🎥 Case Study: Generate a Mind Map from any Text

You can use InfraNodus to build mindmaps automatically using text input. The advantage of this approach to mindmapping is that you don't have to decide beforehand what the central idea is. Rather, our powerful graph analysis algorithms will detect the most influential clusters and elements for you. You can then have an overview of your ideas and detect the structural gaps: the areas of your mind map with a high potential for new interesting discoveries.

See the video tutorial below to learn how it works or follow the step-by-step process:

Also, see the article on mind mapping with InfraNodus.

Mind Mapping with Text, Step-by-Step Process

Step 1: Create a Graph

Open a new graph in InfraNodus. Use the Live Ideation app or the Mind Mapping app. 

Step 2: Add the Data or Text (do not submit it yet!)

Once you see an empty graph, start typing your text or copy and paste some ideas from an already existing one. Do not yet submit it.

For example:

In supervised machine learning, you feed the features and their corresponding labels into an algorithm in a process called training. During training, the algorithm gradually determines the relationship between features and their corresponding labels.

If you add the texts as it is above, every word will be visualized as a node in the mind map. In case you want to control which entities are actually visualized, you can use [[wiki-links]] and add them as you are writing the text. When you type in "[[", InfraNodus will automatically place your cursor right after and add "]]" in the end, so you can type in the name of your concept. For example:

In [[supervised machine learning]] the algorithm determines the relationship between [[features]] and their corresponding [[labels]]

In this case above you will have 3 nodes created on the graph.

Step 3: Highlight the Nodes using #hashtags or [[wiki-links]] and Submit

If you don't select anything in the text and submit it as it is (do not do it yet), then every word (apart from the stop words) will be represented as a node in the graph and every co-occurrence will be represented as a relation. 

If you select the words that you want to visualize, they will be marked as the hashtags and only those words will be added into the graph. All the words appearing in the same statement will be connected.

For example, if you select a few concepts like this:

In supervised #machine_learning, you feed the #features and their corresponding #labels into an algorithm in a process called #training. During #training, the algorithm gradually determines the #relationship between #features and their corresponding #labels.

You can also use [[wiki-links]] to highlight the concepts in the text. Simply type in "[[" and add your concept, for instance:

In supervised [[machine_learning]], you feed the [[features]] and their corresponding [[labels]] into an algorithm in a process called [[training]]. During [[training]], the algorithm gradually determines the [[relationship]] between [[features]] and their corresponding [[labels]].

You will add 5 distinct nodes and 9 relations between them: 

InfraNodus connects not only the concepts that are next to each other but also the words that have 1 or 2 concepts between to emulate the natural process of reading. The closer are the concepts, the stronger is the connection. You can switch this off in the settings and instead choose to connect only the concepts that are next to each other. 

Notice, how "training" and "labels" have the strongest connection. This is because that connection occurs the highest number of times in the graph. The next one is "training" and "features" and "labels" and features".

As you can see, this already provides a good representation of how the concepts are connected.

Note: For more control over this behavior, you can operate in single triplets of concepts. For example: 
In supervised #machine_learning, you feed the #features and their corresponding #labels. 

Here you will only have 3 nodes (machine_learning, features and labels) and they will all be connected to one another. 

Step 4: Add More Text

Now that we added the first statement, we can add more stuff. For example:

In unsupervised learning, the goal is to identify meaningful patterns in the data.  

Let's now select "unsupervised learning", which will be converted into a concept like #unsupervised_learning, then also #patterns and #data. So we get something like this:

In #unsupervised_learning, the goal is to identify meaningful #patterns in the #data.  

Then submit this statement and you will get something like this:

Step 5: Editing a Mind Map

Suppose now we realize that actually, we want to have a separate node for #machine_learning and separate nodes for #supervised and #unsupervised. 

In that case we can simply

a) click the statement we want to edit, then
b) click the Edit link
c) edit the statement
d) save it back into the graph

The result will look like this:

As you can see, now #machine_learning is the main concepts and it's connecting two distinct clusters: on supervised training (using features and labels) and unsupervised training. 

Step 6: Use Analytics

Once you add more data, you will see some interesting patterns forming in the Analytics panel. It will show you the main topical clusters and the most influential nodes, using the metrics derived from network science (mainly betweenness centrality and community detection algorithms based on graph modularity). 

Step 7: Use Insight Generation Tool

You can also use the Insight Generation Tool to get a recommendation as to what could be a next interesting idea to think of in relation to this topic. This tool is based on identifying the structural gaps in the discourse network and asking the questions that are designed to bridge those gaps and to generate new ideas.