Luc Moreau
  • Home
  • About
  • Research
  • Publications
  • Contact
  • Research Highlight
  • All
  • Projects
  • Standard
  • Software
  • Technologies
  • <p>PLEAD, A provenance driven and legally grounded explanation for automated decisions, was an interdisciplinary research collaboration between technologists, legal experts, commercial companies, and public organisations between 2019 and 2022. The project investigated how provenance could help explain the logic that underlies automated decision-making to the benefit of data subjects as well as help data controllers demonstrate compliance with the law. It developed <b>Explainability by Design</b> — a holistic methodology characterised by proactive measures to include explanation capability in the design of decision-making systems. <p><ul><li><a href='https://plead-project.org/'>https://plead-project.org/</a></li> <li><a href='https://arxiv.org/abs/2206.06251'>A Methodology and Software Architecture to Support Explainability-by-Design</a></li> <li><a href='https://www.kcl.ac.uk/news/explainability-by-design'>Explainability by design (highlight)</a></li><li><a href='https://arxiv.org/abs/2206.04438'>A taxonomy of explanations to support Explainability-by-Design</a></li></ul>
  • <p>With multidisciplinary expertise in the areas of artificial intelligence, machine learning, knowledge representation, crowdsourcing, the Internet of Things and human-computer interaction, the ORCHID team has driven the science of <b>human-agent collectives</b> (HACs), and it has shown how HACs can be engineered and applied to real-world applications in the critical domains of energy systems, disaster response and citizen science. Specifically, in this project I led the work area on the accountable information infrastructure that allows the veracity and accuracy of seamlessly-blended human and agent decisions, sensor data and crowd-generated content to be confirmed and audited. A key outcome is a machine learning technique that uses provenance to predict the quality of artifacts.<p> <p><ul><li><a href='http://www.orchid.ac.uk/wp-content/uploads/2016/04/ORCHID-Final-Report-FINAL.pdf'>Orchid Final Report</a></li> <li><a href='https://eprints.soton.ac.uk/364593/1/CACM%2520HAC%2520ARTICLE%2520%2520final.pdf'>Communications of the ACM article</a></li> <li><a href='http://dx.doi.org/10.1613/jair.5098'>JAIR paper on the Orchid Final Demonstrator</a></li><li><a href='https://www.aaai.org/ocs/index.php/HCOMP/HCOMP13/paper/viewFile/7388/7406'>Provenance-Based Machine Learning</a></li></ul>
  • <p>Society is progressively moving towards a socio-technical ecosystem in which the physical and virtual dimensions of life are more and more intertwined and where people interaction, more often than not, takes place with or is mediated by machines. The project's goal was to move towards hybrid systems where people and machines tightly work together to build a smarter society. A project focus has been on <em>Rideshare</em>, a car pooling application, in which the governance aspects were studied. Using provenance, our work has been on providing an account of what the system does, and in particular, how it manages participant's reputation. The experience with this application and others, led to the design of a <b>charter for smart platforms</b>. From an engineering perspective, we use a template mechanism to generate provenance. </p><p><ul><li><a href='https://eprints.soton.ac.uk/410307/1/SmartSocietySocialCharterforSmartPlatforms_final.pdf'>Charter for smart platforms</a></li><li><a href='http://dx.doi.org/10.1007/978-3-319-08681-1_8'>An auditable reputation service for collective adaptive systems</a></li><li><a href='https://eprints.soton.ac.uk/405025/'>Provenance template system</a></li></ul>
  • A social machine is a socio-technical system, therefore comprising both humans and technology, interacting and producing outputs or actions which would not be possible without both parties present. The aim of SOCIAM is to build social machines that solve the routine tasks of daily life as well as the emergencies. Its aim is to develop the theory and practice so that we can create the next generation of decentralised, data intensive, social machines. Understanding the attributes of the current generation of successful social machines will help us build the next.
  • <p>Paul and I co-chaired the W3C Provenance Working group resulting in the recommendation PROV. Following this, we wrote a book on PROV. The book is short and sweet, some 110 pages.</p> <ul><li><a href='http://www.provbook.org'>An Introduction to PROV Web Site</a></li></ul>
  • <p>While the benefits of provenance are becoming well understood, an important aspect is the software engineering methodology to create provenance-aware applications. The key objectives of this activity is to conceive principled methodologies to design software that creates quality provenance, and techniques to mimize the programming effort imposed on the programmer. We designed the first provenance-oriented methodology <b>PRIME</b>, <em>Provenance Incorporation Methodology</em>, which we have used in practice in several applications. We have also designed <b>prov-template</b> a templating mechanism, allowing the <em>schema</em> of provenance to be specified declaratively, while an expansion algorithm takes care of instantiating the schema with concrete values.</p><ul><li><a href='https://eprints.soton.ac.uk/267450/'>PRIME Methodology</a></li><li><a href='https://eprints.soton.ac.uk/405025/'>PROV-Template paper</a></li><li><a href='https://lucmoreau.wordpress.com/2017/03/30/prov-template-a-quick-start/'>PROV-Template Quick Start</a></li></ul>
  • <p>Provenance can help us gain insights in the behaviour of applications and the data they produced. We have successfully developed <b>analytics techiques</b> that can help extract knowledge from provenance. <b>Provenance Networks Analytics</b> is a machine learning technique applied to provenance, which has successfully demonstrated, for example, that quality of data can be derived from its provenance. <b>Provenance Summarization</b> has been shown to successfully extra common patterns and outliers in provenance.</p><ul><li><a href='https://www.aaai.org/ocs/index.php/HCOMP/HCOMP13/paper/viewFile/7388/7406'>Provenance Networks Analytics<a/</li><li><a href='http://dx.doi.org/10.4204/EPTCS.181.9'>Provenance Summarization</a></li></ul>
  • <p>We have developed a series of <b>software packages</b> and <b>services</b> to process provenance, many of them are freely available. ProvToolbox and ProvPi are two open source libraries to process provenance in Java and Python, respectively. The others are services built on these libraries.</p><ul><li><a href='http://lucmoreau.github.io/ProvToolbox/'>ProvToolbox</a></li><li><a href='https://pypi.python.org/pypi/prov'>ProvPi</a></li><li><a href='https://provenance.ecs.soton.ac.uk/store/'>ProvStore</a></li><li><a href='https://provenance.ecs.soton.ac.uk/validator/view/translator.html'>Prov Translator</a></li><li><a href='https://provenance.ecs.soton.ac.uk/validator/view/validator.html'>Prov Validator</a></li></ul>
  • <p><b>Provenance</b> is a record that describes the people, institutions, entities, and activities involved in producing, influencing, or delivering a piece of data or a thing. <b>PROV</b> is a World Wide Web Consortium's <b>standard</b> for sharing provenance on the Web. I co-chaired this standardization working group during 2011-2013, which resulted in PROV for which we can now find flagship deployments. A recent addition to the family of specification is PROV-JSONLD.</p><ul><li><a href='https://www.w3.org/TR/prov-primer/'>The PROV Primer</a></li><li><a href='https://www.w3.org/TR/prov-dm/'>The PROV Data Model (PROV-DM) specification</a></li><li><a href='http://www.provbook.org/'>The PROV book: An introduction to PROV</a></li><li><a href='https://openprovenance.org/prov-jsonld/'>The PROV-JSONLD Serialization</a></li></ul>