IfI Research Talk Series

The IfI Research Talks will be held at the date indicated below. They usually start at 13:15 (check exact time in the table below) and last for about 30 minutes.

* please contact the speaker for online access

Speaker: Prof. Dr. Dan Olteanu

Abstract:

Cardinality estimation is the problem of estimating the size of the output of a query, without actually computing the query. The cardinality estimator is a critical piece of any query optimizer, and is often the main culprit when the optimizer chooses a poor plan.

In this talk, I will overview LpBound, our new cardinality estimator that computes a guaranteed upper bound on the size of the query output. Besides common statistics used for cardinality estimation, such as the sizes of the input relations, the domain sizes for various columns, histograms, and most common values, LpBound also uses Lp norms on the degree sequences of the join columns. The computed cardinality bound is an optimal solution of a linear program whose constraints encode such statistics and Shannon inequalities.

I will also briefly touch on an experimental evaluation of LpBound for a variety of queries from public benchmarks. LpBound takes time and uses memory very close to the traditional estimators used by PostgreSQL, DuckDB, and a commercial database system. Yet the range of LpBound overestimates is much smaller than the range of the estimates of the traditional estimators, which can both underestimate and overestimate by orders of magnitude. When fed the LpBound cardinalities, PostgreSQL can take less query computation time than when using its own estimates.

This is joint work with Mahmoud Abo Khamis (RelationalAI), Chris Mayer (U Zurich), Dan Suciu (U Washington), and Haozhe Zhang (U Zurich).

Project web page: https://www.ifi.uzh.ch/en/dast/research/LpBound.html

Note: Our theoretical work on LpBound appeared in ACM PODS 2024 and received the SIGMOD Research Highlights Award. Our systems work on LpBound has appeared in ACM SIGMOD 2025 and just received the SIGMOD Best Paper Award.

Speaker: Dr. André N. Meyer

Abstract:

In the evolving landscape of hybrid work, balancing individual focused work with team collaboration presents a significant challenge for knowledge workers. This talk introduces FlowTeams, a novel approach designed to enhance presence awareness within hybrid teams by encouraging the explicit scheduling and sharing of work schedules. By integrating physical and digital displays, FlowTeams consolidates information on co-workers' availability, work hours, and locations, thereby facilitating better team alignment of workdays and reducing intrusive interruptions. Insights from a six-week field experiment involving professionals across 10 hybrid teams will be discussed, demonstrating how FlowTeams supports a more structured and less stressful collaboration environment while preserving individual productivity.

Speaker: Prof. Dr. Claudio Tessone

Abstract:

Blockchain technology has revolutionised the way information, particularly economic transactions, are verified in a decentralised manner. This has been achieved through economic incentives (not always properly designed). These incentives range from protocol-level cryptocurrencies to governance structures and user-generated tokens. However, these incentive mechanisms operate within rigid, pre-programmed frameworks designed by a small set of actors, effectively embedding a form of technocratic control. While incentives are central to blockchain functionality, their real-world impact has often been overshadowed by speculative dynamics rather than sustainable economic utility. We will examine the extent to which current blockchain architectures and incentive models fulfil their promise of decentralisation and economic transformation. Or if they are merely reinforcing financial speculation, to levels unseen before.

Speaker: Dr. Sven Helmer

Abstract:

Relational database systems stopped being the one-size-fits-all solution for data management a while ago. We give a brief overview of this development and then highlight one of the research areas of the Database Technology Group on non-relational data, namely fast access of graph-based data. In recent years, graphs have gained importance as a means to represent data, which is also reflected in the inclusion of Property Graph Queries in the new SQL:2023 standard.

More concretely, we show how to make the Software Heritage Archive, which crawls and stores data on publicly available software repositories, more accessible. Furthermore, we develop efficient and practicable algorithms and data structures for fundamental graph queries, such as connectivity and reachability queries.

Speaker: Prof. Dr. Anikó Hannák

Abstract:

In this short talk, I will introduce the work of the Social Computing Group over the last five years through a selection of our projects. There will be something for everyone, including user-centred studies on various online technologies, the impact of recommender systems, and the use of simulations to complement empirical research.

Speaker: Prof. Dr. Jürgen Bernard and Prof. Dr. Martin Volk

Abstract:

The SDG Research Scout infrastructure project (DSI) uses Natural Language Processing (NLP), Artificial Intelligence (AI), and interactive visual data analysis (IVDA) technology to identify content relevant to the UN Sustainable Development Goals (SDGs) in the University of Zurich's (UZH) scientific publications. This initiative provides UZH researchers and sustainability experts with an infrastructure for mapping and assessing the relevance and contribution of scientific work to the 17 SDGs, such as "Clean Water and Sanitation," "Gender Equality," and "Responsible Consumption and Production."

