Geranium - Nexa Center for Internet & Society

Website: https://geranium.polito.it/
Status: frozen
Period: January 2019 – September 2021
Funding: in-kind
Funding organization: N/A
Person(s) in charge: Giovanni Garifo (Researcher), Giuseppe Futia (Researcher), Antonio Vetrò (Senior Researcher).

Executive summary

Geranium is a research project carried out by the Nexa Center for Internet & Society with the Research Area of Politecnico di Torino (PoliTO), aiming at developing a platform to: (i) reduce the barriers for accessing scientific publications by citizens; (ii) create new collaboration opportunities between researchers. The first achievements of the project are a (i) novel academic Knowledge Graph (KG), (ii) a search engine to explore the publications of PoliTO researchers and (iii) a novel recommendation system based on Graph Neural Networks (GNNs). An alpha version of the Geranium platform is available at: https://geranium.polito.it/.

For comprehensive details about the platform, refer to the publication by Garifo et al.: “The Geranium Platform: A KG-Based System for Academic Publications”. The research project is currently suspended due to the lack of fundings.

Background

PoliTO publications are currently available through the Institutional Research Information System (IRIS – https://iris.polito.it/) platform developed by the Cineca, a not-for-profit Consortium made up of 67 Italian universities, 9 Italian Research Institutions, 1 Polyclinic and the Italian Ministry of Education. The IRIS platform allows to explore publications of research institutions such as PoliTO, combining a search mask based on keywords with different filters. Nevertheless, the platform lacks a semantic infrastructure to contextualize and connect information related to different publications and authors. Moreover, the IRIS platform does not provide any mechanism to facilitate collaborations between the authors and departments, which share common research interests.

Objectives

The goal of this research project is to design and develop a semantic platform for the PoliTO publications available on IRIS. The current work is focused on the improvement of the Web application to explore contents related to the papers and the researchers, introducing advanced filtering mechanisms. Moreover, we intend to increase the accuracy of the implemented recommendation system, whose goal is to suggest valuable connections between researchers.

Results

We developed the alpha version of the Geranium web platform built on top of a novel academic KG, which semantically connects the information related to the publications available on IRIS. Geranium allows to explore details on authors and publications, through automatically extracted topics from the article abstracts. The topic extraction is performed by TellMeFirst, a tool developed by the Nexa Center, that employs Wikipedia and semantic technologies for text classification tasks. Moreover, Geranium currently provides a recommendation mechanism, implemented using GNNs, which suggest authors with common research interests. The first version of Geranium has been developed by: (i) Giovanni Garifo (Lead Software Developer & ML Engineer), (ii) Giuseppe Futia (Project Lead), Mehdi Khrichfa (Backend Developer), Cosma Alex Vergari (Web Developer).