ADTEC
收藏资源简介:
ADTEC是由CyberAgent和奈良科学技术研究所共同创建的,用于评估搜索引擎广告文本质量的统一基准数据集。该数据集包含五个任务,涵盖了广告文本的接受性、一致性、性能估计、A3识别和相似性评估。数据集来源于实际的广告操作流程,包括人工创作和自然语言生成模型的输出。创建过程中,数据经过了严格的预处理和标注,确保了数据的质量和实用性。ADTEC数据集主要用于解决广告文本质量评估的问题,特别是在大规模广告操作中的自动质量评估需求。
ADTEC is a unified benchmark dataset jointly created by CyberAgent and the Nara Institute of Science and Technology for evaluating the quality of search engine advertising texts. This dataset includes five tasks covering acceptability, consistency, performance estimation, A3 recognition, and similarity assessment of advertising texts. The dataset is sourced from real-world advertising operation workflows, encompassing both human-authored ad copies and outputs from natural language generation models. During its creation, the data underwent strict preprocessing and annotation to guarantee its quality and practical utility. The ADTEC dataset is primarily developed to tackle the problem of advertising text quality assessment, especially the demand for automated quality evaluation in large-scale advertising operations.

- 1AdTEC: A Unified Benchmark for Evaluating Text Quality in Search Engine AdvertisingCyberAgent 和 奈良科学技术研究所 · 2024年



