Conférences  >  Informatique  >  Gestion des connaissances, Big Data Computing  >  Allemagne

Mathematisches Forschungsinstitut Oberwolfach (MFO, Oberwolfach Research Institute for Mathematics)

Réseaux de neurones et intelligence artificielle, apprentissage automatique,    Calcul scientifique et science des données

Schloss Dagstuhl - Leibniz-Zentrum für Informatik

The central topic of our Dagstuhl Seminar is the development, implementation, and evaluation of new AI technologies to support decision-making in management. A distinguishing feature is therefore innovative algorithms from the field of AI (e.g. explainable AI, generative AI, large language models, probabilistic ML, causal ML, etc.) that enable new insights in practice and beyond. This holds great potential for informing and improving decision-making.

Schloss Dagstuhl - Leibniz-Zentrum für Informatik

Conversational Agents (CA) as frontends to Information Retrieval (IR) and Recommender Systems (RS) become more popular in everyday life, with a wider range of users and usages. The latest developments in Large Language Models (LLMs) will have tremendous consequences, especially for the workplace and education. In this Dagstuhl Perspectives Workshop, we want to focus on the evaluation of these conversational systems, as appropriate methods are still missing. The quality of these systems is limited in terms of personalization, veracity and correctness, bias, transparency, trustworthiness, and understandability. Thus, evaluation methods must address these shortcomings. Furthermore, user- and usage-oriented aspects should become a more prominent and integral component in evaluations, as the user population as well as the tasks these systems are used for become more heterogeneous. For this reason, the topic-centric view of relevance has to be extended to a broad range of facets which are important for the different usage scenarios. Therefore, suitable evaluation criteria have to be specified, which form the basis for defining appropriate measures. Most importantly, the range of evaluation methods must be revisited and extended, as popular methods like the Cranfield approach or crowdsourcing must be complemented by new evaluation methods and strategies specifically tailored to this new type of system.

Traitement automatique des langues, Linguistique informatique,    Mise en œuvre et applications de l'intelligence artificielle

Schloss Dagstuhl - Leibniz-Zentrum für Informatik

Modern algorithms for data analysis require the use of advanced probabilistic methods to achieve desirable scalability and accuracy guarantees. At the same time, modern data-analysis tasks require the use of advanced statistics to handle challenges such as testing for multiple hypotheses or identifying dependencies among data points, such as in time series or graphs. Probabilistic methods are also at the core theoretical computer-science areas, such as sublinear algorithms and average-case analysis. To obtain efficient data-analysis algorithms, probabilistic methods require careful balancing of theoretical and practical aspects. This Dagstuhl Seminar brings together researchers interested in statistical and probabilistic methods to design and analyze scalable algorithms for discovering knowledge in large and rich datasets. We plan to cover the following topics, among others.

Statistique, Théorie des jeux,    Calcul scientifique et science des données

Schloss Dagstuhl - Leibniz-Zentrum für Informatik

Data-sharing across domain and infrastructure boundaries is becoming more common, and significant work has been done in aligning many of the standard model describing the information and data being exchanged. In order to guarantee the scalability and practicality of data-sharing implementations, these standard models need to be aligned with the emerging technology architectures designed to support exchange networks at this scale. This workshop will look at how the standard metadata models for cross-domain use - including DDI-CDI and related models - and architectural approaches can best be used to leverage the strengths of both.

Schloss Dagstuhl - Leibniz-Zentrum für Informatik

Through a series of Dagstuhl hosted workshops, and the CODATA-DDI collaboration, we have developed a framework for cross-domain interoperability, published by the EC-funded WorldFAIR project. This framework is being employed both within data intensive science and and between the scientific world and the world of official data, collected to inform policy making, including in relation to the Sustainable Development Goals and other components of the UN Agenda. The effective use and uptake of such a framework will require methodologies to evaluate implementations. This workshop will explore such approaches, building on current work for FAIR assessment, and the experience gained from implementation in a range of case studies, including those in WorldFAIR. The workshop will focus on improvements to evaluation metrics, and to the metadata exchange frameworks themselves.

Schloss Dagstuhl - Leibniz-Zentrum für Informatik

The vision of NFDI4DS is to support all steps of the complex and interdisciplinary research data lifecycle in Data Science and Artificial Intelligence. The overarching objective of NFDI4DS is the development, establishment, and sustainment of a national research data infrastructure for the Data Science and Artificial Intelligence community. The key idea is to work towards increasing the transparency, reproducibility and fairness of Data Science and Artificial Intelligence projects, by making all digital artifacts available, interlinking them, and offering innovative tools and services. Within this Dagstuhl Research Meeting, the NFDI4DS partners will reflect on their progress so far, and will work on their long-term strategy.

