Our third release of the year—and our 30th for Everlaw—is a big one!
1. Automatic Foreign Language Detection and Translation
Nowadays, it’s not uncommon for your document collection to include at least some content in a language other than English. You can use the new “Language” term in the search interface to find documents based on the languages they contain, including the option to focus on documents that are primarily in a given language.
When you’re reviewing a document with foreign-language text, you can flip between regular hit highlighting and foreign-language highlighting in the text view by clicking on the “Languages” tab. From this tab, you can see all foreign-language text highlighted, and you can translate that text into English with the push of a button. While no substitute for professional translation, this machine translation can help you better understand the contents of the document and potentially prioritize it for more extensive review.
With this first release, Everlaw’s language detection and translation feature supports 53 of the world’s most widely-used languages.
2. Improved and Expanded Prediction Engine
As data volumes grow, predictive coding is an increasingly indispensable tool to help you prioritize documents for review, QC the work you’ve done, and possibly avoid having to manually review every document. With this release, we’re giving you more information from Everlaw’s Prediction Engine so you can better understand how the system is performing, as well as the ability to generate prediction models around any pair of classifications in your case.
On the Document Analytics tab under Case Analytics, you’ll notice that the Prediction Engine now gives you specific, actionable guidance on what you can do to improve the accuracy of the predictions for the model. Below that and the familiar Distribution chart, we’ve added a new Training Coverage chart that shows you how well your documents are understood by the Prediction Engine. It has three dimensions. The color gradient tells you how many documents are in each cell, with darker colors indicating more documents. On the horizontal axis, predicted relevance is shown as it is on the distribution chart above. On the vertical axis, the documents are grouped according to how well the contents of those documents are covered in the training set. More coverage means that the Prediction Engine will have a better understanding of the words in those documents and their impact on relevance. This, in turn, will generally yield more accurate predictions.
Perhaps most importantly, the power of Everlaw’s Prediction Engine is no longer limited to predicting document ratings; now you can create a model for any pair of classifications in your case. You can define what classification you’ll use for relevant and irrelevant documents in this model using the familiar visual search interface. Both relevant and irrelevant documents should be ones you’ll affirmatively categorize as such as part of review—that way, the Prediction Engine can make better predictions about what distinguishes relevant from irrelevant documents. Once you’ve created your new model, you can access it from the model drop-down list, enjoying all of the same information about how it’s working and how to improve performance.
3. User-driven StoryBuilder Enhancements
We’ve made two major enhancements to StoryBuilder, based entirely on your feedback. First, we’ve added the ability to import an existing outline from a Word document; simply select the Import button and point it to your .docx file.
We’ve also added a new export option: PDF (with images). This will generate a single PDF file that includes both the outline and the images for every document you’ve referenced in that outline. The images are included in the order in which they are first referenced in your outline. We hope this simplifies the process of creating binders for depositions or other offline activities.
4. Improved Batch Updating of Similar Documents in the Review Window
As you know, the Context Panel to the left of the review window shows you all of the documents that are related to the document at hand, and gives you the ability to apply the same ratings and codes across all of them at once. Based on your feedback, we’ve replaced the “Apply to Selected” batch coding button at the bottom of the Context Panel with an “Update Selected” button that brings up the familiar batch coding panel. It is pre-populated with the potential batch changes, giving you a clear sense of what will change before you commit those changes.
As usual, we’ve also made many other tweaks and improvements throughout Everlaw, including more informative cards on the home page, the ability to filter live user activity by user group, and a column indicating your current row number in the results table. Enjoy all of these improvements, and keep your feedback coming so we can prioritize your needs on our roadmap!