Developers insert logging statements into source code to monitor system execution, which forms the basis for software debugging and maintenance. For distinguishing diverse runtime information, each software log is assigned with a separate verbosity level (e.g., trace and error). However, choosing an appropriate verbosity level is a challenging and error-prone task due to the lack of specifications for log level usages. Prior solutions aim to suggest log levels based on the code block in which a logging statement resides (i.e., intra-block features). However, such suggestions do not consider information from surrounding blocks (i.e., inter-block features), which also plays an important role in revealing logging characteristics.

To address this issue, we combine multiple levels of code block information (i.e., intra-block and inter-block features) into a joint graph structure called Flow of Abstract Syntax Tree (FAST). To explicitly exploit multi-level block features, we design a new neural architecture, Hierarchical Block Graph Network (HBGN), on the FAST. In particular, it leverages graph neural networks to encode both the intra-block and inter-block features into code block representations and guide log level suggestions. We implement a prototype system, TeLL, and evaluate its effectiveness on nine large-scale software systems. Experimental results showcase TeLL’s advantage in predicting log levels over the state-of-the-art approaches.