Topic Modeling and Sentiment Analysis of Customer Reviews, Voice of Customer

InfraNodus will reveal the main topics and sentiment in customer reviews, discover the structural gaps and use GPT AI to generate new product ideas.

Watch an Introduction


Text mining and topic modeling using network analysis

InfraNodus Features

You can perform sentiment analysis of product review data or survey responses, revealing the main topics in positive / negative reviews. You can also identify the structural gaps in the customers' discourse and use the built-in AI to generate new product ideas.

Survey Responses and Product Reviews

You can upload your survey responses and product / customer reviews as well use the Amazon import.

Sentiment Analysis of Customer Reviews

Use the built-in BERT AI model to identify the sentiment for each customer review (positive, negative, neutral).

Reveal the Main Topics and their Relations

Reveal the main topics in the positive / negative reviews and the context where they appear using the network graph.

Use the GPT AI to Generate New Ideas

Detect the structural gap in the customers' discourse and use the built-in AI to generate new product ideas.


Study the markets in multiple countries and languages: English, German, French, Spanish, and Italian.

Fully Shareable

You can download hi-res copies of the graphs to add to a presentation. You can also export a URL or an embed code to add it to your website.

Market Comparison

You can compare the different graphs together to see how they overlap or how they differ. Use the filters to reveal the missing parts.

Advanced Analytics

Our text network analysis algorithms will reveal the main topics, the most influential terms, and latent ideas in any discourse.

Review & Survey Responses Analysis Workflow:


Import the customer reviews, survey responses, VoC data, and detect the clusters of main topics inside. Unlike other tools, InfraNodus shows you the results in context (e.g. "battery" connected to "life").

Topic modeling of customer review data

Use the advanced BERT AI model to reveal the sentiment in your responses / reviews. Filter the results by sentiment to see what the negative reviewers are talking about to improve your products or challenge your competitors.

Sentiment analysis of customer reviews

Use the integractive graph to find the parts of the customers' discourse that seems interesting to you. Zoom in and find the reviews that contain a certain term.

Get the relevant excerpts using the graph

Identify the structural gap (the ideas that are not yet connected) and use the AI to generate ideas that would fulfil this gap.

Identify the structural gap - what ideas are not connected yet

Sentiment Analysis Workflow

Step by Step Customer Review Analysis Workflow:

Check out this help article from our support center, which demonstrates a workflow for performing sentiment analysis of customer product review data:

Sentiment Analysis


Try it With the Real Graphs:

The graph below is a visualization of Tesla customer reviews from 2018, filtered by the low rating, so you can see what the customers who give the car a lower rating don't like about it:

  Tesla Reviews (2018, 1 & 2 stars)

The advantage of InfraNodus is that it doesn't just show you the keywords, but also the interactive relations between them, so you can click on those terms and better understand the context of the customers' sentiment or of a market discourse.

You can also import and analyze customer product reviews from any Amazon site, so you can study your competition or simply use it to make a more informed decision.

Sign Up to Try with Your Data

How Text Network Analysis Works

Network analysis and graph visualization help you cope with the informational overload and make sense of data.

Here is how it works:


1. Add Any Text or Data

You can use this approach for your research, creative writing, marketing feedback study, SEO purposes, news analysis, etc.

Add any text (speech-to-text available), notes, or import your Gephi graph, Evernote data, Twitter feed, or Google search results snippets.

The text is converted into a graph where the words are the nodes and the co-occurrences are the connections between them, so you can see all your disjointed bits and pieces of data at once, as a big picture.

More in our Whitepaper
Create a Text Network Graph

2. Get a Visual Overview

The network will then be generated from the text or data you added. The most influential words in the graph (nodes with the highest betweenness centrality) are shown bigger, while the words (nodes) that occur more often together are aligned into topical clusters and have distinct colors.

The graph will show you not only the main topics and the most influential terms, but also the relations between them, making it a much better visual summary tool than the traditional tagclouds.

Overview Data with Text Network Visualization

3. Generate Insight

InfraNodus will identify the structural gaps in the network: parts of the graph that could be connected but are not.

It will propose you the different ways how you can bridge these gaps to produce insight and generate new ideas.

You can also export your data as a CSV file (with the InfraNodus insights) for further analysis or to use as a training data for your neural network applications.

Generating Insight with Structural Gaps

Our Methodology in Detail

To learn more about the algorithms used in InfraNodus and for citations, please, refer to this peer-reviewed paper:

Paranyushkin, D (2019). InfraNodus: Generating Insight Using Text Network Analysis, Proceedings of WWW'19 The Web Conference, (ACM library, PDF).

