the provided corpus is a human tagged collection of pairs of sentences marked as being in a paraphrase relationship.It is in json format which adheres to the following schema:

[
    {
         "first"  : "string",
         "second" : "string",
         "tag"    : "int",
         "date"   : "string"
    }
]

Where (first, second) is the sentence pair, tag is a binary value, representing if the pair is in paraphrase relationship, and date is the time the tagging was created in yyyy_mm_dd format