In this help article we explain how you can compare the different groups of statement or text entries and see how they are different or similar to each other.
You may be interested to do that when you want to compare the information that appears in different news sources or if you need to see what's the difference between customer reviews for a product that has a high rating vs a product that has a low rating.
Below is a video tutorial that explains the approach in detail. Alternatively, you can also read the step-by-step instructions that follow.
1. Tagging the Statements in the Graph to Filter them Later
Every statement that you add into InfraNodus graph can be tagged manually or automatically. As a result, you get a list of statements in each graph that can belong to the different tag categories.
For example, when you import the top articles from the top 5 newspapers using the InfraNodus RSS app, you will have a list of 80 statements (20 per each newspaper). Each of those statements has a tag, which is the name of the newspaper source it's coming from (e.g. The Guardian or The New York Times). As a result, you can also filter your statements using the menu at the top:
2. Tags Merge — selecting several tags to filter the statements and the part of the graph that belong to those statements
The first approach allows you to select several tags and to filter the graph and the statements to see only those that are tagged with a specific tag.
For instance, if you want to only see the articles from The Guardian and The New York Times, you need to
1) select one of the categories in the Filter menu
2) press and hold the Shift button
3) select the other categories you want to merge it with (holding the Shift button)
4) To deselect, click "All the Statements"
Then you will only see the statements and the graph of the statements that belong to those two categories.
You can also choose multiple categories.
2. Tags Overlap — see the statements and the graph of the statements that belong to several different categories
This option allows you to see the statements and the graph of the statements that belong to several different categories. The graph will highlight the nodes that appear in all the filters you choose.
For example, if you imported the news articles from 5 different sources, you can choose to see what are the topics that both The Guardian and The New York Times are writing about.
To do that, go to the filter menu above the statements and
1) Select the category you want to see first
2) Press and hold the Command (on a Mac) or the Ctrl button (on PC)
3) Choose the category you want to overlay
4) Repeat if necessary, holding the Command / Ctrl button
5) To deselect, click "All the Statements"
You can select more than 2 filters. You will then see the statements that belong to ANY of those selected tags and the full graph for all the filters. However, only the nodes that appear in ALL the different categories of statements you chose will be highlighted on the graph.
If you want to see the full graph (without the highlight), you need to choose the "merge" button as explained in the previous section (2).
4. Tags Differences — what are the differences between the different categories — "text diff"
This last option enables you to see the differences between the different categories. E.g. you can highlight the nodes that exist in, say, The Washington Post articles, but not in The Guardian.
To do that, click the "differences" button in the left menu and then first choose the filter tag category you want to treat as the "main" one (e.g. Washington Post), then, secondly, choose the filter tag category you want to see the differences for (e.g. The Guardian). In that case, you will see highlighted the words / nodes that are present in the Washington Post but NOT in The Guardian articles.
To see the differences between two categories, go to the Statements Filter menu and then
1) Select the first category you want to analyze
2) Press and hold the Alt button
3) Choose the other category that you want to compare with the first one
4) InfraNodus will highlight the nodes that are present in the first category you chose, but not in the second
5) To reverse, click "See All Statements" and repeat the other way round.
5. Comparing Different Graphs to Each Other
Sometimes you might not know directly how to separate your data into categories using tags and instead import the different datasets as separate graphs.
For instance, one graph could contain all the 1- and 2-star reviews of a product, while another graph could contain all the 4- and 5-star reviews.
In that case, you can also the different graphs to each other using the same approaches as above (i.e. — "merge", "overlap", and "diff"). To learn more about this approach, please, see our help article on this subject: How to Compare Text Graphs to Find the Similarities and Differences