dondon2475848

Dataset

英文

1. CNN/Daily Mail

   • 2015-Hermann et al. - Teaching machines to read and comprehend
   • 2016-Nallapati et al.-Abstractive text summarization using sequence-to-sequence rnns and beyond
   • Nallapati等人有定義評估的步驟，後續如要使用可以follow他們的研究
   • dataset contains 287,113 training examples, 13,368 validation examples and 11,490 testing examples. After limiting the input length to 800 tokens and output length to 100 tokens, the average input and output lengths are respectively 632 and 53 tokens.

2. the New York Times dataset (NYT)

• 2008 - Evan Sandhaus - The new york times annotated corpus.

dondon2475848

"""
程式參考自：
https://gist.github.com/bwhite/3726239
https://gist.github.com/gumption/b54278ec9bab2c0e0472816d1d7663be
差異：新增「 sum (2^rel_i - 1) / log2(i + 1) 」的版本
作者：Jie Dondon
版本：ndcg_dondon_20180201_v2
"""

import numpy as np