# Convolve Epoch-1 — Data

来源：https://www.kaggle.com/competitions/convolve-epoch1/data

采集时间：2026-10-19T07:50:19Z

This contest is about **Log Anomaly Detection**

In computing, logging is the act of keeping a log of events that occur in a computer system, such as problems, errors or just information on current operations. These events may occur in the operating system or in other software. A message or log entry is recorded for each such event. Log Anomaly Detection is simply detecting anomalies in logs deposited by softwares using Machine Learning.

Anomaly is anything that is different from what is usually perceived as normal - an exception. In software engineering, anomaly can be defined as occurrence of rare or unexpected events that does not fit into the normal patterns and hence a suspicious one. Examples could be:
- Unexpected failure of a service
- Sudden increase/decrease of user activity in a customer facing system
Reasons for these anomalies could be different, like:
- Entry of unperceived inputs to the system
- System running out of memory

You have been given a training dataset and a testing dataset containing logs generated by softwares. The training dataset is in **JSON** format where each key is a single software log  and coresponding value is the label for that log.The labels for logs are **"abnormal" and "normal"**. The testing dataset is in **CSV** format with 2 columns corespondinig to a Unique ID and a software log. Your task is to train a ML model on the given training data that can predict whether a given log in testing data is an anomaly or normal. Unique identifier is assigned to each log in testing dataset for ease of evaluation and submission. 

Please note that software logs have a massive number of lines and can be very difficult to handle without crashing memory and RAM.


## Files

*   **train.json** - the training dataset
*   **test.csv** - the test dataset
*   **sample_submission.csv** - a sample submission file in the correct format