The Text Mining add-on allows Google Spreadsheet users to:
- enhance their textual spreadsheets content by automatically extracting named entities (such as places, persons, organizations, concepts, events, etc... ) and linking them to Wikipedia by using our Entity Extraction API;
- detect emotions and sentiment in short texts using our Sentiment Analysis API, to identify whether the expressed opinion is positive, negative, or neutral.
Using this add-on, your account has 1000 units for free. If you need for more units please contact us at firstname.lastname@example.org.
To use the Text Mining add-on you must first add it to your spreadsheet. The following instructions will walk you through the process:
- create a new Google Spreadsheet (or open an existing one);
- from the menu bar choose: Add-ons > Get Add-ons… ;
- find the Text Mining Add-on from the add-ons gallery and select it;
- from the add-on description page, click the "+" in the top right corner to add this add-on to your spreadsheet;
- a dialog should pop up requesting your permission for the add-on. Click "Accept";
- the add-on is now installed. A Text Mining submenu should now appear in the Add-ons menu.
Extract places, people, organizations and concepts from your spreadsheet
Now you are ready to use the power of Dandelion API to enrich your spreadsheet. Select the cells containing the texts you want to enrich, and navigate through Add-ons / Text Mining / Analyze text.
In most cases the default configuration will suffice. Select the option "Entity Extraction", and just click on the 'Analyze text' button to proceed; if you need more fine tuning you can choose different options to match your needs.
You have two different options to better tune the text analysis of your texts:
- "highlight minimum length": you remove those entities having a highlight shorter than this number;
- "social content optimization": you enable special hashtag parsing and the social mention @ to correctly analyze Tweets and Facebook posts.
All the semantic information extracted from your texts will be available to you in a new sheet, by default titled "Analysis": have fun with it!
Extracted entities listed in this new sheet include several columns:
- "Text": to keep control of the the text analysed;
- "Highlight": the words identified in the submitted text;
- "Confidence": the confidence value is a numeric estimation of the quality of the annotation, which ranges between 0.6 and 1. Entities with a confidence value below 0.6 are hidden (0.6 is the default threshold for this add-on);
- "Entity Name": the conceptual entity as it appears in our system;
- "Types": the types associated with the entity, extracted from Wikipedia;
- "Categories": the corresponding Wikipedia categories of every entity;
- "Wikipedia URL": the link to the Wikipedia page, which you can use to pull additional data about the entity, and enrich your content.
Detect sentiment and emotions in your texts
If you select the "Sentiment Analysis" checkbox inside the sidebar, you can detect sentiment and emotions of a text stream, especially useful for reviews and opinions expressed on Social Media
All the results of the Sentiment Analysis performed on your texts will be available to you in a new sheet, which include several columns:
- "Text": the text analyzed;
- "Score": the score for the text emotional polarity (-1.0 totally negative, 0.0 neutral, 1.0 totally positive);
- "Sentiment": the overall sentiment detected. The possible values are negative, neutral or positive.