InfraNodus text network analysis tool can be used to conduct research on any topic using Google Scholar. This can be very useful for literature review or for gaining an initial understanding of a certain knowledge domain. The advantage of using InfraNodus is that it combines text analysis, data visualization, and AI to help you gain a general overview of a subject, reveal high-level ideas present within, and also — dive into any specific topic deeper, in order to reveal underlying ideas.
Step 1: Choose a Topic and Open the Google Scholar App in InfraNodus
Choose a topic and open the Google Scholar app in InfraNodus.
This app allows you to choose any search query and then import the top 40 (or more) search results for this query.
Step 2: Visualize the Results
The snippets obtained from Google Scholar usually contain the scientific paper title and a short snippet containing the most relevant part of the paper. The text in those snippets will be represented as a network graph where the concepts are the nodes and their co-occurrences (in the same title or excerpt) are the connections. To learn more about how our algorithm works, check How it Works section of InfraNodus where you can also find a reference to the peer-reviewed paper that you can cite in your research.
Step 3: Hide the Words from the Search Query in the Graph
By default, InfraNodus will hide the words contained your search query from the graph, because most search results tend to be too biased to those concepts (as you can see in the Thematic Diversity index that's low, because the search results in this case — for "deep sleep" – are biased towards "deep" and "sleep".
If these words are not hidden or if you want to hide more words that seem obvious, you can use the "Reveal Underlying Ideas" function in the Analytics Panel > Topics > Most Influential Keywords or simply select those nodes directly on the graph and hide them (read an article on how to do that if it's not clear how it works).
Step 4: Reveal High-Level Ideas using GPT-3 AI
After a few iterations, you will get a "cleaner" graph and you can use the built-in GPT-3 AI to interpret the topical clusters detected and generate high-level ideas that describe them well. To do that, click the "conceptual field" button on the graph or choose Analytics > Reveal High-Level Ideas.
You can use this insight to get a general overview of a research domain (in our case — "deep sleep") and see which other topics and ideas it is comprised of.
Once you find a topic that is interesting for you, you can go on a tangent and explore it deeper using the graph.
Step 5: Write Down Your Ideas
At this point, you have a good overview, so it's a good moment to write down some ideas. You can use the Project Notes / Your Insights panel for that. Simply write down your thoughts or copy the data from the Analytics panel automatically, so you have an overview of the main topics contained inside.
Step 6: Zoom Into Specific Topics
To zoom into a specific topic or concept you find interesting, click on it in the graph and you will see the Google Scholar search results that contain those concepts that you selected. You can then see in which context they were used and what other concepts they relate to (Analytics > Relations):
Step 7: Go on a Tangent — Exploring Beyond the Graph
So far, we followed our cognitive variability framework, which promotes alternation between zooming out and zooming in. This allows you to have a general overview and to also get hold of nuance and specifics.
We now will shift to another mode of research where we switch between intents. So far, we've been operating in the realm of "focus", trying to concentrate on a specific subject ("deep sleep"). Now that we found some interesting subtopics inside this discourse, we can go on a tangent and explore this idea in more detail.
In order to do that, we can select some specific nodes and use the Selected Nodes > Binocular menu in the graph to automatically generate Google search results for the subject. We can then add them to the graph or switch the AI mode and instead query GPT-3 Ai on the subject:
In the example above, we selected the nodes "exercise bedtime health benefits" and clicked the Binocular button to generate Google search results for this query to see if there are actually any benefits of exercise prior to bedtime. As it turns out, there are.
We can then enhance these results by clicking "Edit the AI Query" and choosing another AI module (e.g. GPT-3 assertions or facts) or even using the AI to challenge this combination of concepts.
Step 8: Enhance Your Concept Graph: Import More Data
You may also choose to import more data into your graph. For example, you might want to add more results for "exercise benefits for deep sleep". In order to do that, simply go to +Add / Import then choose Import > Scientific Papers > Google Scholar.
Make sure to import your search results to the same graph as you are researching originally. Your Scholar search results will automatically import with the new search query you use as a tag, so you can then later use it to filter the results in order to see the fresh results you imported and combine them with other filter options (e.g. other search queries, citation numbers, etc).
Step 9: Reveal the Structural Gap in Your Discourse
After you finish exploring the graph and importing data from Google Scholar, you can use the Analytics > Gap Insight panel to reveal the structural gap in this discourse.
A structural gap will usually indicate the topics in this discourse that should be connected but are not and propose to think of an idea or a research question that could bridge them together. This usually leads to innovative ideas in research fields as it combines clusters of ideas that are relevant to the discourse but that are not yet well-connected.
Step 10: Finalize Your Findings
Add more ideas to Project Notes (see Step 5 above). Once you're done, export them to finalize in a text editor or re-import them as a new InfraNodus graph to develop those ideas further.
Try It On Your Topic of Interest
Try this approach for any research topic on www.infranodus.com — if you don't have an account, you can join a free trial for 2 weeks.