No AI Without Human Intelligence!

May 10, 2019 by Stefan Summesberger

SEMANTiCS 2019 Workshops & Tutorials Chair Irene Celino is the Head of the Knowledge Technologies group at Cefriel in Milano, Italy. The main areas of her research are Human Computation and Semantic Interoperability where she is principally focused on putting back the “human in the loop” in information processing workflows. In this interview, Irene shares some insights into these areas and gives us a preview of what to expect from this year’s workshops and tutorials.

You are an expert in the fields of Web, Semantic Web and Linked Data and  Human Computation. What is Human Computation? Which area of this domain do you work in and what are the most recent innovations there?

Human Computation is a broad field of research aimed at exploiting people capabilities in information management. It spans from the involvement of the general audience in collecting data from the field (as in citizen science), to outsourcing simple and repetitive tasks like image labeling to a distributed crowd on the Web (as in crowdsourcing), up to engaging people in several different steps of the research process, like providing “human” explanations to automatic machine predictions (as in explainable AI). In all those cases, humans complement and enrich what machines can do: artificial intelligence still needs human intelligence!

My focus is on facilitating the cooperation between humans and machines, by providing people with simple yet effective tools to contribute. I have been researching and adopting games with a purpose and gamification approaches to engage and retain human contributors and, lately, I have been focusing on exploiting storytelling and conversational interfaces to improve data collection.

Are you interested in Human Computation and Semantic Interoperability? Register for SEMANTiCS 2019!

Your recent work is on City Planning, Citizen Science and Games with a Purpose. I guess you worked a lot with public data there. How is the situation on availability, quality and actuality of public data in relation to the past? Do you see further unsettled demands? Which and Where?

In every context, being it public open data or even private business information sources, data quality is the number one issue. Any piece of knowledge requires a clear understanding of its “semantics” to enable reuse, fusion, integration and exploitation. Up to today, there is no silver bullet solution to automatically improve data quality: while in a few cases simple scripts can improve accuracy (e.g., by transforming dates-times into a canonical reference form), in most cases human interpretation is required. That’s exactly the case for Human Computation, which will be more and more employed to put back the “human in the loop”. Just to make a simple example, any machine learning algorithm requires a labelled training set, large enough to “teach” the machine to produce automatic predictions; and who better than a knowledgeable person can create such training set?

What are the most exciting business cases in the areas you are active? Any projects that we should look out for?

I am currently involved in a European cooperative research project named ACTION - Participatory science toolkit against pollution: a collective endeavour to provide effective tools and methods to support citizen science initiatives around Europe to fight pollution. We aim at supporting both the scientists in their open research process and citizens willing to contribute to environmental and awareness campaigns.

My team at Cefriel is busy in providing CONEY, our Conversational Survey toolkit to design and administer questionnaires through an engaging chat interface. We mix user experience with FAIR data and services to achieve a more impactful science and a truly responsible and reproducible innovation.

At SEMANTiCS 2019 you are chairing the Workshops and Tutorials Track. The call for Workshops and Tutorials is closed by now - can you give us a preview of what to expect? What are the topics that we will be able to get hands-on experiences at SEMANTiCS 2019? Workshops? Tutorials?

As in the best tradition of SEMANTiCS, also this year we will have a rich and varied program of workshops and tutorials, spanning different expertise level and diverse application domains: knowledge engineering, natural language processing, data integration, transportation, agriculture, patents, licenses, industry 4.0 and robotics… I’m sure that SEMANTiCS attendees will have a hard time to decide which track to follow!

Discuss about Human Computation and Games with a Purposewith Irene. Register for SEMANTiCS 2019!

About SEMANTiCS

The annual SEMANTiCS conference is the meeting place for professionals who make semantic computing work, and understand its benefits and know its limitations. Every year, SEMANTiCS attracts information managers, IT-architects, software engineers, and researchers, from organisations ranging from NPOs, universities, public administrations to the largest companies in the world. http://www.semantics.cc