Search research articles

Search research articles

Related Experiment Videos

From paragraph to graph: latent semantic analysis for information visualization.

Thomas K Landauer¹, Darrell Laham, Marcia Derr

¹Department of Psychology, University of Colorado, Boulder, CO 80309-0345, USA. landauer@psych.colorado.edu

Proceedings of the National Academy of Sciences of the United States of America

|March 24, 2004

Summary

This summary is machine-generated.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by

Same journal

In This Issue.

Proceedings of the National Academy of Sciences of the United States of America·2026

Same journal

Long-term cultural continuity across the Neanderthal-modern human sequence at Üçağızlı II Cave, northern Levant.

Proceedings of the National Academy of Sciences of the United States of America·2026

Same journal

Dolphins use names to remember whom to avoid.

Proceedings of the National Academy of Sciences of the United States of America·2026

Same journal

Retraction for Shaked and Frenkel, Curiouser and curiouser: Meningeal lymphoid structures in the aging brain.

Proceedings of the National Academy of Sciences of the United States of America·2026

Same journal

Small but mighty: The outsized role of small water bodies in the global carbon cycle.

Proceedings of the National Academy of Sciences of the United States of America·2026

Same journal

Functional traits produce conditional outcomes in different community contexts.

Proceedings of the National Academy of Sciences of the United States of America·2026

See all related articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Exploring high-dimensional semantic spaces is crucial for understanding text. Latent semantic analysis combined with dynamic visualization offers a powerful method to reveal complex linguistic relationships missed by traditional techniques.

Area of Science:

Information Science
Computational Linguistics
Data Visualization

Background:

Traditional text analysis relies on manual links (keywords, citations).
Semantic similarity of whole documents offers a more comprehensive approach.
Latent semantic analysis (LSA) reduces dimensionality but has limitations.

Purpose of the Study:

To explore high-dimensional semantic spaces for text analysis.
To overcome limitations of low-dimensional LSA visualizations.
To leverage human visual perception for discovering hidden linguistic patterns.

Main Methods:

Utilized latent semantic analysis (LSA) for dimension reduction.
Developed a high-dimensional dynamic viewer with projection pursuit.

Related Experiment Videos

Integrated user control and human visual system capabilities.

Main Results:

LSA correlations with human judgments peak around 300 dimensions.
Low-dimensional LSA (2-3 dimensions) fails to capture all relevant relationships.
The proposed dynamic visualization approach successfully reveals insights missed by computational algorithms.

Conclusions:

Linguistic meaning is inherently high-dimensional.
High-dimensional dynamic visualization is essential for exploring complex semantic spaces.
Combining computational methods with human perception enhances text analysis discovery.