Graph-Massivizer: Sustainable Graph Processing of Extreme Data
Graphs are widely used to model complex systems such as social networks, transportation networks, and biological systems. However, as the size of these graphs grows, processing and analyzing them becomes increasingly challenging. This is where Graph-Massivizer comes in. This research project aims to develop algorithms that can efficiently handle massive graphs with billions of vertices and edges. Ultimately, this will support a climate-neutral and sustainable economy based on graph data.
Software toolkit & commercialization goals
The Graph-Massivizer project consortium will develop innovative products, particularly for use cases that can be commercialized after the project ends. These products will be based on a software platform consisting of five integrated tools for extreme data processing that will:
- translate extreme data streams or follows heuristics to generate synthetic data and persist it within a graph structure.
- use probabilistic reasoning and AI algorithms for graph pattern discovery, low-footprint graph generation, and low latency error-bounded queries.
- help co-design the most promising processing infrastructure with guaranteed performance and energy consumption estimates for specific workloads.
- use operational data centres and national energy supplier data to simulate sustainability profiles for operating graph workload analytics at scale.
- use the performance and sustainability models to deploy and orchestrate the graph analytics workloads on the computing continuum.
Furthermore, the project consortium aims to create an integrated platform that is user-friendly and easy to deploy in enterprise environments, using the metaphactory knowledge graph platform as a basis. The platform will tightly integrate the tools developed by Graph-Massivizer to provide a comprehensive offering.
Use cases validation in finance, environment protection, manufacturing, and high-performance computing sectors
To ensure applicability and scalability in real-world scenarios and the feasibility of commercial solutions developed on top of metaphactory as a result of the Graph-Massivizer project, the project partners will validate the innovative toolkit on four use cases that cover the economic, societal and environmental sustainability pillars:
- sustainable green finance,
- global environment protection foresight,
- green artificial intelligence (AI) for the sustainable automotive industry, and
- data centre digital twin for exascale computing.
These use cases tackle extreme data processing and massive graph analytics challenges and are, therefore, a perfect fit for the Graph-Massivizer toolkit.
Across these use cases, Graph-Massivizer aims to improve analytics efficiency by 70% and energy awareness for extract-transform-load (ETL) storage operations by 30%. Furthermore, it aims to demonstrate a possible two-fold improvement in data centre energy efficiency and over 25% lower greenhouse gas (GHG) emissions for basic graph operations.
Consortium & funding
Led by the University of Klagenfurt and composed of 12 partners from 8 EU countries, the project brings together the world-leading roles of European researchers in graph processing and serverless computing and uses leadership-class European infrastructure in the computing continuum.
The project is funded by 'Horizon Europe', the European Union’s key funding programme for research and innovation. Among many other R&D topics, Horizon Europe tackles climate change and helps to achieve the UN’s Sustainable Development Goals. It also aims at boosting the EU’s competitiveness and growth.
KIRA: AI methods for optimized control of electric traction drives
The market for electrically powered vehicles is steadily growing and will play a central role in reaching climate neutrality. However, the limited range of electric vehicles remains a major obstacle for many customers. In addition to the use of increasingly large, resource-intensive batteries, the efficient use of the available energy is an alternative way to increase this range.
The KIRA project aims to improve the operation of electric traction drives through the use of AI-based methods. To this end, the project consortium is working towards developing innovative control and activation concepts that will lead to an increase in efficiency, an increase in power density, a reduction in noise development, and an increase in torque accuracy.
As part of the KIRA project, metaphacts will focus on the formal representation of domain-specific knowledge using knowledge graphs for improved model building. Knowledge graphs allow for explicit knowledge representation at a conceptual level and provide the basis for the development of components needed to deliver explainable AI. metaphacts develops hybrid approaches for the integration of ML-based methodologies with knowledge representation.
IIDI - Intelligent information services for domain-specific Web portals
Obtaining relevant information on technological and market developments quickly and in a targeted manner is becoming an increasingly important success factor for companies and their technology and innovation management (TIM). However, dynamic technological change coupled with an exponentially growing flood of information is presenting companies with major challenges. For example, employees in most innovation-related jobs spend five to ten hours a week researching relevant information.
The main goal of the IIDI research project is to develop specialized, reusable and extensible functional digital services that can be modularly orchestrated as needed and can support, e.g., data aggregation, preprocessing, or semantic text analysis. These information services will be enriched with ML processes and can integrated into Web portals to deliver relevant information to end users in a timely manner. As an application area, the IIDI research project will focus on the domain of hydrogen technology for smart city portals.
metaphacts has a leading role in the IIDI project as a technology and integration partner. The metaphactory platform is central to the overall architecture of the solution to be developed and piloted.
KompAKI: Competence center for work and artificial intelligence
Artificial Intelligence (AI) is increasingly being used in software applications and has the potential to revolutionize the way we work. However, research regarding the impact of new technology on the workplace is often performed after said technology has already been developed and implemented. In KompAKI (Kompetenzzentrum für Arbeit und künstliche Intelligenz im Rhein-Main-Gebiet), the research consortium is researching new potential for human-centered AI applications and their business models.
The knowledge gained at the Competence Center will be validated in pilot projects with partner companies mainly from the production sector. All project results will be made available to the regional working world and university education through various information and transfer formats during the project period.
metaphacts is involved in central issues within the joint project, focusing on "cooperative artificial intelligence". The main characteristics of cooperative AI is that it should be possible for users without knowledge in the field of artificial intelligence to train learning systems ("Automated AI"), but also that learning systems can explain themselves ("Explainable AI") and adapt through interaction with the user ("Interactive AI"). Specifically, metaphacts' goal is to develop methods for cooperative AI using knowledge graphs that provide explanation components and support the creation of new interfaces for non-expert users to interact with knowledge-based intelligent systems.
FROCKG: Fact checking for large enterprise Knowledge Graphs
Enterprise Knowledge Graphs underpin business-critical decisions, thereby influencing financial markets, the economic environment, and many more aspects of our lives. Determining the veracity of the facts in these Enterprise Knowledge Graphs is hence mission-critical.
Together with partners from industry and academia, in FROCKG we aim at building a framework to enable companies to ensure they are basing business decisions on reliable and true data. For this, we are developing accurate and time-efficient approaches to quantify and provide evidence for how true or untrue facts in Enterprise Knowledge Graphs are likely to be. A prototype of this framework will be integrated in commercial solutions and deployed in three use cases: finances, cultural heritage and linked open data.