vue4logs -- Automatic Structuring of Heterogeneous Computer System Logs

Understand

Computer system log data is commonly used in system monitoring, performance characteristic investigation, workflow modeling and anomaly detection.

• Log data is inherently unstructured or semi-structured, which makes it harder to understand the event flow or other important information of a system by reading raw logs.
• The process of structuring log files first identifies the log message groups based on the system events that triggered them, and extracts an event template to represent the log messages of each event.
• This paper introduces a novel method to extract event templates from raw system log files, by using the vector space model commonly used in the field of Information Retrieval to vectorize log data and group log messages into event templates based on their vector similarity.