Hi everyone,<\/p>\n
I\u2019m currently conducting a research project comparing traditional search engines (e.g., Google) and LLM-based conversational search tools (e.g., ChatGPT, Perplexity.ai) in the context of consumer search behaviour \u2014 specifically, how users search for and choose products like smartphones when factors such as price and features moderate their decisions. I intend to conduct a controlled experiment to collect search behavior of approximately. 100 participants providing causal evidence, but still want to validate those insights using external datasets or benchmarks.<\/p>\n
I\u2019m looking for publicly available datasets that capture one or more of the following aspects:<\/p>\n
- \n
- User\u00b4s background, including age, gender, education, employment, nationality, residence, prior knowledge of AI tools, and shopping-related tools.<\/li>\n
- Search behavior logs (queries, clicks, scrolls, or multi-turn interactions).<\/li>\n
- Conversational or query reformulation datasets \u2192 datasets where users ask follow-up questions or clarify queries.<\/li>\n
- Consumer choice or e-commerce data (based on price or features).<\/li>\n
- User attitude or satisfaction survey data (e.g., perceived trust, relevance, ease of use, usefulness, overload, decision confidence, and handling contradictory information).<\/li>\n<\/ul>\n
Also open to:<\/p>\n
- \n
- Suggestions for benchmark datasets used in Conversational Search<\/em> or Retrieval-Augmented Generation (RAG)<\/em> evaluations<\/li>\n
- References to recent arXiv or TREC publications releasing such data<\/li>\n<\/ul>\n
If anyone here knows of datasets that bridge search interactions \u2014 or newer LLM-integrated conversational search datasets \u2014 I\u2019d really appreciate your input. Thanks in advance!<\/p>\n<\/div>\n
submitted by \/u\/Dismal_Priority_2381 <\/a>
[link]<\/a><\/span> [comments]<\/a><\/span><\/p> - References to recent arXiv or TREC publications releasing such data<\/li>\n<\/ul>\n
- Suggestions for benchmark datasets used in Conversational Search<\/em> or Retrieval-Augmented Generation (RAG)<\/em> evaluations<\/li>\n
