Wolfram Computation Meets Knowledge

Wolfram Summer School


Valeria Mazzeo

Science and Technology

Class of 2019


Valeria graduated with a Master’s degree in Theoretical Physics from the University of Catania, Italy. Currently she is working at SAS Institute in Dublin (Ireland) and pursuing a PhD in Complex Systems. Her academic research interests include complex networks and multi-agent models applied to socioeconomic systems and data and text mining. Valeria’s interests span from science to literature.

Computational Essay: The Dot-To-Dot Poetry

Project: Text Summarization with GPT-2


Every day, millions of articles and posts unfold on the web, spanning from sports to politics to science. But what would happen if AI systems became better than humans, learning from the huge amount of data and obtaining very high performance across many different tasks, using neural models and deep learning techniques based on natural language processing (NLP)? In this project, we are going to work with the CNN dataset, proposing a new model called GPT-2 (which stands for Generative Pretrained Transformer). This model is trained to predict (or generate) the next token in a sequence of tokens in an unsupervised way, by using a transformer architecture.

Future Work

Future work would be to investigate methods to optimize the neural network architecture in order to achieve more accuracy on different datasets and topics. Another possible direction for future research would be to improve the robustness of the deep learning model by adding noise.