The system offers interactive exploration of SDG-related publications, enabling flexible access to UZH research. It serves as a pilot model, highlighting the potential of integrating AI into sustainability research and fostering human-AI collaboration, based on IVDA methods for the critical and transparent observation of explainable AI output. Furthermore, the SDG-based classification can be incorporated into the UZH repositories, enhancing publication discovery and supporting sustainability initiatives.

In this talk, Jürgen Bernard (IfI, IVDA) and Martin Volk (CL, NLP) will present and demonstrate project results, describe their interdisciplinary method of how they joined forces together with Lorenz Hilty (IfI, UZH Sustainability Responsible), and invite the audience to a discussion on the necessary steps regarding SDG-awareness at UZH and the general public, enabled through  the SDG Research Scout.

Speaker: Prof. Dr. Alberto Bacchelli

Abstract:

In the majority of software companies and projects, code review is a crucial step in the development process. This involves having developers, other than the original author, manually evaluate any changes to the code before it is integrated into the production software. Despite the widespread adoption of code review in software engineering, there has been a noticeable lack of innovation in the way it is conducted over the past decade or so. This stagnation persists even though there are ample opportunities for refining and rethinking conventional practices.

In this presentation, I will share a series of valuable insights that my team and I have gained through extensive research on code review over the past ten years. Our findings have the potential to significantly enhance current methodologies and tools, leading to more efficient and effective code review processes.

Speaker: Prof. Dr. Giorgia Ramponi

Abstract:

We explore the problem of imitation learning (IL) in the context of mean-field games (MFGs), where the goal is to imitate the behavior of a population of agents following a Nash equilibrium policy according to some unknown payoff function. IL in MFGs presents new challenges compared to single-agent IL, particularly when both the reward function and the transition kernel depend on the population distribution. In this presentation, departing from the existing literature on IL for MFGs, we introduce a new solution concept called the Nash imitation gap. Then we show that when only the reward depends on the population distribution, IL in MFGs can be reduced to single-agent IL with similar guarantees. However, when the dynamics are population-dependent, we provide a novel upper bound that suggests IL is harder in this setting. To address this issue, we propose a new adversarial formulation where the reinforcement learning problem is replaced by a mean-field control (MFC) problem, suggesting progress in IL within MFGs may have to build upon MFC.

Speaker: Prof. Dr. Abraham Bernstein

Abstract:

Recommender systems and social networks are often faulted to be the cause for creating Echo Chambers – environments where people mostly encounter news that match their previous choices or those that are popular among similar users, resulting in their isolation inside familiar but insulated information silos. Echo chambers, in turn, have been attributed to be one cause for the polarization of society, which leads to the increased difficulty to promote tolerance, build consensus, and forge compromises. To escape these echo chambers, we propose to change the focus of recommender systems from optimizing prediction accuracy only to considering measures for social cohesion.

This proposition raises questions in three spheres: In the technical sphere, we need to investigate how to build “socially considerate” recommender systems. To that end, we develop a novel recommendation framework with the goal of improving information diversity using a modified random walk exploration of the user-item graph. In the social sphere, we need to investigate if the adapted recommender systems have the desired effect. To that end, we present an empirical pilot study that exposed users to various sets (some diverse) of news with surprising results. Finally, in the normative sphere, these studies raise the question what kind of diversity is desirable for the functioning of democracy.

Reflecting the consequences of these findings for our discipline, this talk highlights that computer science needs to increasingly engage with both the social and normative challenges of our work, possibly producing a new understanding of our discipline. It proposes similar consequences for other disciplines in that they increasingly need to embrace all three spheres.

Speaker: Prof. Dr. Renato Pajarola

Abstract:

In this talk I will give a brief overview over the research activities in the Visualization and MultiMedia Lab (VMML), focusing in particular on our work on transforming 3D scans into digital architectural CAD models. "As-built" architectural building information models (BIMs) are critical to represent the current state and condition of building structures in the fields of architectural construction, building maintenance, or civil engineering. However, reconstructing an architectural CAD model of the interior rooms of an existing building from scanned 3D point cloud data is a challenging research problem. An effective approach must be able to faithfully capture and model the architectural structures and separate permanent components (e.g. walls) from clutter (e.g. furniture), while at the same time dealing with defects (e.g. noise) in the input data. To solve this ill-posed problem, many methods assume constrained input environments which significantly restricts their applicability in real-world settings. We present an approach that allows us to reconstruct 3D interior architectural models, significantly increasing the applicability to more realistic scenarios.

