AAAI 2019 Spring Symposium on Combining Machine Learning with Knowledge Engineering
March 25–27, 2019 · Stanford University, Palo Alto, California, USA
AAAI-MAKE 2019 brought together researchers and practitioners in machine learning and knowledge engineering to explore AI systems that are explainable, compliant, and grounded in domain knowledge.
View the 2019 symposium program and papers
About the symposium
Machine learning enables systems to learn from real-world data, including settings where explicit knowledge is incomplete or tacit. Knowledge engineering makes domain expertise, rules, and constraints explicit, supporting reasoning, compliance, and explanations. AAAI-MAKE 2019 explored how these complementary approaches could be combined to create more robust and trustworthy AI systems.
Regulated and high-stakes application areas, including banking, insurance, and healthcare, motivated the discussion. The symposium explored how explicit knowledge might guide data-driven learning, improve performance with sparse or noisy data, and make the resulting decisions more understandable.
Topics
- Knowledge representation and reasoning; ontologies; rule-based systems; Semantic Web
- Machine learning, deep learning, and neural networks
- Knowledge engineering and knowledge management
- Causal explainability and learning cause-and-effect relationships
- Ontology learning and using knowledge to guide machine learning
Format and keynote speakers
The three-day program included keynotes, peer-reviewed full and position papers, posters and demonstrations, discussions, and an AAAI plenary session.
- Doug Lenat (Cycorp): It’s Statistical Learning and Knowledge Engineering All the Way Down — March 25
- Frank van Harmelen (Vrije Universiteit Amsterdam): Keynote — March 26
- Aurona Gerber (University of Pretoria): AAAI plenary report — March 26
Organizing committee
- Peter Clark — Allen Institute for Artificial Intelligence, Seattle, USA
- Aurona Gerber — University of Pretoria, South Africa
- Knut Hinkelmann (co-chair) — FHNW, Switzerland
- Doug Lenat — Cycorp, Austin, USA
- Andreas Martin (co-chair) — FHNW, Switzerland
- Frank van Harmelen — Vrije Universiteit Amsterdam, Netherlands
Symposium administration
Charuta Pande (proceedings), Devid Montecchiari (media and proceedings), and Emanuele Laurenzi (media), FHNW School of Business.
Program committee
Andrea Polini; Andreas Martin; Andreas Schmidt; Ansgar Bernardi; Ashutosh Modi; Aurona Gerber; Barbara Re; Deshen Moodley; Dominik Bork; Ela Pustulka-Hunt; Elena Miron; Emanuele Laurenzi; Erik Mueller; Frank van Harmelen; Hans Chalupsky; Hans Friedrich Witschel; Hans-Georg Fill; Holger Wache; Ivan Jovanovikj; Jonathan Gordon; Ken Barker; Knut Hinkelmann; Kyriakos Kritikos; Ludger Van Elst; Marcela Ruiz; Matt Gardner; Matthias Baldauf; Michael Walch; Nathanael Chambers; Peter Clark; Peter Yeh; Rainer Telesko; Robert Andrei Buchmann; Sheraz Ahmed; Stephan Jüngling; Thomas Hanne; Thomas Meyer; Todor Mihaylov; Ulrich Reimer; Uwe Riss; Wilfrid Utz.
Milestones
- July 19, 2018 — Symposium proposal accepted
- November 18, 2018 — Extended submission deadline
- December 7, 2018 — Author notifications
- February 22, 2019 — Camera-ready deadline
- March 25–27, 2019 — Symposium at Stanford University
Proceedings
Accepted papers were intended for publication in the open-access CEUR Workshop Proceedings, subject to publication permissions and presentation requirements. For the program and paper titles, see Program and Papers—2019.
AAAI-MAKE 2019 formed part of the AAAI Spring Symposium Series of the Association for the Advancement of Artificial Intelligence (AAAI). Contact: mail@aaai-make.info.
