How to understand the Excel File.


Explorer Excel report consists of 3 sheets:

Summary: structured results according to the project's topics and groups. 

Data: detailed sentiment analysis for every text from the original file. 

Metadata: general information about the project and analysis (which filters were applied, which ignore terms were configured etc).


And you can also configure the contents by adding a Target Concept, Keywords and custom concepts before generating your report.


The Data report sheet is row based: each input text is represented on its own row with all associated analysis results data on the same row in different columns. All rows of the original file will be included in the report regardless of how you configure the report.

Here is a list of the columns of the Data sheet:

  1. First come the original columns of the project input file. The data is the same as in the input file.
  2. Next there will be a column for every pinned group, and each of its topics, and every separate pinned topic. The values in the columns will be either 0 (zero) or 1 (one) signifying whether the text of a certain row belongs to a certain group or topic. The column header will look like this: 'Group: room', 'Topic: room'. This means that one verbatim can have multiple Topics and The Explorer will categorize the topics respectively.
  3. Next there will be a column for each of the standard eight sentiments that are configured for each language that the Explorer supports. The headers will look like this: 'SENT: SKEPTICISM', 'SENT: FEAR', 'SENT: VIOLENCE', 'SENT: HATE', 'SENT: NEGATIVITY', 'SENT: LOVE', 'SENT: POSITIVITY', 'SENT: DESIRE'. The values will be the measurements of sentiment of each type for each text. If you decided to use a target concept when you created the report, the sentiment will be restricted to those sentences containing any of the target terms of the chosen concept in each text and you can look at the column for the target concept to tell you which texts matched at all. If you didn't choose a target concept the values will be for the complete texts. Values will range from zero to infinity.
  4. Next you will get a column for each of the custom concepts you have defined in your account and chosen to include in the report. They are measured the same way as the sentiments (taking your optional target concept into consideration); if you chose a target concept, its column will tell you which texts have sentences that match that concept. The headers will look like this: 'Concept: concept_name'.
  5. Next follows four different keywords columns:
    1. Keywords: this column will contain the significant keywords of each text. Keywords are words or multiword expressions that carry significant meaning in the text. The format of this and the other keywords columns is comma separated list like this: '[word, other word, third word]'. The column header is 'Keywords'.
    2. Qualified Keywords: this column will contain expressions from the previous list that have also been qualified by looking them up in Wikipedia. If they exist as pages in Wikipedia they are considered qualified. The column header is 'Qualified Keywords'.
    3. Expanded Keywords: this column contains words that are semantically similar to the words in the first list. This is produced by looking up the words in the Gavagai Lexicon to find related words. The column header is 'Expanded Keywords'.
    4. Qualified Expanded Keywords: this column will contain expanded keywords that have been qualified by Wikipedia in the same way as for qualified keywords. This column is guaranteed to contain expressions that do not appear in the text but that are highly relevant representatives (and widely used such representatives) of the text. The column header is 'Qualified Expanded Keywords'.
  6. Next you will get a column containing a summary of the text. The summary will be a three sentence or less summary of the text. The column header is 'Summary'.