StairwAI

Stairway to AI: Ease the Engagement of Low-Tech users to the AI-on-Demand platform through AI

(description)

Facilitating user engagement in the AI4EU platform

The goal of the AI4EU on-demand platform is to bring together AI stakeholders and resources, overcoming fragmentation issues and encouraging the acceleration of AI-based innovations. Among the most fundamental objectives of the platform is to improve functionality to allow for the easy engagement of its stakeholders, namely SMEs. The EU-funded StairwAI project is targeting low-tech users with the goal of facilitating their engagement in the AI4EU platform. To that end, it will design and implement a series of AI-based services, such as natural language processing, constraint solving, optimisation and machine learning for horizontal and vertical matchmaking as well as reputation and fairness mechanisms to improve the matching results.

(objective)

The most fundamental need of the AI4EU on-demand Platform at this stage in its development is the introduction of improved functionality to allow for the easy engagements of it core targets stakeholders, namely SMEs. While additional resources and strong domain-specific solutions are also needed, it will be a wasted enterprise without a simple mechanism to match users to these assets. What is needed is a Stairway to AI, a linking bridge between users in a low-tech level to the higher-level AI resources that have the potential to transform both their business.

The StairwAI project targets low-tech users with the goal of facilitating their engagement on the AI on-demand Platform. This will be achieved through a new service layer enriching the functionalities of the on-demand platform and containing: (1) a multi-lingual interaction layer enabling conversations with the Platform in the user’s own language, (2) a horizontal matchmaking service for the automatic discovery of AI assets (tools, data sets, AI experts, consultants, papers, courses etc.) meeting the user business needs and, (3) a vertical matchmaking service that will dimension and provision hardware resources through a proper hardware provider (HPC, Cloud and Edge infrastructures).

These services are designed and implemented by using techniques in an AI for AI fashion. The AI techniques deployed in the development of the services are natural language processing for the multi-lingual interaction, constraint solving, optimization and machine learning for horizontal and vertical matchmaking, knowledge representation for organizing the platform AI assets, reputation and fairness mechanisms to improve the matching results.

StairwAI will have a tremendous impact on the sustainability, collaboration opportunities, accessibility and fairness of the AI on-demand Platform, enabling the definition of proper business models for the uptake of AI bringing new value for EU industry.

(fields of science) natural language processing • business models • machine learning

Info

(funding body) EU Horizon 2020 Framework Programme (H2020)
(programme) H2020-EU.2.1.1. – INDUSTRIAL LEADERSHIP — Leadership in enabling and industrial technologies - Information and Communication Technologies (ICT)
(topic) ICT-49-2020 – Artificial Intelligence on demand platform
(funding scheme) IA – Innovation Action

Partners

Alma Mater Studiorum – Università di Bologna (UNIBO) — coordinator
- Michela Milano (Coordinator), Roberta Calegari, Giovanni Ciatto, Michele Lombardi, Andrea Omicini, Federico Sabbatini
University College Cork – National University of Ireland, Cork (UCC)
Universitat Politècnica de Catalunya (UPC)
Huawei Technologies Deutschland GmbH (HUA)
Technische Universiteit Eindhoven (TUE)
Bonseyes Community Association (BCA)
Fundingbox Accelerator sp. z o.o. (FBA)
- (subunit) Fundingbox Communities SL (FCS)
Istituto Nazionale di Fisica Nucleare (INFN)
Tilde SIA (TIL)
Stichting EGI (EGI)
Thales SIX GTS France (THA)

