AmazonQAC
收藏资源简介:
AmazonQAC是一个大规模的查询自动完成(Query Autocomplete, QAC)数据集,源自真实的亚马逊搜索日志。该数据集包含用户输入的前缀序列和最终搜索词,以及丰富的会话元数据,如时间戳和会话ID。训练集包含3.95亿个样本,测试集包含2万个样本。数据集支持上下文感知的查询完成研究,提供真实、大规模和自然的用户行为数据。数据集的结构包括训练集和测试集,每个数据条目包含查询ID、会话ID、前缀序列、首次输入前缀的时间、最终搜索词、搜索时间以及搜索词的流行度。测试集还包括用户过去的搜索词及其时间戳。数据集的评估指标包括Success@10和Reciprocal Rank@10。所有数据均已匿名化,并移除了个人身份信息(PII)。
AmazonQAC is a large-scale Query Autocomplete (QAC) dataset derived from real Amazon search logs. This dataset contains user-input prefix sequences and final search queries, along with rich session metadata such as timestamps and session IDs. The training set includes 395 million samples, while the test set contains 20,000 samples. It supports research on context-aware query completion by providing real, large-scale and natural user behavior data. The dataset is structured into training and test splits, with each data entry comprising query ID, session ID, prefix sequence, timestamp of the first prefix input, final search query, search time, and the popularity of the search query. The test set additionally includes the user's past search queries and their corresponding timestamps. The evaluation metrics for this dataset include Success@10 and Reciprocal Rank@10. All data has been anonymized, and personally identifiable information (PII) has been removed.