Speaker: Prof. Dr. Martin Volk

Abstract:

In the first part of my talk I will look back on my group’s activities of building and exploiting large multilingual corpora with texts that span more than 100 years, and on our research in machine translation.

In the second part, I will present our recent work on a large collection of 16^th century letters from and to the Zurich reformer Heinrich Bullinger. We investigated handwriting recognition, automatic translation, name recognition and linking, and topic classification for these 12,000 letters. I will show that ChatGPT performs well on most these tasks even though we are dealing with Latin and Early New High German. I will also speculate on how ChatGPT can reconstruct lost letters in this collection.

Speaker: Prof. Dr. Ingo Scholtes

Abstract:

Graph Neural Networks (GNNs) have become a cornerstone for the application of deep learning to data on complex networks. However, we increasingly have access to time-resolved data that not only capture which nodes are connected to each other, but also when and in which temporal order those connections occur. A number of works have shown how the timing and ordering of links shapes the causal topology of networked systems, i.e., which nodes can influence each other over time. Moreover, higher-order models have been developed that allow us to model patterns in the resulting causal topology. While those works have shed light on the question how the time dimension of temporal graphs influences node centralities, community structures, or diffusion processes, we lack methods to incorporate those insights into state-of-the-art graph learning techniques.
 
Addressing this gap, we introduce De Bruijn Graph Neural Networks (DBGNNs), a time-aware graph neural network architecture for temporal network data. Our approach accounts for temporal-topological patterns that unfold via causal walks, i.e., temporally ordered sequences of links by which nodes can influence each other over time. We develop a graph neural network architecture that utilizes De Bruijn graphs of multiple orders to implement a message passing scheme that follows a non-Markovian dynamics, which enables us to learn patterns in the causal topology of dynamic graphs.

Speaker: Prof. Dr. Manuel Günther

Abstract:

When image classification methods leave academic lab settings, they will be confronted with items that they have never seen before.

Open-set recognition tries to balance between correctly classifying samples from known categories, while not getting confused by unknown items.
In this presentation, I will give a brief introduction to current methods applied to improve open-set recognition.

Furthermore, I will present our new large-scale evaluation framework for open-set image recognition tasks, and the latest findings from these evaluations.

Particularly, when unknown items are very dissimilar to the known classes, they can be rejected easily.
On the other hand, when known and unknown classes come from the same domains, none of our research efforts in the last 10 years provides any real advantage.

Speaker: Prof. Dr. Dan Olteanu

Abstract:

In this talk I will overview recent efforts of the DaST (Data Systems and Theory) group on the problem of maintaining real-time analytics over large and continuously evolving relational databases.

We looked at the maintenance problem from different angles. We addressed several fundamental questions, including:

• Which database queries can be maintained in worst-case optimal time?
• Can we trade the query answer time for fast update time?
• How can we maintain models trained over changing relational data?

Our answers include:

• An algorithm with the best possible (worst-case) computational complexity for maintaining the number of triangles in a graph under insertions and deletions of edges.
• The computational complexity trade-off between update time and the answer time for a large class of queries.
• A new algebraic structure called the covariance ring that defines the shared maintenance of the covariance matrix for features in the database.

Equipped with a theoretical understanding of the maintenance problem, we focused on the design and implementation of maintenance systems:

• F-IVM (https://github.com/fdbresearch/FIVM) is our open-source state-of-the-art system. Using one thread of a commodity machine, F-IVM can maintain queries and linear regression models over database joins with a throughput of ten million updates per second.
• The RelationalAI commercial engine (https://relational.ai) incorporated some of our theoretical and system design ideas.

This is joint work with DaST members Ahmet Kara and Haozhe Zhang, Milos Nikolic of U. Edinburgh, and Henrik Barthels, Mohamed ElSeidy and Niko Göbels from the RelationalAI IVM team.

Speaker: Prof. Dr. Jürgen Bernard

In his Talk, Prof. Jürgen Bernard motivates the benefits of human-machine collaboration to solve data science and analysis challenges. Along these lines, he outlies typical bottlenecks in the human-machine collaboration, and gives indications on how his research helps solving these problems.

In particular, Jürgen Bernard emhpasizes the potentials of leveraging preferences of humans when it comes to data analysis, and discusses how interactive machine learning systems can replay the information that was learned from humans expressing feedback.

In two real-world examples, Jürgen Bernard shows how bi-directional human-machine feedback can be applied in practice, to leverage data preferences: for the preference-based creation of item rankings and human-centered similarity search.

Links: IVDA Group  |  Digital Society Initiative