Skip to content

Latest commit

 

History

History
66 lines (50 loc) · 3.3 KB

README.md

File metadata and controls

66 lines (50 loc) · 3.3 KB

DOI

alt text

If you end up using this code or the data, please cite our paper:

@unknown{unknown,
author = {Arya, Shreyash and Uberoi, Anannya and Dhawan, Sarthika and Chakraborty, Tanmoy},
year = {2019},
month = {02},
pages = {},
title = {“I am Kalam” - Analyzing and Generating Kalam's Answer Patterns},
doi = {10.13140/RG.2.2.28964.09602}
}

Cite work here!

'I am Kalam' - Reliving Kalam’s Words

💡 The work was presented at the Workshop on AI for Computational Social Systems (ACSS 2019), IIIT-Delhi.

Analyzing answer pattern of APJ Abdul Kalam and responding to a query following his answering pattern. We are applying RNNs to generate answers to user queries.

Dataset: Dataset has been scrapped from interviews available on various websites form the google search results.
Files: dataset/ directory containes different extracted data forms.
Code: code/ directory contains codes from IR-IE model, seq2seq model, preprocessing and evaluation.

IR-IE model


$ python sen2vec_my.py

** sent2vec library needs to be installed from https://github.com/epfml/sent2vec.<br>
** pre trained model <a href='https://drive.google.com/file/d/0B6VhzidiLvjSOWdGM0tOX1lUNEk/view'>torontobooks_unigrams.bin</a> need to be downloaded and kept in same directory.

seq2seq model


$ python main.py 

to train the system and save the model named as model.npz.<br>
Set inference_mode=1 for testing purpose and run python main.py.

References


Tada! (:) ✌️👽