In this repository one can find all related code to the Master thesis 'Finding Frequently Asked Questions' by David Zomerdijk. This research was carried out during a Master degree in Artificial Intelligence at the University of Amsterdam.
The thesis was done in cooperation with ABN AMRO bank, all code pertaining ABN AMRO data is not publicly available.
The directory contains the following subdirectories:
- analysis: contains code to explore the topic models latent space
- data: contains some of the public datasets
- lm: contains code for the language models
- lm_cond: contains language models that can be conditioned
- readers: contains readers classes, these can convert reader data into a useful object for the models
- saved: a directory where models are saved
- tm: contains code for several topic models
- lstm_encoder: contains code that uses the bi-lstm encoder for Prodlda