Chrysanthi Kosyfaki

About Me

I am a postdoctoral research fellow in the Department of Computer Science and Engineering at Hong Kong University of Science and Technology. Previously, I was a postdoctoral research associate at the University of Hong Kong. I obtained my Ph.D. degree from University of Ioannina, Computer Science and Engineering Department, working with Prof. Nikos Mamoulis.

Research Interests

My research interests are in the area of graph analytics with a particular focus on a new class of networks called Temporal Interaction Networks (TINs). In these networks, a quantity (flow) transfers from one vertex to another within a time window. TINs can model dynamic systems where entities exchange quantities, such as financial transactions, transportation flows, or communication data. Over the past few years, I have been developing efficient solutions to optimize classical problems—such as max-flow computation—by introducing the temporal dimension.

I have also explored the Text-to-SQL domain, focusing on addressing challenges related to text ambiguity in natural language database queries.

Currently, I am working on provenance tracking for streaming systems. As real-time analytics pipelines grow in complexity, understanding how outputs are derived from inputs becomes critical for debugging, auditing, and explainability. I am developing how-much provenance, a new model that efficiently counts input contributions across dataflow paths rather than enumerating individual tuples. This approach incorporates temporal window-based tracking and path compression techniques to scale to high-throughput streaming workloads, and supports queries that allow users to identify which inputs (e.g., from which paths or time windows) contributed to a given output.