A computer-assisted approach to filtering large numbers of documents for media analyses

Abstract

Media analysts are challenged to acquire selections of documents that are representative of their topics of interest. Conventional search and selection processes are often constrained because of an inability to efficiently filter large amounts of potentially relevant documents and thus pose the risk of introducing bias. We describe a computer-assisted approach to increase the probability of identifying all articles relevant to a topic (in this case, marijuana), and provide an evaluation of its effectiveness in reducing bias while minimizing time expenditure. Using our system, we filtered 23,755 articles in 24.4 h. Relative to conventional processes, a substantial reduction in bias was achieved. Our system significantly reduced the risk of bias while retaining efficiency and accuracy in document selection.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 4 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.