skip to content

Now Introducing Data Visualization and a Predictive Coding System Upgrade

by Everlaw


We at Everlaw are proud to announce three major updates to our ediscovery platform, and just in time for our trip to Legalweek New York.

Data Visualizer

Everlaw’s latest update includes our brand new Data Visualizer, a tool that allows any user to create interactive visualizations from any set of documents, bringing to life information about document dates, metadata, contents, formats, review activity, and predicted relevance.

Viewing your dataset by date, for example, will give you a visualization of the spread of documents created (or modified, accessed, etc.) over a particular period of time.

Viewing your dataset by document type will inform you of the nature of the dataset. Is this mostly email you’re dealing with, or is there a large portion of chat files you’ll need to wade through?

It’s all there for you with Data Visualizer.

Learn more about data visualization in Everlaw in our knowledge base.

Predictive Coding System Upgrade

Long-time users will notice our predictive coding system has a fresh new look. But in addition to the visuals, the new system underwent a major upgrade under the hood. The system now makes it easier to create and train new, more defensible prediction models, provides more guidance on how best to make use of a model’s predictions, and makes the evaluation of a model’s performance both richer and more interactive.

Easier model creation

We’ve changed the way prediction models are created. In the past, you specified the relevant and irrelevant criteria for a model. Now, you’ll set the criteria for documents considered “reviewed”—or those within the universe of documents you’d like to be considered for training. You can then specify the subset of those documents which are “relevant.” All documents that are “reviewed” and not “relevant” will be considered “irrelevant” for the model.

Update the model on demand

Now, you’ll be able to initiate an immediate update of your model from the predictive coding page at the click of a button.

Performance metrics at any relevance boundary

Select any relevance boundary and see the precision, recall, and F1 values at that boundary. This allows you to easily select a body of documents based on your desired precision and recall levels. For example, if you want to ensure that all responsive documents are captured by your search, you may choose to increase your recall cutoff score, ensuring that you cast the widest net possible. On the other hand, if you’re only interested in seeing documents that the model predicts are likely to be hot, you may increase your precision cutoff score, ensuring that you only see documents that are likely to be relevant to your case.

Read Everlaw's Beginner's Guide to Predictive Coding for a thorough introduction to technology-assisted review.

New Homepage Organization Tools

This release brings the introduction of a much-anticipated feature: homepage folders! Organize your personal dashboard just the way you’d like by creating folders that can store multiple cards. And while you’re at it, share folder contents with your colleagues while also managing card permissions.

Check out these feature along with a few more:

  • Searching against a prior search

  • Support of Google Drive files

  • Sorting by billable size

  • Interface changes to the organization admin dashboard

More details on all of these additions can be found here. Please try them out on your next project!