2016/06/24 by Christina Lioma, Birger Larsen, Lioma, Christina +5 · 3 citations
Computer Science · #Advanced Text Analysis Techniques #Artificial intelligence #Computer science #Data science #Information Retrieval and Search Behavior #Information retrieval #Knowledge management #Mobile Crowdsensing and Crowdsourcing #Political science #Relevance (law) #Topic Modeling #cs.IR
paper · pdf · doi:10.48550/arxiv.1606.07660
published in arXiv (Cornell University) (Cornell University) · Neu-IR '16 SIGIR Workshop on Neural Information Retrieval, July 21, 2016, Pisa, Italy
openalex publication_date 2016/06/24 · arxiv created 2016/06/27 · arxiv updated 2016/06/28 · openalex created_date 2021/08/02 · openalex updated_date 2026/07/28
What if Information Retrieval (IR) systems did not just retrieve relevant\ninformation that is stored in their indices, but could also "understand" it and\nsynthesise it into a single document? We present a preliminary study that makes\na first step towards answering this question. Given a query, we train a\nRecurrent Neural Network (RNN) on existing relevant information to that query.\nWe then use the RNN to "deep learn" a single, synthetic, and we assume,\nrelevant document for that query. We design a crowdsourcing experiment to\nassess how relevant the "deep learned" document is, compared to existing\nrelevant documents. Users are shown a query and four wordclouds (of three\nexisting relevant documents and our deep learned synthetic document). The\nsynthetic document is ranked on average most relevant of all.\n