LabeledBinaryExtraction
objects from a plain text source. The format of the source should be (for each extraction): - Line 1: Source sentence tokens - Line 2: Source POS tags - Line 3: Source NP Chunk tags - Line 4: Argument1 tokens - Line 5: Argument1 range start and length - Line 6: Relation tokens - Line 7: Relation range start and length - Line 8: Argument2 tokens - Line 9: Argument 2 range start and length - Line 10: label (either 0 or 1) For example: Bush was US President . NNP VBD NNP NNP . B-NP O B-NP I-NP O Bush 0 1 was 1 1 US President 2 2 1 This gets loaded into a LabeledBinaryExtraction
object representing (Bush, was, US President) with a positive label (=0).
@author afader
|
|
|
|