Skip to Main Content

Text Analysis

Dartmouth College Library's guide to text analysis tools and platforms

Build and Analyze Datasets

Constellate provides how-to guides, reference materials, and tutorials to help you get started.  Follow the steps below from Constellate's user quickstart guide.

Step 1:  Identify a topic that interests you and see what content is available for analysis right from the search box

When you click search, you are going to come straight into the Dataset Builder with your initial keywords filtered.  

    Hint: Building a coherent dataset is key to any text analysis. If you are just exploring, we recommend building a very tight, subject specific dataset. You can do this with some robust keyword filtering, but you might also want to focus on one or two titles. The publications in Reveal Digital are often very narrow and using one of those publications as a source can provide interesting and informative results.

Step 2: Play with the filters on the left side of the Dataset Builder to narrow down your topic of interest. As you change them, your visualizations on the right side will change.  (You may also read a detailed description of these filters.)

Step 3: You may download any of those visualizations as images or download the data behind them, by selecting the three vertical dots on the top right of each.  You can also create a shareable URL directly to the visualization from that menu.  You may read more about the ins and outs of our visualizations.

Step 4: When you are happy with your configurations and want to save your dataset, go up to the top and select “Build”.  This is going to save your search and give it a unique identifier, so you can get back to it again easily and share it with others.

Step 5: The page changed!  You are now looking at your Dataset Dashboard.  This is where all the datasets you have created will live.  If you ever want to build a new dataset, click “Build a new dataset” up at the top of the page.

Step 6 (Optional): This would be a good time to log in.  You have a couple of options under that Login button at the top right:

    Work with bigger datasets: If you look at the very top right of your webpage, does it say “Log in through your institution”?  If so, click the Log in button and choose “Find your Institution to increase access”.  You can then use the JSTOR institution finder to connect to your institution’s network.  If your institution participates in Constellate, you’ll now be able to work with datasets of up to 50,000 items.

    Save your dashboard across devices: By default, information about your datasets is being stored in a cookie in your browser.  If you create a user account and log-in, your dashboard and dataset information will be saved to the Constellate database and you’ll have access to the same dashboard from any device (as a bonus, you can use that same account to do some extra stuff over on JSTOR, too.)

Step 7: You can stop at the immediate visualizations of your dataset or you can dig further.  From your Dashboard, you may click on the dataset ‘name’ to go back to its configuration and visualizations in the Dataset Builder.  You could also choose to do deeper analysis by selecting the Analyze button or download the data by way of the Download button.

Step 8: If you want to further analyze your dataset, you can work with it in the Constellate Analytics Lab with Jupyter Notebooks.  If you select the Analyze button, you will be presented with some analysis options.  We recommend new users select the “Tutorial” options.  If you are new to coding and/or Jupyter Notebooks, the most important thing to know to get started is that you want to click on that “Run” button at the very top (the one that looks like a play button), and just run each cell.  You will get results!  We recommend you read the documentation as you go, but it isn’t necessary.  

Step 9 and the final step in this Quickstart: If the Analytics Lab and Jupyter Notebooks are too big a jump for you, click on the “Tutorials” link in the top menu above and work through the lessons (must be connected via Dartmouth's vpn).