1. Overview
You can export your results by manually generating an output file. Click on the downward pointing arrow icon at any dataset, choose ‘Export’, and then select the file type you would like to generate from the dropdown menu.
You can choose a Results or a Semantic Excel, a .PDF, an .SPSS or a .CSV file.
Your new export will appear in the list of exports for that particular dataset, and once the generation is complete (its status changes from blue to green), you can download it by clicking on the downward pointing arrow next to it. The previously generated exports are also available in this popup, so you don’t need to generate a new one every time.
2. Semantic Excel
The Semantic Excel file is a very detailed and rich output of your analysis results in a structured format. It focuses on the insights of text analysis and has an emphasis on recognized topics, entities, custom labels and sentiment analysis.
Our goal with it is to provide an output which offers the same insights and richness as our dashboard, but in a static file format, enabling you to carry out further data manipulation and calculation. The Excel document contains nine tabs, where the data is structured to serve different use cases.
- The Summary tab shows the most frequent labels, their polarity, and some basic statistics. For each label, you’ll find its summarized Opinion Index, and the distribution of the verbatims according to polarity.
- The Texts tab lists each analyzed verbatim, along with its statistics, such as its polarity score, the recognized positive and negative phrases, along with the recognized topics and entities.
- The Recognized labels tab lists the 100 most frequent labels recognized in the data, with its frequency and opinion index.
- The Text-label matrix tab helps you filter verbatims, based on the recognized labels or Custom Labels listed for each of them. Keep in mind that the values in the entity columns represent not only the occurrence of the entity of the specific label but the assigned sentiment score as well.
- The Associations tab shows the most important relations between recognized labels, and how many times they were mentioned together.
- The Custom Labels tab lists the frequency and Opinion Index statistics of your Custom Labels.
- The Custom Label coding tab helps you discover more about your custom labels by listing every recognized case. This is useful for market researchers, because in each row, there is only one verbatim with one Custom Label. If a verbatim has multiple Custom Labels, the verbatim is listed in multiple rows too, each time with a different Custom Label.
- The Custom Label priority tab lists each verbatim and their recognized Custom Labels in the order of recognition for that verbatim.
- The opinion correlations tab is a matrix showing the strength of correlations between topics and their effect on the overall opinion index. This correlation is not only calculated for the opinion index, but all data with the Number dimension type, as well as NPS®, CES and CSAT.
3. Results Excel, SPSS, PDF
The Results Excel file lists all of your verbatims and all related metadata, for example date or NPS®. Unlike the Semantic Excel export, this export contains only the most important text analysis results, like Opinion Index per verbatim and the recognized custom labels.
The output for IBM SPSS Statistics will be a .sav file. When you open it, you can see that there are three tabs: the overview, the data view and the variable view.
The overview shows basic statistics of your output, such as the number of variables and cases (records), the measurement level of your variables, and distribution charts of each variable’s results.
The data view lists every single record in your dataset by variables, which correspond to dimensions (e.g. Excel columns, survey questions) in the original dataset.
The variable view lists your variables and their characteristics. Each dataset dimension has its own variable, or multiple variables in case of verbatims and complex survey questions.
Variable ‘Name’ corresponds to the header of the data view. It is prefixed meta_ for metadata dimensions and q_ for all others. All verbatim dimensions have an additional variable for opinion index postfixed with _oi and as many additional variables as many custom labels the dimension has. Complex survey questions such as matrix questions also have multiple variables, one for each matrix row. Variables belonging to the same dimensions share the same number after q_.
Variable ‘Label’ is the dimension name (e.g. question title in case of survey) by default, but for verbatim custom label variables, it’s the custom label name.
For category dimensions, ‘Values’ contain the category codes and their corresponding value, e.g. in matrix cases you will see 1 - not checked, 2 - checked.
Make sure to check the ‘Measure’ column and choose the right type for each variable if you want to work with the data in SPSS, because by default, they’re all set to ‘Nominal’.
The exported PDF file is a basic overview, it essentially contains charts from the Results dashboard. Therefore, its content depends on what kind of data you have available in that dataset. However, if you need the charts as separate files, you can export them in .png format individually. This is true for text analysis charts as well, which will not be shown in a PDF export.