Principal Components Analysis on Plain Text - Beta (TAPoRware)
This tool applies Principal Components Analysis rules to a text to generate relationships between words and text units. It works best with large texts where users can specify units of over 500 words. HTML and XML versions are not currently avaiable.
This tool’s functionality has been incorporated into Voyant available online at www.voyant-tools.org.
Documentation
Creator(s): Geoffrey Rockwell et. al.
Creator's Site: http://geoffreyrockwell.com/
Creator's Email: grockwel@ualberta.ca
Repository:
Attributes
- Type of analysis: Uncategorized
- Type of license: Free
- Web Usable: Run in Browser
- Ease of Use: Moderate
- Warning: No longer supported
- Tool Family: TAPoRware
- Historic Tool (developed before 2005): Discontinued (no longer available)
- TaDiRAH Methods: Uncategorized
List this tool appear on
Rating
Suggested Tools
Comments
Principle Components Analysis (Beta) is a free, web-based tool designed to run in a browser window. It is easy to use, designed generate relationships between words and text units in a user-specified plain text document, either hosted at a web address or uploaded from the user's files. The relationships are visualized in a Java applet that places words on an interactive scatter plot.
To display the applet, users must have Java 1.4.2_8 or later installed for their browser. In order to have the applet display in a separate tab or browser window, users must also tell their browser’s pop-up blocker to permit popups from the TAPoR website. The tool can handle novel-length texts, though the tool’s processing will slow proportional to the size of the text. A number of options are available for interacting with the graph, including zoom, rotate, pan and changing the view axis.
Users may specify a number of restrictions to limit the list. The subtext options allow one to break the text into paragraphs, by percentage or into chunks by number of words. The words themselves may be limited by the number of the highest frequency words to appear, and by whether or not to exclude words according to either the Glasgow stop list or a user-specified list.
In addition to the Java applet, the tool’s results can be displayed as an HTML table or as tab delimited text. The tool gives best results for documents over 500 words and a number of text units below 100.
This tool currently has a few minor problems. When identifying words, it tends to include punctuation such as quotation marks, and opening quotes are appended to the word they appear adjacent to. The Java applet is sensitive to the volume of other traffic on the user’s network and its processing may be slowed or it may fail to load when traffic is heavy. If it fails to load, refreshing the page or resubmitting the tool will generally resolve the issue. In addition, while it will accept HTML and XML documents, it strips all tagging and processes it as if it were plain text.
Despite these limitations, Principal Components Analysis (Beta) is an interesting, interactive way to view the distribution of the top words in a document.