Tutorial for Coconut Libtool: A Web-based Textual Analysis Guide

This tutorial is ideal for researchers, librarians, and students looking to quickly analyze bibliometric data or thematic patterns in textual resources, facilitating tasks such as literature review analysis, bibliometric studies, and thematic exploration.

Step 1: Access Coconut Libtool Website

To begin, open the Coconut Libtool website (you can either click the hyperlink “Coconut Libtool website” or copy the URL directly into your browser: https://www.coconut-libtool.com/). Coconut Libtool significantly aids digital humanities researchers in textual data exploration, enabling quick extraction of thematic insights, uncovering keyword trends, and visualizing complex relationships within texts. This tool is entirely web-based, requires no coding skills or installation, and is completely free, making it highly accessible for analyzing textual data.

I selected Coconut Libtool due to its intuitive interface, comprehensive text mining features, and visual interactivity, making it ideal for researchers and students in digital humanities who wish to analyze textual data without deep technical expertise. This tutorial is perfect for researchers, librarians, and students conducting bibliometric studies, literature review analyses, or thematic exploration of large text datasets.

Step 2: Choose a File for Analysis

Next, click “Sample Files” to view recommended data formats and examples provided by Coconut Libtool. For this tutorial, we’ll select the Web of Science sample file (webofscience.txt) as our dataset for the following analysis steps. This ensures we’re using a suitable and properly formatted file.

Step 3: Check Your File for Compatibility

Click “Check File” to verify that your chosen file (supported formats: .txt or .csv) is compatible with Coconut Libtool’s analytical methods. Our selected sample file, webofscience.txt, successfully passed the compatibility check, indicating it supports all available analytical options: Keyword Stem, Sunburst, Topic Modeling, Burst Detection, Bidirected Network, and Scattertext.

Step 4: Start Analyzing Your Text

After confirming that your file is compatible, click “Start Analyzing”. You will see a menu displaying all available analytical methods and a “How to” section providing clear explanations and instructions for each method. For example, selecting “Topic Modeling” will give you detailed guidance on how to apply this approach to explore themes within your textual data effectively.

Step 5: Analyze Text Using Keyword Stem

Let’s start analyzing the text with Keyword Stem. Click “Go to Keywords Stem”, upload your file (webofscience.txt), select your method (e.g., Lemmatization), and choose the column (e.g., Author Keywords). After processing, results will appear clearly structured, and you can download them by clicking “Press to download result”. This approach efficiently identifies fundamental root forms of keywords, enhancing semantic clarity. It’s particularly useful for bibliometric studies (e.g., using VOSviewer or Bibliometrix) because it simplifies keyword management and improves analysis accuracy, especially with large keyword sets. For other texts, simply rename the column to ‘Keyword’ in your file.

Step 6: Topic Modeling Analysis

Next, let’s explore Topic Modeling. After uploading your file (webofscience.txt), select a method: pyLDA (classic, general texts), Biterm (short texts like tweets), or BERTopic (advanced, semantic-rich texts). Here, we chose pyLDA, set topics to 5, selected the column (Abstract), and removed unnecessary words. After submitting, you’ll receive interactive visualizations like the Intertopic Distance Map and keyword lists per topic, clearly identifying themes like academic libraries and professional practices.

Topic modeling helps librarians rapidly organize and categorize large textual collections—such as archives, emails, or reference inquiries—by automatically identifying main themes. This improves resource discovery, user-service customization, and decision-making about library resources.

Step 7: Bidirected Network Analysis

Next, we’ll explore the Bidirected Network tool. After uploading the file (webofscience.txt), select a method (such as Lemmatization), choose the column (e.g., Author Keywords), and set parameters like support and confidence.

Upon analysis, you’ll first obtain a downloadable table showing keyword relationships in terms of antecedents (preceding keywords), consequents (subsequent keywords), support (frequency of keyword pairs appearing together), and confidence (probability of one keyword appearing given another). This table allows you to identify strong keyword associations within your text clearly.

You can further visualize these keyword relationships through a network visualization, clearly showing interconnectedness between keywords and highlighting central themes and their associations within your textual dataset.

Overall, Bidirected Network analysis helps identify meaningful two-way keyword relationships, improving your understanding of thematic structures in textual data.

Step 8: Sunburst Visualization

Now let’s explore the Sunburst Visualization tool. After uploading your file(webofscience.txt), you’ll instantly generate an intuitive, hierarchical visual representation of your data over time. The resulting interactive Sunburst chart clearly shows relationships across different hierarchical levels—such as publication type, journal name, and publication years. This intuitive visualization allows you and users to easily explore complex relationships, trends, and structures, significantly enhancing data comprehension, information retrieval, and supporting informed decision-making.

Step 9: Burst Detection

Next, let’s explore the Burst Detection feature. After uploading your file(webofscience.txt), choose parameters such as the number of top words (e.g., 10), visualization type (line graph), calculation method (running total), and column (e.g., Abstract).

Upon submission, the tool identifies terms with significant frequency spikes (“bursts”) over time, generating individual line graphs for each detected term. Each graph displays the frequency of a keyword over years, highlighting periods (bursts) where usage significantly increased, alongside their respective “weight,” indicating the intensity of each burst.

Overall, Burst Detection is valuable for identifying emerging trends and sudden increases in attention or interest within textual datasets, beneficial for understanding changes and evolutions in research topics.

Step 10: Scattertext Visualization

Finally, let’s explore Scattertext, an open-source visualization tool that highlights linguistic differences between two categories of text. After uploading your file(webofscience.txt), select the column to analyze (e.g., Abstract), and define two labels manually to categorize your documents into distinct groups for comparison.

Once processed, Scattertext generates an interactive scatterplot visualization, with each axis representing the frequency rankings of terms within each document category. Words located further from the origin reflect stronger associations with their respective categories. The visualization clearly demonstrates the linguistic differences and key terms uniquely associated with each category, providing intuitive insights into the thematic contrasts in your dataset. You can export the resulting plot as an SVG file for further use or presentations.

Further Resources for Exploring Coconut Libtool

Tutorial Video
Coconut Libtool Tutorial (YouTube)

Reading
Official Recommended Reading for Coconut Libtool