Project data
| Project start date | 2024-12-01 |
| Project end date | 2027-11-30 |
| Project cost | 16918839.55 PLN |
| Project amount of funding | 9200000 PLN |
Lider
- POWSZECHNA KASA OSZCZĘDNOŚCI BANK POLSKI SPÓŁKA AKCYJNA
Business Partner
- The National Information Processing Institute
The project’s advanced AI R&D is being led by AI Lab, a dedicated laboratory at OPI PIB that specialises in Polish neural language models.
The project seeks to implement intelligent, cutting-edge methods that rely on neural language models to process and analyse unstructured data. The partners plan to develop a new generative model for the Polish language that will be optimised for use in the financial sector. They also plan to create a dedicated service platform that generates and analyses linguistic content related to banking.
This will lead to improved text clustering and retrieval, better result summaries, simplified language, and more advanced conversations with users.
The project will deliver:
- a universal architecture model for PKO BP’s service platform
- a data model, physical data organisation, and tangible structures
- representation and generative models for the Polish language that are optimised for use in the financial sector
- a new clustering method
- a new few-shot classification method
- a new text representation encoder model
- a new reranker of banking-related search results
- a natural language search-result-based response generator
- a domain-specific text generation model for multiple diverse instructions.
The project will address the needs both of PKO BP’s clients and of the bank itself. The new solutions will help identify and resolve issues that arise during conversations with bank clients. They will also support PKO BP internally by enabling consultants to answer questions without having to rely on expert helpdesk employees.
The project is co-funded by the European Union under the European Funds for Smart Economy programme.
