The discovery of temporal causality is essential for identifying the root causes of problems in real-world systems. In practice, causality can arise from two primary sources: time series (TS) data and discrete events. Although many studies have explored these two modalities independently, existing approaches typically focus on one or the other. To our knowledge, no existing framework unifies causal relationships derived from both TS and events. In this paper, we propose UniCausal, a unified framework for constructing hierarchical causal graphs from both TS and event data. Our approach integrates symbolic state-change events derived from raw TS with predefined discrete events, enabling multi-level causal inference across heterogeneous data sources. We further advance causal discovery by introducing a novel deep model that incorporates symbolic context conditions, enabling more interpretable and context-aware causal reasoning. To demonstrate the framework’s practical utility, we present a case study in which UniCausal is combined with a Large Language Model (LLM) to perform interpretable root cause analysis. This case study emphasizes application-oriented reasoning in a real-world industrial scenario, highlighting the potential of UniCausal for transparent, human-in-the-loop diagnostics. While our case study focuses on industrial systems, the framework is broadly applicable to domains such as IoT, healthcare, and any other field where TS and event data co-exist.