12 Publications / StairwAI

Logic Programming library for Machine Learning: API design and prototype (paper in proceedings, 2022) — Giovanni Ciatto, Matteo Castigliò, Roberta Calegari
Symbolic knowledge extraction from opaque ML predictors in PSyKE: Platform design & experiments (article in journal, 2022) — Federico Sabbatini, Giovanni Ciatto, Roberta Calegari, Andrea Omicini
Semantic Web-based Interoperability for Intelligent Agents with PSyKE (paper in proceedings, 2022) — Federico Sabbatini, Giovanni Ciatto, Andrea Omicini
Evaluation Metrics for Symbolic Knowledge Extracted from Machine Learning Black Boxes: A Discussion Paper (paper in proceedings, 2022) — Federico Sabbatini, Roberta Calegari
Hypercube-Based Methods for Symbolic Knowledge Extraction: Towards a Unified Model (paper in proceedings, 2022) — Federico Sabbatini, Giovanni Ciatto, Roberta Calegari, Andrea Omicini
Symbolic Knowledge Extraction from Opaque Machine Learning Predictors: GridREx & PEDRO (paper in proceedings, 2022) — Federico Sabbatini, Roberta Calegari
Clustering-Based Approaches for Symbolic Knowledge Extraction (paper in proceedings, 2022) — Federico Sabbatini, Roberta Calegari
Shallow2Deep: Restraining Neural Networks Opacity through Neural Architecture Search (paper in proceedings, 2021) — Andrea Agiollo, Giovanni Ciatto, Andrea Omicini
On the Design of PSyKE: A Platform for Symbolic Knowledge Extraction (paper in proceedings, 2021) — Federico Sabbatini, Giovanni Ciatto, Roberta Calegari, Andrea Omicini
GridEx: An Algorithm for Knowledge Extraction from Black-Box Regressors (paper in proceedings, 2021) — Federico Sabbatini, Giovanni Ciatto, Andrea Omicini
A Mechanism for Reasoning over Defeasible Preferences in Arg2P (paper in proceedings, 2021) — Giuseppe Pisano, Roberta Calegari, Andrea Omicini, Giovanni Sartor
Graph Neural Networks as the Copula Mundi between Logic and Machine Learning: A Roadmap (paper in proceedings, 2021) — Andrea Agiollo, Giovanni Ciatto, Andrea Omicini

11 Talks / StairwAI

On the Evaluation of the Symbolic Knowledge Extracted from Black Boxes (talk @ AAAI Spring Symposium 2023 – AITA: AI Trustworthiness Assessment, 27/03/2023) — Roberta Calegari (Federico Sabbatini, Roberta Calegari)
Hypercube-Based Methods for Symbolic Knowledge Extraction: Towards a Unified Model (talk @ WOA 2022, 01/09/2022) — Federico Sabbatini (Federico Sabbatini, Giovanni Ciatto, Roberta Calegari, Andrea Omicini)
Symbolic Knowledge Extraction from Opaque Machine Learning Predictors: GridREx & PEDRO (talk @ KR 2022, 04/08/2022) — Roberta Calegari (Federico Sabbatini, Roberta Calegari)
Logic Programming library for Machine Learning: API design and prototype (talk @ CILC 2022, 01/08/2022) — Giovanni Ciatto (Giovanni Ciatto, Matteo Castiglio, Roberta Calegari)
Clustering-Based Approaches for Symbolic Knowledge Extraction (talk @ KR 2022 – ML Session @ KR 2022, 31/07/2022) — Roberta Calegari (Federico Sabbatini, Roberta Calegari)
Grounding Trustworthy AI: from high-level principles to technical requirements (talk @ IJCAI-ECAI 22, 25/07/2022) — Roberta Calegari (Roberta Calegari, Michela Milano, Catelijne Muller)
Semantic Web-based Interoperability for Intelligent Agents with PSyKE (talk @ EXTRAAMAS 2022 @ AAMAS 2022, 09/05/2022) — Federico Sabbatini (Federico Sabbatini, Giovanni Ciatto, Andrea Omicini)
A Mechanism for Reasoning over Defeasible Preferences in Arg2P (talk @ CILC 2021, 07/09/2021) — Giuseppe Pisano (Giuseppe Pisano, Roberta Calegari, Andrea Omicini, Giovanni Sartor)
On the Design of PSyKE: A Platform for Symbolic Knowledge Extraction (talk @ WOA 2021, 01/09/2021) — Federico Sabbatini (Federico Sabbatini, Giovanni Ciatto, Roberta Calegari, Andrea Omicini)
GridEx: An Algorithm for Knowledge Extraction from Black-Box Regressors (talk @ EXTRAAMAS 2021 @ AAMAS 2021, 03/05/2021) — Federico Sabbatini (Federico Sabbatini, Giovanni Ciatto, Andrea Omicini)
Shallow2Deep: Restraining Neural Networks Opacity through Neural Architecture Search (talk @ EXTRAAMAS 2021 @ AAMAS 2021, 03/05/2021) — Andrea Agiollo (Andrea Agiollo, Giovanni Ciatto, Andrea Omicini)