Per-year keyword diversity (Pielou's J') along each flow trajectory

Recomputes, from the documents in each sniff_trajectory_braid() trajectory's own group-year nodes, the normalized Shannon entropy (Pielou's evenness J') of their keyword distribution, one value per year of the trajectory's life. A falling series marks thematic convergence, a rising one diversification. Unlike sniff_entropy(), which scores a static sniff_groups() partition by group-year, this follows the trajectory's actual membership, so an absorbed tributary is scored on its own documents rather than its destination group's.

Usage

sniff_trajectory_entropy(flow, keywords)

Arguments

flow

A sniff_trajectory_braid() object.

keywords

A long data frame with columns document_id and keyword, one row per document-keyword pair (e.g. the corpus DE field split on ; and keyed by the membership id).

Value

A tibble, one row per trajectory: traj_id, type, group, and keyword_entropy, a list-column whose each cell is a tibble with columns year and keyword_entropy (Pielou's J' over that node-year's documents, in [0, 1]; 0 for a single keyword, NA when the year's documents carry none).

Details

This is keyword entropy (thematic diversity), distinct from the dynamics dest_entropy (the dispersion of an absorbed cohort across destination groups); the dynamics column is named keyword_entropy to keep the two apart.

See also

sniff_entropy(), sniff_trajectory_cct(), sniff_trajectory_dynamics(), sniff_trajectory_braid()

Other trajectory analysis: sniff_trajectory_cct(), sniff_trajectory_community(), sniff_trajectory_confluence(), sniff_trajectory_contribution(), sniff_trajectory_destination(), sniff_trajectory_dynamics(), sniff_trajectory_emergence_owners(), sniff_trajectory_formation(), sniff_trajectory_hubs(), sniff_trajectory_self_sufficiency()