Réseaux de neurones et intelligence artificielle, apprentissage automatique,    Calcul scientifique et science des données

Schloss Dagstuhl - Leibniz-Zentrum für Informatik GmbH

Process mining and visual analytics are separate disciplines with the common goal of helping humans gain insight into and extract knowledge about relevant phenomena from complex data. Process mining (PM) is a rapidly growing discipline blending machine learning and data mining concepts with ideas taken from the field of business process management (BPM). It utilizes event data recorded by IT systems that support business process execution for a variety of tasks, from the automated discovery of graphical process models to operational support. Visual Analytics (VA) is a multidisciplinary approach that combines interactive, visual, and analytical methods to make complex phenomena more comprehensible, facilitate new insights, and enable knowledge discovery. VA research happens at the intersection of data mining and knowledge discovery, information visualization, human-computer interaction, and cognitive science.

Schloss Dagstuhl - Leibniz-Zentrum für Informatik GmbH

In today’s digital landscape, complexity grows with increasing data volume and degree of interconnection. A suitable data abstraction is crucial for comprehending and navigating this dense network of connections. Starting from Euler’s pioneering work on The Bridges of Konigsberg in 1735, graphs have steadily evolved as a robust and adaptable conceptual framework. Graphs are universal representations of concepts, where nodes are markers for distinct entities and edges delineate their interrelations, further enriched with detailed annotations when necessary. Graphs are successful in various domains, like bioinformatics, e-commerce, logistics and transportation networks, urban planning, and even pandemic analysis or vaccine development (e.g., during COVID-19).

Schloss Dagstuhl - Leibniz-Zentrum für Informatik GmbH

The increasing amount of data being collected, stored, and analyzed induces a need for efficient, scalable, and robust methods to handle this data. Representation learning, i.e. the practice of leveraging neural networks to obtain generic representations of data objects, has been shown effective for various applications over data modalities such as images and text. More recently, representation learning has shown initial impressive capabilities on structured data (e.g. relational tables in databases), for a limited set of tasks in data management and analysis, such as data cleaning, insight retrieval, and data analytics. Most applications traditionally relied on heuristics and statistics, which are limited in robustness, scale, and accuracy. The ability to learn abstract representations across tables unlocked new opportunities, such as pretrained models for data augmentation and machine learning, that address these limitations. This emerging research area, which we refer to as Table Representation Learning (TRL), receives increasing interest from industry as well as academia , in particular in the communities of data management, machine learning, and natural language processing. This growing interest is a result of the high potential impact of TRL in industry given the abundance of tables in the organizational data landscape, the large range of high-value applications relying on tables, and the early state of TRL research so far.

Schloss Dagstuhl - Leibniz-Zentrum für Informatik GmbH

This summer school is an intensive course on software techniques for data management and database management. Topics include database command translation and query optimization, query execution and indexing, row- and column-oriented storage structures and algorithms, optimistic and pessimistic concurrency control, logging and recovery, replication and high availability, scalability and cloud computing, and more. The goal of lectures and presentations is to understand classic techniques and to link them to industrial reality and to research opportunities. The target audience are recent MS graduates who have already taken 2-3 courses covering data management software and who are embarking on a career in industrial development or in academic research in related topics.

Evènements pour étudiants en informatique,    Conception et gestion des bases de données, Moteurs de recherche et extraction de connaissances à pa

Schloss Dagstuhl - Leibniz-Zentrum für Informatik GmbH

The Autumn School for Information Retrieval and Information Foraging (ASIRF) focuses on recent developments in Information Retrieval and Information-Seeking behavior and their applications. Information Retrieval deals with searching in huge amounts of unstructured data. In many professional fields of work and everyday life, access to text data via search engines is one of the most important forms of access to knowledge. Information Foraging aims to model the entire process from the awareness of an information problem through selecting appropriate information sources to the actual search.

Students and lecturers will live together for a week at Schloss Dagstuhl, which guarantees the school's traditionally highly communicative and interactive atmosphere. The intended audience of ASIRF are doctoral students and young researchers working in Information Retrieval, Information Foraging, or using concepts or methods from these fields in their research

Evènements pour étudiants en informatique,    Conception et gestion des bases de données, Moteurs de recherche et extraction de connaissances à pa

Schloss Dagstuhl - Leibniz-Zentrum für Informatik GmbH

More and more, we find that the provenance of data is a key missing ingredient in the information set made available to us. Provenance - the origination and intention of data as collected - provides significant challenges for modern networks. While standards exist for describing data and its analysis - and even for parts of the provenance chain - there is no standard mechanism for tying together all relevant information across the series of steps which fully describe the origination of data. This workshop explores how the many contextualising factors needed to support responsible and accurate data reuse can be described, allowing for the needed chaining of provenance steps.