The Search Bar, also known as the Search Filter, allows you to search for granular results.
To use this filter to search for a specific word, click on the Search Bar and start typing.
Depending on the word you enter, a choice of up to three result types, which you can filter by, will appear below the Search Bar.
Topics - Selects results that have been identified as Topics.
Opinions - Selects results that have been identified as Opinions.
Keywords - Similar to a traditional search, this will return all Records containing the specific word you entered.
For example, in a dataset containing reviews for a food deliver app, we typed in "address" and the results display three options: Topics, Opinions and Keywords.
The Topic “address” includes other words under its umbrella, such as “wrong address” and “post code”. Note: users are able to change how results are grouped.
If we select "address" as a Topic as our filter, I will get the results for all the grouped variations. Notice how all Comments have a variation of “address” as a Topic:
If I select "address" as a Keyword, I will get a wider range of results, including Comments that don't even mention the word "address". Here the keyword "address" is mentioned somewhere in the Record itself but not necessarily as a Topic.
Notice how the Topics may refer to other things included in the Comments mentioning "address".
Once you select the type of result you want to see: Topic, Opinion or Keyword click on it. It will show up as a filter right below the Search bar.
It's possible to add as many Search filters as you want while text mining. Below, I added "address" as a Topic and "wrong" as an Opinion.
Notice how the results reduced from 20 Comments to 1 Comment. This is because combining filters of different types (e.g. Topic and Opinion) only returns results where both instances are present (applies “AND logic”). For results of the same type (e.g. Topic and Topic), the results aggregate with all terms (applies “OR logic”).
Finally, to remove the Search filters, click on the "X" located on each filter label.