Towards Query Logs for Privacy Studies: On Deriving Search Queries from Questions

Detailed query histories often contain a precise picture of a person’s life, including sensitive and personally identifiable information. As sanitization of such logs is an unsolved research problem, commercial Web search engines that possess large datasets of this kind at their disposal refrain from disseminating them to the wider research community. Ironically, studies examining privacy in search often require detailed search logs with user profiles. This paper builds on an observation that information needs are also expressed in the form of questions in online Community Question Answering (CQA) communities. We take a step towards understanding the process of formulating queries from questions to form a basis for automatic derivation of search logs from CQA forums. Specifically, we sample natural language (NL) questions spanning diverse themes from the StackExchange platform, and conduct a large-scale conversion experiment where crowdworkers submit search queries they would use when looking for equivalent information. We also release a dataset of 7,000 question-query pairs from our study.