top of page

Technical Directions

Text-as-Data Analysis Tool - Technical Directions

About the Tool

The Text-as-Data Analysis Tool is a Shiny Web Application developed by Courtney Consulting LLC in 2023. Aimed at making text analysis as straightforward as possible, it provides functionalities like keyword analysis, sentiment analysis, and frequency counts. Running on the shinyapps.io server, this tool leverages the power of the R programming language to turn your text into insightful data.

Preparing your Text Files

  1. Understanding Text Analysis: Get a good grasp of text analysis basics. The tool handles different segments of text, and understanding the types of analysis it can perform will help ensure you get the results you desire.

  2. Gather Text Segments: Before uploading, ensure your text file is well-structured. You should have your segments clearly marked, as this would affect the dropdown selections in the tool.

  3. Data Sensitivity: Make sure that the text you intend to upload does not contain personally identifiable or sensitive information.

 

Using the Text-as-Data Analysis Tool

  1. Choose Your File: Begin the process by selecting your text file via the "Choose a text file" button.

  2. Select Text Segment: Use the dropdown menu that appears to select the segment of the text you'd like to analyze.

  3. Analysis Options: After segment selection, a new set of options will appear. You can now select the type of analysis you want—whether keyword analysis, sentiment analysis, or frequency counts.

  4. Generate Analysis: Click on "Analyze" to start the magic. You’ll be shown graphs and metrics that encapsulate the selected text segment's characteristics.

  5. Reset for New Analysis: If you wish to analyze another segment or another text entirely, click the "Start Over" button to clear all selections and results.

Interpreting Results from the Text-as-Data Analysis Tool

So, you've got your graphs and metrics after running an analysis. Great! But what do these numbers and shapes mean? Let's break it down.

Keyword Analysis

  1. Frequency Chart: This is a bar chart showing the most commonly occurring words in your selected text segment. The higher the bar, the more often the word appears. It gives you a quick glance at the main topics in your text.

  2. Word Cloud: A fun and visual representation of the frequency data. Words that appear more often are shown in larger fonts. It's a great way to identify central themes quickly.

 

Sentiment Analysis

  1. Polarity Score: Ranging from -1 to +1, this score tells you the overall sentiment of the text. A score closer to +1 indicates positive sentiment, while a score closer to -1 suggests negative sentiment.

  2. Subjectivity Score: This measures the amount of personal opinion in the text, ranging from 0 to 1. A higher score means the text contains more opinions as opposed to factual information.

  3. Sentiment Over Time: If your text spans different time periods or segments, this line chart will show you how sentiment changes over time. Look for any spikes or dips—they're points of interest!

 

Frequency Counts

  1. Count Table: This table will list all unique words in the text, along with the number of times each appears. It's a more detailed version of the frequency chart and is sortable.

  2. Bigram and Trigram Tables: These tables list common two-word and three-word phrases in the text. This can help identify common phrases or idioms in the text, giving more context than single words.

 

Multi-Dimensional Scaling (MDS) Plot (if applicable)

  1. MDS Plot: This is a scatter plot that places words that are contextually similar close to each other. It helps in identifying clusters of related terms or concepts in your text.

 

Extra Tips

  1. Cross-reference: Always good to cross-reference these results with your initial objectives. Are the most common keywords what you expected? Does the sentiment align with the tone of the text?

  2. Context Matters: Remember that all these metrics should be understood in the context of your specific text and objectives. A high frequency of a certain word could be significant or irrelevant depending on what you're analyzing.

  3. Multiple Runs: Don't hesitate to run multiple analyses, tweaking your text or segments to compare results. Sometimes you catch new insights on a second or third look.

 

References

bottom of page