Call for Participation
Join us … and work together on a combined machine learning and knowledge engineering based AI! AAAI-MAKE 2019 invited researchers and practitioners from machine learning and knowledge engineering to explore AI that is explainable, compliant, and grounded in domain knowledge.
The symposium format included presentations of accepted full and position papers, posters, panel discussions, demonstrations, and plenary sessions with breakouts where required, fostering interaction among participants.
Submissions and publication requirements
- Full papers (up to 12 pages) and position papers (3–5 pages) were peer-reviewed.
- Poster proposals required an extended abstract (1–2 pages).
- Discussion proposals (1–2 pages) were to describe their topic, questions, and moderator; panel discussions also needed agreed panel members.
- Demonstration proposals (1–2 pages) were to focus on research or business applications rather than product advertising.
- Submissions had to follow the AAAI Author Kit formatting instructions, go through EasyChair, and undergo program-committee peer review.
- Authors of accepted papers were required to submit camera-ready versions, grant publication permission, and present at the symposium for publication.
Publication arrangements: submissions accepted into the proceedings were published in CEUR Workshop Proceedings Volume 2350. Browse all published 2019 papers and abstracts.
Important Dates
- Extended submission deadline: November 18, 2018
- Notification of authors: December 7, 2018
- Registration for authors and invitees: February 8, 2019
- Camera-ready submission: February 22, 2019
- General registration (first come, first served): March 1, 2019
- Late registration: March 25, 2019
- Symposium: March 25–27, 2019
News
- March 27, 2019 — Final day of symposium with closing session of Knut Hinkelmann and Andreas Martin.
- March 26, 2019 — Second day of symposium with Frank van Harmelen’s keynote and Aurona Gerber’s AAAI plenary session.
- March 25, 2019 — Symposium opening and Doug Lenat’s keynote.
- February 9, 2019 — Registration opened to active participants and interested individuals on a first-come, first-served basis.
- January 27, 2019 — Camera-ready and publication-permission instructions made available.
- January 26, 2019 — First preliminary symposium program published.
- January 11, 2019 — Registration opened for accepted authors, invited speakers, symposium participants, and invitees.
- December 7, 2018 — Review process completed and authors notified.
- November 18, 2018 — Extended submission deadline passed; review process began.
- October 29, 2018 — Submission deadline extended to November 18.
- September 29, 2018 — Second Call for Participation.
- September 7, 2018 — News article by Cycorp, Inc.
- August 31, 2018 — FHNW School of Business blog article.
- August 23, 2018 — WikiCFP announcement.
- August 22, 2018 — EasyChair Smart CFP announcement.
- August 21, 2018 — First Call for Participation.
- August 9, 2018 — SSS-19 EasyChair submission site opened.
- July 19, 2018 — Symposium proposal accepted.
Research Rationale
Knowledge-based systems can make expertise explicit, commonly through logic-based representations, and thereby explain conclusions. Although engineering these representations can entail significant initial effort, symbolic machine learning and ontology learning offer approaches to reduce that effort. Conversely, machine learning methods can derive useful patterns from real-world data when rules are unknown or knowledge is tacit.
In applications where the consequences of decisions are significant, explainability, auditability, and compliance matter. Domain-specific legal and regulatory requirements cannot reliably be inferred from training data alone. The symposium examined the combination of explicit knowledge and statistical learning, including approaches that could improve learning from sparse data and increase robustness against noise.
Participants were encouraged to identify promising areas for collaboration and practical, near-term synergies between the communities.
Published proceedings and symposium outcomes
The 2019 proceedings are available as CEUR Workshop Proceedings, Volume 2350, edited by Andreas Martin, Knut Hinkelmann, Aurona Gerber, Doug Lenat, Frank van Harmelen, and Peter Clark. See also the editors’ preface.
The preface reports more than 70 participants, 28 full and position papers, and four posters and demonstrations. It also records the discussion of hybrid knowledge-based and machine-learning approaches and the closing conversation about continuing the emerging community.
For submission details, see the EasyChair Call for Participation.