Research areas

Recent research topics.

My research focus on several domains enumerated below. Its results can be found in my papers and theses of my students

The studies are carried out in ImpactAI Research Group, HumaNLP Research Group, ENGINE - The European Centre for Data Science, as well as Emognition Research Group  , and formerly Social Network Group.

  • Impact of AI and LLMs on humans, especially long-term
  • Manipulation of LLMs
  • Responsible recommender systems.
  • Large Language Models (LLM)
  • Self-learning LLMs
  • Hallucination of LLMs
  • Emotion recognition from physiological signals and wearables (affective computing)
  • Personalized LLMs. Personalized reasoning for subjective textual content (personalized NLP), among others hate speech, humor, aggressiveness, controversy, conformity, offensivness, etc.
  • Active learning for NLP
  • Emotions and textual content.
  • Machine learning, especially multimodal deep learning, representationj learning (graphs, signals, text), reasoning for linguistic and social networks, collective classification. 

    Former interest domains; recently not explored intensively:

    • Complex networks, especially social networks and social network analysis (SNA).  The main, recent research areas in the domain of SN include:
      • Spread of influence in complex networks, especially in temporal ones
      • Models for temporal networks
      • Analysis of multilayer social networks; their extraction; measures in multilayer SNs; clustering and prediction of groups, etc.
      • Classification in social networks, collective classification
      • Sentiment and emotion analysis
      • Social networks in evaluation of organizational structures
      • Application of social network analysis in marketing, telecommunication, latent knowledge acquisition, and analysis of criminal networks 
    • Prediction of future research domains taht will develop the most in the nearest future.
    • Evaluation of research grants.
    • Multiple model systems and hybrid prediction (classification), multi-label classification, especially for complex structures in output (structured output prediction), e.g. sequences, graphs.
    • Data analyses for medical data, see TRANSFoRm - Translational Research and Patient Safety in Europe. 
    • Recommender systems
      • Project ROSA (Remote Open Site Agents) - web recommender system - the part of the general concept of "adaptive web sites". The prototype not maintained any more.
      • The extension of ROSA was AdROSA - the system for personalized web adversiting (banners).
      • WindOwls - prototype e-commerce on windsurfing, in which 5 recommendation methods were integrated in the personalized and adaptive way; implemented by Paweł Kołodziejski.
      • The International Workshop on Recommender Agents and Adaptive Web-based Systems - RAAWS 2005. A special issue International Journal of Computer Science & Applications, Vol. 3 Issue 3, 2006.
      • Data mining in the web environment - web mining; association rules mining, especially negative ones.
    • Databases for simple devices - modular databases.
    • Aplications and the future of languages derivered from XML.
    • Data security: electronic signature and intrusion detection systems.
    • Content management systems (CMS).
    • Information Retrieval in the Internet environment.
    • Polski