CNSVRE: A Query Reformulated Search System with Explainable Summarization for Virtual Research Environment

Computational notebook environments have drawn broad attention in data-centric research applications, e.g., virtual research environment, for exploratory data analysis and algorithm prototyping. Vanilla computational notebook search solutions have been proposed but they do not pay much attention to the information needs of scientific researchers. Previous studies either treat computational notebook search as a code search problem or focus on content-based computational notebook search. The queries being considered are neither research-concerning nor diversified whereas researchers' information needs are highly specialized and complex. Moreover, relevance evaluation for computational notebooks is tricky and unreliable since computational notebooks contain fragments of text and code and are usually poorly organized. To solve the above challenges, we propose a computational notebook search system for virtual research environment (VRE), i.e., CNSVRE, with scientific query reformulation and computational notebook summarization. We conduct a user study to demonstrate the effectiveness, efficiency, and satisfaction with the system.