Search papers, labs, and topics across Lattice.
This paper introduces the concept of answer-reconstruction search density (ARSD), a novel metric that quantifies the minimum number of distinct query actions needed to generate a synthesized answer in conversational systems. By distinguishing between query compression and source compression, the authors provide a more nuanced understanding of the efficiency of information retrieval processes. The key finding reveals that traditional retrieval metrics fail to capture the complexity of search work involved in producing conversational answers, highlighting the need for ARSD in evaluating system performance.
Conversational systems may obscure the true search effort behind synthesized answers, with ARSD revealing the hidden complexity of query actions required for effective information retrieval.
Conversational systems can collapse a visible sequence of web queries, result inspections, and source comparisons into a single synthesized answer. Existing retrieval metrics evaluate ranking, effort, or factual support, but they do not quantify the minimum conventional search work represented by a completed answer. We define answer-reconstruction search density (\ARSD): the minimum number of distinct query actions required, under a fixed and dated reconstruction policy, to support a target share of atomic retrievable answer units. A parallel page-density measure separates query compression from source compression.
