forked from KEMT/zpwiki
53 lines
1.8 KiB
Markdown
53 lines
1.8 KiB
Markdown
---
|
|
title: Legal Data Processing
|
|
category: [project]
|
|
tag: [legal,nlp]
|
|
---
|
|
|
|
# Legal Data Processing Scientific Project
|
|
|
|
Goal:
|
|
|
|
- Help the slovak judges
|
|
|
|
Possible applications:
|
|
|
|
- agent retrieval agent to search in laws and processdings.
|
|
- question answering system
|
|
|
|
Plan:
|
|
|
|
- Develop a demo agent for legal and proceeding retrieval anad answering.
|
|
- Develop a summarization system - retrieve and highlight important parts of document. Generate a summary.
|
|
|
|
|
|
People:
|
|
|
|
- Daniel Hládek
|
|
- [Martin Šarišský](/students/2021/martin_sarissky) - agentový systém
|
|
- [Matej Ščišľak](/students/2021/matej_scislak) - sumarizácia, dotrénovanie LLM
|
|
- [Maxim Zatyrka](/students/2023/maxim_zatyrka) - využitie grafovej informácie ( Tezaurus)
|
|
- [Oleksandr Dorybohov](/students/2023/oleksandr_dorybohov) - agent pre volanie REST
|
|
- [Simona Bobrovčanová](/students/2023/simona_bobrovcanova) - sumarizácia
|
|
|
|
Tasks:
|
|
|
|
- Create index of Slovak laws
|
|
- Create index of the Slovak Court proceedings.
|
|
- Fine-tune Slovak embedding model for the legal domain.
|
|
- Evaluate LLMS for the legal domain.
|
|
- Design legal data pre-processing (anonymization, de-anonymization)
|
|
- Create a knowledge graph (tezaurus)
|
|
- Create legal data summarization (extractive and abstractive) system.
|
|
- Create agent that utilizes OpenAPI
|
|
|
|
Links:
|
|
|
|
- OpenAPI pre prístup k údajom o súdoch a o rozsudkoch. https://obcan.justice.sk/pilot/api/ress-isu-service/swagger-ui/index.html
|
|
- https://www.justice.gov.sk/sluzby/register-partnerov-verejneho-sektora/open-data/
|
|
- https://rpvs.gov.sk/opendatav2/swagger/index.html
|
|
- Databáza zákonov a register opatrení https://static.slov-lex.sk/
|
|
- Tezaurus slovenského práva https://www.slov-lex.sk/tezaury/terminy - aj vzťahy ku právnym predpisom
|
|
- Znalostný graf údajov verejnej správy https://znalosti.gov.sk/
|
|
|