TransTVDiag is a Transformer-based multi-modal failure diagnosis method for microservice systems. It includes Failure Type Identification (FTI) and Root Cause Localization (RCL).
Our method is an enhancement of TVDiag. The experiments are based on the GAIA dataset.
Kiss, Áron, and Nehéz, Károly. "A Graph Transformer-Based Framework for Multi-Modal Failure Diagnosis in Microservice Systems," International Journal of Cloud Applications and Computing (IJCAC) 16, no.1: 1-28. https://doi.org/10.4018/IJCAC.402208
@article{Kiss2026,
title = {A Graph Transformer-Based Framework for Multi-Modal Failure Diagnosis in Microservice Systems},
volume = {16},
ISSN = {2156-1826},
url = {http://dx.doi.org/10.4018/IJCAC.402208},
DOI = {10.4018/ijcac.402208},
number = {1},
journal = {International Journal of Cloud Applications and Computing},
publisher = {IGI Global},
author = {Kiss, Áron and Nehéz, Károly},
year = {2026},
month = mar,
pages = {1–28}
}Feel free to leave messages in email: aron.kiss@uni-miskolc.hu