GR-06 - ELECTRONIC DESIGN AUTOMATION - EDA
Machine Learning techniques for text synthesis from newspapers
Thesis in external company
Reference persons EDOARDO PATTI
External reference persons Giovanni Garifo (email@example.com), Christian Camarda (firstname.lastname@example.org)
Thesis type EXPERIMENTAL
Description The scope of the thesis is studying the state of the art of text synthesis techniques and developing a framework for the synthesis of large texts. The framework developed will be based on Machine Learning techniques at the core. In particular, the candidate will study and analyse both supervised and unsupervised algorithms for the synthesis of texts in the news articles domain. The candidate will explore the fine tuning of already existing models via transfer learning, and also the training of newly built models based on already existing frameworks before starting the design and implementation phase of a synthesis framework for the domain of interest.
The candidate will be required to:
1. study the state-of-art in text synthesis techniques
2. design and develop a software framework for text synthesis
3. design, develop and test a microservice that expose a text synthesis API
4. build a final “demo” embedding the results of the text synthesis API into an already existing web application.
The thesis will be supported by the startup Column (https://www.column.news). Column is a startup founded in Turin and hosted at the Innovative Companies Incubator of the Polytechnic University of Turin (I3P – www.i3p.it). The team is made of young researchers, engineers, journalists and marketers: all of them motivated and passionate. Column builds a platform for personalised, high quality news, addressed to young people. The services is a mix of technological and social innovation, the outcome of the union between high competences in the ICT world and journalism, with a strong inclination to the analysis of the social needs in the world of news. Column collaborates with high-profile national publishers and we intend to grow always more, until we become a landmark in Italy in the field of personalised quality news.
Required skills programming skills in Python,
Knowledge on Machine Learning (optional)
Deadline 11/10/2023 PROPONI LA TUA CANDIDATURA