Large-scale Heterogeneous Information Integration: Using machine-learning techniques for schema integration across multi-model data
Multi-Model Query Languages: Development of a type-aware dataflow-style query language that operates over relational, graph, document, vector databases and analytical functions
Query Optimization across Multiple Data Models: Development of new cross-model query operators and cross-model optimization based on learned cost models
Construction of Knowledge Graphs: Using ML-based techniques to create ontologically grounded virtual knowledge graphs
Conversational AI Systems
Proactive Conversational Recommendation: Development of a model to estimate user’s knowledge gap in a conversation and provide adequate information to bridge the gap
Mission-Driven Conversation Flows: Development of algorithms to steer a conversational consultant to meet a sequence of targets
Adaptive User Modeling: Development of a semantic graph model to incrementally build a user profile from a sequence of conversational text.
Knowledge Representation and Semantic Search
Entity and Event-Driven Knowledge Representation: Entity and event and relationship learning from heterogeneous information sources
Progressive Semantic Search Engine: Development of algorithms to enable knowledge-based information seeking tasks in digital libraries
Question Answering Over Regulatory and Legislative Documents: Development of a semantic model for representation of legal knowledge and enabling question answering based on this model
Innovation Landscape Analysis: Development of an analytical platform to explore innovations from patents, publications and grants
AI-Driven Decision Support
Business Ideation and Recommendation: Development of conversational recommendation engine that enables potential entrepreneurs to decide on businesses that suit their personal profiles as well as market conditions
Domain-Specific Advisory Systems: An extension of the recommendation engine to be tune for new business domains
Context-Aware Information Delivery: An algorithm to balance complexity of information content and cognitive overload in a conversational recommendation system
Information Security and Privacy for Heterogeneous Information
Access Control for Multi-Model Data: We design fine-grained access control models that support heterogeneous data systems, enabling secure and policy-driven access to structured, semi-structured, and unstructured data across multiple models.
Privacy-Preserving Query Processing: We develop privacy-aware query validation and execution techniques that allow users to retrieve data securely, ensuring sensitive information is protected and regulatory requirements are met.
Secure Knowledge Integration: We integrate information from diverse structured and unstructured sources while preserving semantic integrity and enforcing data security policies to ensure consistent and trusted knowledge synthesis.
Constraint-Aware Data Handling: We implement data management systems that respect regulatory, contextual, and logical constraints, ensuring that all operations — from storage to access — adhere to compliance requirements and support trustworthy decision-making.
Digital Health
Sensor-Driven Health Monitoring: Developing scalable platforms for collecting, processing, and analyzing continuous physiological data from wearable and clinical sensors to enable early detection, personalized care, and real-time interventions.
Secure and Interoperable Infrastructure: Building secure, privacy-preserving architectures that integrate diverse health.
AI/ML for Digital Medicine: Applying machine learning and knowledge graphs to uncover patterns in multimodal health data, supporting predictive modeling, patient stratification, and data-informed clinical decision-making.
Critical and Non Critical Intervention: We design adaptive digital health systems that differentiate between critical and non-critical interventions, enabling timely, context-aware responses across acute and chronic care scenarios.
