsonl" to the tfds folder.
Follow-Up Queries
Method And Limits
- Load the catalog entry and preview rows from the processed dataset file.
- Infer numeric, categorical, time, and location fields from real columns.
- Generate a small set of defensive Plotly chart specifications from that profile.
- Expose each chart idea as a query link so the report can be rerun or exported in Helix.
This report is intentionally reproducible. It uses the local catalog metadata and generated chart specifications rather than claiming external conclusions beyond the dataset.