Nofar Carmeli
Technion, Israel and DI ENS, Université PSL, CNRS, Inria, France
Accessing Answers to Conjunctive Queries with Ideal Time Guarantees
(sponsored by SECAI)
When can we answer conjunctive queries with ideal time guarantees? We will start this talk by examining different kinds of query-answering tasks and the connections between them. These tasks include enumerating all answers, sampling answers without repetitions, and simulating a sorted array of the answers. From a data complexity point of view, the ideal time guarantees for these tasks are constant time per answer following a linear preprocessing (required to read the database input). Our goal is to avoid the polynomial preprocessing required to produce all answers. We will then have an example-based discussion of the complexity landscape for these tasks. In particular, we will see how self-joins, constraints and unions can play a crucial role in determining the complexity and designing efficient algorithms.
Francesca Toni
Imperial College London, United Kingdom
Knowledge representation and reasoning in the time of data-centric AI
(sponsored by EurAI)
Data-driven AI has grown massively in the last 10 years or so, predominantly due to increased processing power availability, big data and powerful statistical and probabilistic models to support machine learning and reasoning as vector (rather than symbol) manipulation. In this talk I will explore the role that knowledge representation and reasoning (KR) may have in this landscape, as well as (to a lesser extent) the role that data-centric AI may have for KR. Specifically, I will focus on how KR can support the need for data-centric AI “models” to be verified and explained, so as to overcome any artifacts and biases that may be present in these “models”. Also, KR can contribute to “hybrid” data-centric AI “models” integrating symbolic reasoning components with statistical/neural modules. In addition to these examples of how KR can support data-centric AI, I will also describe uses of data-centric AI for knowledge elicitation. Overall, data-centric AI is an important area of AI research and the KR community can gain lots by engaging with this landscape.
Anni-Yasmin Turhan
(Joint keynote with NMR 2023)
Dresden University of Technology, Germany
Brushing-up description logics to cope with imperfect data
(sponsored by CPEC)
For logic-based applications where data is not curated, but generated automatically, noisy or erroneous data can clearly be an obstacle for reasoning under classical First-order semantics. In recent years several approaches have been investigated for reasoning in Description Logics that deal with this problem — often by changing the underlying semantics. In this talk I will discuss different reasoning problems using non-standard semantics, such as defeasible or approximative semantics, that can preserve useful logical reasoning even in the presence of imperfect data.