Analyze Student Engagement from WebVTT Transcripts • engager

engager

Note: This README.md is automatically generated from README.Rmd. After making changes to README.Rmd, run devtools::build_readme() to update the README.md.

engager analyzes participation in Zoom WebVTT transcripts and produces privacy-supporting metrics, summaries, plots, and export files for instructor review.

Documentation

Installed help: run help(package = "engager")
Package vignettes: run browseVignettes(package = "engager")
Issue tracker and contribution discussion

Quick Start

Installation

devtools::install_github("revgizmo/engager")

One-function beginner workflow

library(engager)

transcript_file <- system.file(
  "extdata/test_transcripts/intro_statistics_week1.vtt",
  package = "engager"
)
results <- basic_transcript_analysis(
  transcript_file,
  output_dir = tempfile("engager-basic-")
)
head(results$analysis)
print(results$plots)

Composable workflow

library(engager)

transcript_file <- system.file(
  "extdata/test_transcripts/intro_statistics_week1.vtt",
  package = "engager"
)

# Load and process once
transcript <- load_zoom_transcript(transcript_file)
processed <- process_zoom_transcript(transcript_df = transcript)

# Summarize the processed data
metrics <- summarize_transcript_metrics(
  transcript_df = processed,
  names_exclude = c("dead_air")
)

# Plot and export with privacy-supporting defaults
plot <- plot_users(metrics, metric = "n", facet_by = "none", mask_by = "name")
print(plot)
invisible(write_metrics(
  metrics,
  what = "engagement",
  path = tempfile(fileext = ".csv")
))

Vignettes

For detailed workflows and examples, see the package vignettes:

Getting Started: see vignette("getting-started", package = "engager")
Plotting and Analysis: see vignette("plotting", package = "engager")
Essential Functions: see vignette("essential-functions", package = "engager")
Privacy, Ethics, and Institutional Review: see vignette("privacy-ethics-review", package = "engager")

What the Package Does

The engager package provides tools for:

Loading and Processing Zoom Transcripts: Convert Zoom .transcript.vtt files into analyzable data
Calculating Engagement Metrics: Measure participation by speaker/student
Name Matching and Cleaning: Match transcript names to student rosters
Visualization: Create plots to analyze participation patterns
Exporting: Write privacy-supporting participation metrics and summaries

Note: The package specifically processes .transcript.vtt files (the canonical Zoom transcript files). Other Zoom file types like .cc.vtt (closed captions) and .newChat.txt (chat logs) are not currently supported but may be added in future versions.

Key Functions

Core Processing

load_zoom_transcript() - Load raw Zoom transcript files (.transcript.vtt)
process_zoom_transcript() - Process and consolidate transcript data
summarize_transcript_metrics() - Calculate engagement metrics
summarize_transcript_files() - Batch process multiple transcripts

Data Management

load_roster() - Load student enrollment data
detect_unmatched_names() - Identify transcript names that need review
match_names_workflow() - Apply the supported exact name-matching workflow

Name Matching (Exact MVP)

Use match_names_workflow() for exact hash-based matching:

library(engager)
roster <- tibble::tibble(
  preferred_name = c("Alice Smith", "Bob Jones"),
  student_id = c("S1", "S2"),
  aliases = c("A Smith; Alice S", NA_character_)
)
transcripts <- tibble::tibble(
  speaker = c("alice smith", "carol"),
  timestamp = as.POSIXct(c("2025-01-01 10:00:00", "2025-01-01 10:01:00"), tz = "UTC")
)
res <- match_names_workflow(transcripts, roster, options = list(match_strategy = "exact"))
res

Analysis and Visualization

plot_users() - Unified plotting with privacy-aware options
summarize_transcript_files() - Generate metrics for multiple transcript files

Diagnostics

Most functions are quiet by default to keep examples/tests clean. You can enable optional diagnostics:

# Enable package-wide diagnostics
options(engager.verbose = TRUE)

# Run an exported workflow with diagnostics enabled
summarize_transcript_metrics(transcript_file_path = transcript_file)

# Turn diagnostics back off
options(engager.verbose = FALSE)

Typical Workflow

Setup: Configure analysis parameters
Load Transcripts: Import and process Zoom transcript files
Load Roster: Import student enrollment data
Clean Names: Match transcript names to student records
Analyze: Calculate metrics and create visualizations
Export: Write privacy-supporting metrics and summaries; save returned plot objects explicitly

Version 0.1.0 provides per-session metrics, summaries, and plot objects. It does not generate a polished course-level engagement report or support longitudinal individual-student reporting.

Privacy Defaults

This package uses privacy-supporting defaults. On load, it sets the session option engager.privacy_level to "mask" (unless you set it yourself). Supported summaries, plots, and writers use this option to mask recognized structured identifier columns with labels such as Student 01. Masking does not redact names or other identifiers embedded in free transcript text, and it is not a legal or institutional compliance determination.

To change the option temporarily for an authorized workflow, save and restore its prior value:

library(engager)
old_privacy_level <- getOption("engager.privacy_level", "mask")
options(engager.privacy_level = "none")
# ... analysis that may include identifiable outputs ...
options(engager.privacy_level = old_privacy_level)

Common structured fields considered for masking include preferred_name, name, first_last, name_raw, student_id, and email. Inspect all outputs before storing or sharing them.

See vignette("privacy-ethics-review", package = "engager") for privacy, ethics, and institutional review considerations.

Contributing

We welcome contributions. Start with the issue tracker.

License

This package is licensed under the MIT License. See the CRAN license stub in LICENSE; the standard MIT terms accompany the distributed package.

engager