The dataset WIKIBIO is available from https://github.com/DavidGrangier/wikipedia-biography-dataset
We preprocess the data following the work of Liu et al. 2018. You can download the original_data from https://github.com/tyliupku/wiki2bio
Firstly, we extract words, keys and positions from the original infoboxes. python prepro.py Then we order the input attributes according to the order in which they appear in the summaries. python prepro/preprocess_order.py
The code will be released soon.
The test results of our model from the test set of WIKIBIO are stored in the test directory test/test.summary_hplan.