Sprinklr/Articles_Search

Articles Search数据集是为了在商业环境中测试检索和答案生成系统而设计的。它包括自然语言查询、相应的答案以及以多种格式（清洁和噪声）呈现的知识库条目链接。该数据集是使用Gemini-2.0-flash合成生成的，生成过程中精心挑选了重要的品牌实体。数据集适用于评估商业相关问题的检索系统以及结合查询理解、知识库检索和答案生成的管道。数据集内容局限于商业相关话题，可能不适用于其他领域。

The Articles Search dataset is designed for testing retrieval and answer generation systems in a business context. It includes natural language queries, corresponding answers, and links to knowledge base entries in multiple formats (clean and noisy). The dataset was synthetically generated using Gemini-2.0-flash, carefully selecting important Brand entities during the generation process. It is suitable for evaluating retrieval systems for business-related questions and pipelines that combine query understanding, KB retrieval, and answer generation. The content of the dataset is limited to business-related topics and may not be generalizable to other domains.