Expected release date: June 22, 2018 (US & AUS), June 29, 2018 (EU & CAN)
This month, we’re releasing significant functionality updates to our search page and results table, allowing you to create more powerful searches. You can group and filter from the search page itself, which opens up a host of more efficient workflows. Additionally, you’ll notice a host of other interface improvements to search. Read below for examples of how you can leverage this suite of new features.
Beyond this exciting update, we’ve included improvements to administrator functionality. When evaluating predictive coding models, you now have the option to select a truly random sample for your holdout set, thus improving the rigor of your performance statistics. Additionally, we can now support multi-page TIFFs upon processed upload, have created a Produced To results table column for production volumes, and have allowed exports of Chronologies to be organized by events.
More Options in search: duplicates, sampling, grouping, filtering
You now have the option to group and filter individual logical containers, as well as negate entire logical containers directly from the search page. These settings can be customized from the More Options tab, at the bottom right of each logical container, which also allows you to specify sampling and duplicate inclusion or exclusion preferences. You can learn more in our help article, as well as below. 
Here are some ways that these new search options can drastically improve your workflows.
Pre-production QA
Need to prepare document families for production while making sure that nothing has been skipped over during review? You can now run a search for all documents, including attachments, that have been coded responsive, and have NOT been coded privileged.
Find all documents that have attachments
You can easily group your search by attachment and then remove the parent documents. This will retrieve all documents that are attachments within your database.
Remove non-inclusive emails
If you are searching in a complete project, you now have the option to remove non-inclusive emails from email threads. This can help you filter out repetitive, superfluous email files so that you can focus your time on the most complete ones.
Deduplicate across all documents that meet your search criteria
Want to deduplicate only within the context of a particular search? With Everlaw’s new search options, you can group your documents with their exact duplicates, and choose to only view one copy of every document in your search results. To learn more about deduplication within Everlaw search, see this deduplication article.
The same grouping and removal options will also be accessible from the results table toolbar. A single Options icon will replace the Duplicates and Group icons.
Search interface improvements
You will also notice a few interface changes that will make your search experience more intuitive. At the top of the builder, there are instructions for how to add logical operators to your search. You can click or drag and drop the operators to include them in your query builder.
You can now negate logical containers. This makes searches like the one below easier and more intuitive. The below search returns all documents that are neither rated nor coded. You can still negate individual search terms, as you could previously!
Finally, we have updated the date range search interface. The new look has a practical component in addition to an aesthetic one. It requires fewer clicks to select the date you need, while still displaying a calendar view.
Predictive coding: additional option for reviewing holdout set
Our predictive coding tool exists to support a diverse range of users, from those who want a quick and simple tool to save time, to those who want a statistically rigorous and defensible model. With this release we are introducing rigorous performance statistics, which ensures that a purely random sample of holdout documents are used to generate your model's performance statistics.
In the previous paradigm, performance statistics were solely generated based on the total set of reviewed holdout documents. This method of generating performance statistics risks using an unrepresentative sample of documents to measure model performance. With the introduction of rigorous performance statistics, you can be confident that your model's performance is representative of the documents in your project. This addition gives you the option to prioritize statistical defensibility or simplicity, based on your objectives for leveraging predictive coding.
For more detail about rigorous performance statistics, read our help article.
Improvements to Chronology exports
When you export documents in a Chronology and choose to include events as an export column, the resulting CSV will automatically group your documents with their events, preventing you from having to do the work yourself!
Additionally, you can now export only a portion of your Chronology by checking the relevant documents, or by filtering on the labels you wish to export.
Administrative improvements
- "Produced To" and "Produced To Volume" results table columns: To better help you identify which original documents correspond to which produced documents, a Produced To field will be generated whenever a document is produced. The Produced To field will contain the Bates number(s) for the produced version of the document. You can access this field from the results table, as well as the document’s history. Another new field, Produced To Volume, will list the productions of which the document was a part.
- Support for multi-page TIFF file upload: Previously, Everlaw supported uploads of single-page TIFF files. With this release, uploads containing multi-page TIFF files are also supported. This article includes more detail on processed data upload requirements.