In this competition, you must identify the target of a pronoun within a text passage. The source text is taken from Wikipedia articles. You are provided with the pronoun and two candidate names to which the pronoun could refer. You must create an algorithm capable of deciding whether the pronoun refers to name `A`, name `B`, or neither.

## Files

Unlike many Kaggle challenges, this competition does not provide an explicit labeled training set. Files are also available on the [GAP Dataset Github Repo](https://github.com/google-research-datasets/gap-coreference). Note that the labels for the test set are available on this page. However, your final score and ranking will be determined in stage 2, against a withheld private test set.

 - **test_stage_1.tsv** - the test set data for stage 1
 - **sample_submission_stage_1.csv** - a file showing the correct submission format for stage 1

## Columns

 - `ID` - Unique identifier for an example (Matches to Id in output file format)
 - `Text` - Text containing the ambiguous pronoun and two candidate names (about a paragraph in length)
 - `Pronoun` - The target pronoun (text)
 - `Pronoun-offset` The character offset of Pronoun in Text 
 - `A` - The first name candidate (text)
 - `A-offset` -  The character offset of name A in Text
 - `B` -  The second name candidate
 - `B-offset` -  The character offset of name B in Text
 - `URL` -  The URL of the source Wikipedia page for the example