curvePlot()

Visualize Temporal Evolution of Keyword Frequencies

The curvePlot() function creates multi-line plots showing how keyword frequencies change over time. It displays all keywords in your corpus simultaneously, with colors and line weights indicating frequency zones, providing a comprehensive view of temporal dynamics.

πŸ”Ή Function Definition

curvePlot(
  data,
  keywords = NULL,
  r = 1,
  themety = "light",
  size_class = NULL,
  x_leg = 0.85,
  x_lab = "years"
)

🎯 Purpose

Understanding how terms evolve over time is fundamental to temporal corpus analysis. The curvePlot() function helps you:

  1. Explore temporal patterns β€” See how keywords rise, fall, or remain stable over time
  2. Identify trends β€” Spot emerging, declining, or persistent concepts
  3. Compare trajectories β€” Observe relationships between different keywords
  4. Detect events β€” Find spikes or drops that correspond to historical events
  5. Visualize zones β€” See how different frequency zones behave temporally
  6. Assess data quality β€” Identify anomalies or irregularities in the data
  7. Generate insights β€” Discover unexpected patterns for further investigation
  8. Create publication figures β€” Produce high-quality visualizations for papers

This function can display either all keywords (comprehensive overview) or selected keywords (focused analysis), making it versatile for both exploration and presentation.

βš™οΈ Arguments

Argument Type Default Description
data List required A list object returned by importData() or normalization(), containing the TDM and corpus metadata.
keywords Character vector NULL Specific keywords to plot. If NULL, plots all keywords in the corpus. Use this to focus on specific terms of interest.
r Integer 1 Interval for x-axis label thinning. r = 1 shows all years, r = 2 shows every 2nd year, etc. Useful for long time series.
themety Character "light" Visual theme:
β€’ "light": Light background (default)
β€’ "dark": Dark background (better for presentations)
size_class Numeric vector NULL Custom line thickness for each frequency zone. If NULL, uses theme-appropriate defaults. Vector length must match number of zones.
x_leg Numeric 0.85 Horizontal legend position (0 = left, 1 = right). Adjust if legend overlaps with curves.
x_lab Character "years" Label for x-axis (e.g., "Years", "Time Period", "Publication Year").

πŸ“Š Plot Components

Visual Elements

Lines: - Each keyword is represented by one line - Line color indicates frequency zone - Line thickness can vary by zone importance

X-axis: - Time periods (years) from corpus metadata - Labels can be thinned for long time series

Y-axis: - Frequency values (raw or normalized) - Scale depends on normalization applied

Legend: - Shows frequency zones with corresponding colors - Position adjustable via x_leg parameter

Theme: - Light: White background, dark text (publications) - Dark: Dark background, light text (presentations)

πŸ’‘ Usage Examples

Basic Usage - All Keywords

library(cccc)

# Import data
corpus <- importData("tdm.csv", "corpus_info.csv")

# Plot all keyword trajectories
curvePlot(corpus)

Plot Specific Keywords

# Import and normalize
corpus <- importData("tdm.csv", "corpus_info.csv")
corpus_norm <- normalization(corpus, normty = "nc")

# Plot specific keywords of interest
curvePlot(
  corpus_norm, 
  keywords = c("algorithm", "network", "data", "cloud")
)

Long Time Series

# Show every 5th year on x-axis
curvePlot(corpus_norm, r = 5)

# Show every 10th year
curvePlot(corpus_norm, r = 10, x_lab = "Decade")

πŸ” Interpreting Temporal Patterns

Common Trajectory Types

1. Emerging Terms πŸ“ˆ

Frequency
    |           ╱───
    |         β•±
    |       β•±
    |_____β•±________Time
  • Start low, increase over time
  • Indicate new concepts or growing importance
  • Examples: β€œcloud computing”, β€œdeep learning”

2. Declining Terms πŸ“‰

Frequency
    |───╲
    |     β•²
    |       β•²
    |         β•²___Time
  • Start high, decrease over time
  • Indicate obsolescence or reduced relevance
  • Examples: β€œfloppy disk”, β€œmainframe”

3. Stable Terms ➑️

Frequency
    |─────────────
    |
    |
    |______________Time
  • Consistent frequency over time
  • Indicate core, persistent concepts
  • Examples: β€œcomputer”, β€œdata”, β€œsystem”

4. Peaked Terms πŸ”οΈ

Frequency
    |      β•±β•²
    |    β•±    β•²
    |  β•±        β•²
    |β•±____________β•²_Time
  • Rise and fall within time period
  • May indicate temporary trends or events
  • Examples: β€œY2K”, specific technology buzzwords

5. Cyclical Terms πŸ”„

Frequency
    |  β•±β•²    β•±β•²
    | β•±  β•²  β•±  β•²
    |β•±    β•²β•±    β•²_Time
  • Periodic fluctuations
  • May indicate recurring themes or seasonal patterns
  • Examples: Conference-related terms, annual events

6. Volatile Terms πŸ“Š

Frequency
    | β•±β•²β•±β•²β•±β•²β•±β•²
    |β•±        β•²Time
  • Highly irregular patterns
  • May indicate noise or inconsistent usage
  • Consider smoothing or filtering

🎨 Visual Interpretation

Zone-Based Patterns

High-frequency zones (typically darker colors): - Represent core vocabulary of the corpus - Trajectories often more stable - Changes indicate major conceptual shifts

Low-frequency zones (typically lighter colors): - Represent specialized or emerging terms - More volatile trajectories - May show sharper rises/falls

Density and Overlap

Dense clustering: - Many terms with similar temporal patterns - May indicate coherent topical evolution

Spread trajectories: - Diverse temporal behaviors - Suggests heterogeneous corpus or multiple themes

πŸ“ˆ Use Cases

1. Initial Exploration

Get a first look at temporal dynamics before detailed analysis.

2. Hypothesis Generation

Identify unexpected patterns that merit further investigation.

3. Trend Validation

Verify expected temporal patterns (e.g., β€œinternet” should rise after 1990s).

4. Comparative Analysis

Compare trajectories of related terms (e.g., β€œanalog” vs β€œdigital”).

5. Event Detection

Find spikes or drops corresponding to historical events.

6. Normalization Assessment

Compare raw vs. normalized frequencies to see normalization effects.

7. Publication Figures

Create visualizations for papers, presentations, or reports.

8. Teaching/Communication

Show collaborators or students how terms evolve in your corpus.

πŸ’‘ Tips & Best Practices

  1. Start with all keywords β€” Get overview before focusing
  2. Use normalization β€” Compare normalized trajectories for fair comparison
  3. Focus selectively β€” After overview, plot specific keywords of interest
  4. Match theme to context β€” Light for publications, dark for presentations
  5. Thin x-axis for long series β€” Prevent label overlap (use r parameter)
  6. Save high resolution β€” Use dpi = 300 or higher for publications
  7. Compare before/after normalization β€” Understand normalization effects
  8. Document patterns β€” Keep notes on interesting trajectories for later analysis
  9. Cross-reference with history β€” Match patterns to known events/developments
  10. Use with other plots β€” Combine with rowMassPlot() and colMassPlot() for full picture

πŸ“š See Also