Complete Labelled News Corpus

23,344 private-market news articles, both relevant and not relevant, each with a headline, source, relevance label and category, and linked to the companies and investors it mentions. Because it holds the articles a classifier should ignore as well as those it should keep, it can be used for training and for evaluation. A live feed is available. Suited to news classification, entity linking, event extraction and retrieval testing. Masked sample available. Licence, permitted use and delivery are confirmed per deal; data about individuals is released only after compliance review.

At a glance

Related datasets