If you want to know more about the algorithms used, here is the first paper we published on the subject:

Paranyushkin, D (2011). Identifying the pathways for meaning circulation using text network analysis, Nodus Labs. (Google scholar)

Become a Supporter!

InfraNodus is open-source software and is developed without external funds or investors. We develop and maintain this tool and the associated research based on your contributions only. We believe it is a working model for crowdfunded software made by users for users.

You can support our work if you become a subscriber and get access to the cloud version or if you become our patron on Patreon. Enterpise, government, and educational versions are also available.

Pricing Options

You can sign up for an account below or request a managed dedicated enterprise solution. We offer a 14-day free trial and you can cancel at any time.

All plans include CSV spreadsheet upload and Amazon reviews import function. The Pro and Premium accounts have extended GPT AI quotas and higher file upload limits.

About InfraNodus: Our Philosophy

One of the most important skills for any business is the ability to listen. InfraNodus is especially designed to provide a deeper insight into a public discourse about your company, revealing high-level ideas as well as hidden insights.

Gregory Bateson coined a beautiful term: "ecology of the mind". What is a mind that is ecological? It has the ability to have an overview, but it can also zoom into any idea. It embraces diversity, but it can also obsess over one thing when needed. It can discover the obvious, but it can also reveal the things that are hidden and ponder the gaps that have not yet been bridged. Rational and poetic.

InfraNodus is a tool that is developed to help you think and listen in this way. It is made to promote ecological dynamics and diversity on the cognitive level. InfraNodus visualises and analyses ideas as a network, revealing the relations and patterns within them, so you can understand the dynamic complexity of how knowledge evolves and explore the nuances of meaning.

How does InfraNodus achieve that? It uses a combination of network theory and social science. In fact, InfraNodus' special sauce is to apply social network analysis techniques to discourse analysis. Once you start seeing the words as people and their co-occurrences as the connections, you can build social networks of concepts, in order to identify the most influential ideas. Then you can target those influential concepts and ideas to connect your own discourse to that of a market.

This can be especially useful for studying the market discourse and customer reviews. With InfraNodus, you can understand the current informational supply in the market, discover what people are actually searching for, analyze your competitors' websites, and better understand customer sentiment. All these insights can be augmented by built-in AI. As a result, you don't only get an in-depth understanding of any market, but can also spot the strong and the weak sides of your competition, and reveal the latest market trends.

InfraNodus can also be particularly helpful for removing cognitive bias and helping you see the nuances nobody else does. Most of the time, people think that if they know the main keywords they get an idea, but this idea is usually superficial. InfraNodus, on the other hand, allows you to get rid of the first, obvious layer of ideas and get to the depths of a discourse, revealing both high-level ideas and the insights hidden within or at the periphery of the market discourse.

It also uses multiple tools from social sciences to help you identify the most effective entrance points into any discourse, which can be particularly helpful if you want to be heard in a crowded space. The insights obtained using InfraNodus can be really helpful for bringing innovative products to the market, the sort of things that are hard to promote using conventional tools.

If you would like to learn more about how InfraNodus works, you can sign up for an account and give it a try or learn more about our approach in the About InfraNodus section where we showcase some of the cutting-edge algorithms and frameworks we use. You may also be interested to find out more about the competitive intelligence approach.

Dmitry Paranyushkin, the founder




InfraNodus is used by researchers, writers, marketing professionals and corporations. Here's what some of our supporters have to say:

Technical Support

Check out our support pages, so you can see how to create the new network graphs, how to read and interpret them, how to discover the hidden patterns in your data and share the results of your research.

1. How to Create a New Graph

The most basic way to create a new graph is to use the hashtags. Each hashtag is a node, their co-occurrence is the connection between them. You can also use normal text to build your text graphs automatically. More on Creating the Graphs

2. How to Read and Interpret Network Visualization

Studies have shown that diagrams and especially network representation of data makes you think more of connections in your data. You can identify the main elements as well as the relations between them and — most importantly — discover the gaps in your knowledge using the graphs. More on Reading the Graphs

3. Find the Right Excerpt in Text

Network graphs are great for non-linear reading. Look at the graph, find the part you like, click on the nodes, and you'll see the excerpt of your original data that contains this exact combination. More on Nonlinear Reading

4. How to Import Text and PDF files

You can import and visualize your text files or PDF documents. In order to do that, just upload the file and InfraNodus will do the rest. More on Visualizing PDFs