Tutorial Objectives and Description

TOPIC

This tutorial introduces the topic of semantic data enrichment, covering theoretical and practical considerations. In particular, the tutorial will provide an explanation of the role that semantics play in data enrichment for downstream AI-based applications, a review of the advantages and limitations of tools, methodologies, and techniques for semantic data enrichment available today, and a practical dive into the creation of data transformations for enriching the data.

OBJECTIVES

Provide evidence for the hypothesis that several semantic approaches can play an important role in supporting high-quality, controllable, scalable tabular data enrichment processes. This objective is related to proposing data enrichment as a key perspective on semantic solutions to support AI-based applications and intercept a strong demand from the market;

Present a link & extend paradigm for data enrichment as a unifying abstraction to develop semantic data enrichment solutions.

Present key aspects of semantic data enrichment, with a particular focus on:

The role of semantic annotation, especially STI and NLP solutions, by reviewing existing solutions and their limitations;
The different phases in the design and deployment of data enrichment pipelines, from discovery to scalable deployment, and the role of algorithms, interactive tools, and web-based infrastructure in both phases;
The role of data pipelines in meeting scalability, repeatability, and engineering constraints;
Reasons for which semantic data enrichment is still relevant in the era of generative AI.

Provide a practical guide to creating a semantic data enrichment pipeline for real-world data enrichment tasks, by using an extensible toolkit (under development in the context of enRichMyData project) that covers key aspects of semantic enrichment.

Discuss open research questions to stimulate further research on data enrichment.

PROGRAM

The tutorial will be a half-day tutorial, requiring approximately three hours including presentations and a hands-on section. The draft schedule proposed for the tutorial is the following:

DATA ENRICHMENT FOR DATA SCIENCE 45 MIN

Topics: challenges; semantics technologies as key enablers; analytics with enriched data: use cases from industry-driven data science projects.

STATE-OF-THE-ART: SEMANTIC DATA ENRICHMENT FOR DATA SCIENCE 45 MIN

Topics: semantic table annotation tools and techniques; doing it with OpenRefine: lessons learned and limitations; data enrichment with batch pipelines: lessons learned and limitations; user-driven data enrichment at scale: the TAO approach;.

A TOOLKIT FOR USER-DRIVEN DATA ENRICHMENT AT SCALE ENDOWED WITH ROBUST ANALYTICAL MODELING 30 MIN

Topics: SemTUI fully modular framework: extending, annotating and enriching data; Expert AI tools: category classification on top of enriched data.

HANDS-ON SESSION: PROCUREMENT DATA ENRICHMENT WITH WIKIDATA IDS 60 MIN

Environment set up; design of the data transformation workflows with SemTUI; batch execution of the transformations on a larger data set; design of larger data set transformation workflows with TAO.

INTENDED AUDIENCE

You should attend the tutorial if you:

Researchers with expertise in semantic technologies, and, in particular in their application to data integration problems (e.g., ontology matching, semantic reconciliation, table annotation), who will discover new industry-driven, real-world application scenarios;

Young researchers that recently joined the semantic web community (e.g., PhD students and postdocs), who will be exposed to challenging problems and possible solution

Data scientists and data engineers, who will learn how semantic technologies can be exploited to support data-driven innovation at scale

SPEAKERS

Matteo Palmonari

Associate Professor

Dumitru Roman

Senior Research Scientist

Flavio De Paoli