corpus.txt: This file contains the content of parsed web pages. Each line represents a single document, with entity mentions being replaced by their corresponding freebase_ids.

entities.txt: This file contains specific information about the entities that appear in the corpus. Each line consists of five entries, separated by tabs, which includes the freebase_id, anchor_text, wiki_url, freebase_name, and description of an entity.

splits/: This directory contains train/valid/test split of corpus.txt used for the experiments.