Passage aggregation retrieval model (PARM) for dense document-to-document retrieval

June 9th, 2021. 12:00 - 1:30 pm

Dense passage retrieval models show great effectiveness gains in first stage retrieval for the web domain. However in the web domain we are in a constrained setting with large amounts of training data and a query-to-passage or a query-to-document retrieval task. For those domains, large amounts of training data are available. Also, prior studies have been limited to query-to-passage or passage-to-passage retrieval tasks. We investigate in this talk dense document-to-document retrieval with limited labelled target data for training, in particular legal case retrieval. For this we introduce dense passage retrieval and propose a paragraph aggregation retrieval model (PARM) for dense document-to-document retrieval.

Location: Zoom

slides