Project ACKnowledge focuses on artificial intelligence (AI) and natural language processing (NLP) fields, emphasizing creating environments for constructing virtual assistants in diverse areas of expertise. This project seeks to overcome the limitations of large language models (LLM) by quickly and securely adapting them to new enterprise applications.

One notable feature of the project is its commitment to innovation and practical and valuable solutions for the real world. Existing LLMs contain many parameters but struggle to answer specific questions about particular business sectors. The project seeks to facilitate adapting AI systems to new application domains within businesses by implementing adaptive modules that can incorporate knowledge from multiple sources.

Additionally, the project concentrates on developing prompt engineering and knowledge graph development to advance the capacity of LLMs to understand and respond to queries in various fields of expertise while considering the cognitive-emotional state of users, along with an iterative dataset and fine-tuning mechanism that incorporates responses of non-present values found in initial information.

Recognizing the importance of protecting personal and commercial information in a business context, Project ACKnowledge prioritizes the privacy and confidentiality of the data employed during the development of this Virtual Assistant Development Kit.

Project ACKnowledge represents a significant step toward improving and expanding generative AI, creating tools and solutions that can easily, quickly, and securely adapt to new business domains.

Project Consortium: Nokia TECSS, UPM-GSI, Technalia.

Funding body: Ministry of Science, Innovation and University.

Project of Public-Private Collaboration - CPP2023-010437

Budget: 366.653€