Goal Recognition Benchmark Dataset

该数据集是一个为评估目标识别任务而创建的基准数据集，它通过适配现有的基准测试，包含了不同级别的可观测性，并引入噪声以测试目标识别方法的鲁棒性。该数据集包含了最优和次优计划，这些计划具有不同的可观测性水平（10%、30%、50%、70%、100%），并相应地包含了带噪声的数据集。整个数据集被划分为四个子集，共包含8,288个目标识别任务。这些任务的目的是进行目标识别。

This dataset is a benchmark dataset developed for evaluating object recognition tasks. It adapts existing benchmark test suites, integrates multiple levels of observability, and introduces noise to test the robustness of object recognition methods. The dataset encompasses both optimal and suboptimal plans, which feature observability levels of 10%, 30%, 50%, 70% and 100% respectively, along with corresponding noisy datasets. The full dataset is partitioned into four subsets, which collectively contain 8,288 object recognition tasks, all dedicated to object recognition.