The Clash of Titans: CAL vs TAR 1 – A Battle for Supremacy

As an attorney, or eDiscovery expert, you may have heard of or worked with TAR 1.0 and TAR 2.0. Would you consider yourself well-versed in the differences between them? Do you know how to select one over the other for a project? Both options are useful in eDiscovery to automate the review process of large sets of data. There are key differences that set them apart. Let’s look at how Continuous Active Learning and Technology Assisted Review 1.0 differ. 

What is Technology Assisted Review (TAR)?

Technology Assisted Review (TAR), also known as predictive coding, uses algorithms to analyze documents and identify patterns in order to classify them as relevant or non-relevant. It includes a training phase where trained reviewers provide feedback on sample documents in order to “teach” the algorithm what constitutes relevance for those documents. This training allows for the algorithm to build a model of what relevance looks like in a document. Once the algorithm training is complete, it can then apply that same knowledge to other documents in the dataset. This allows for more accurate reviews because the algorithm can recognize patterns that human reviewers can miss. It also increases the pace of review by batching the likely-to-be-responsive documents first. 

What Is Continuous Active Learning or “TAR 2.0?”

Continuous Active Learning (TAR 2.0) is an advanced form of Technology Assisted Review. It uses machine learning algorithms to continuously refine its classification decisions at the same time human reviewers are reviewing documents. This means that as each document moves through review, the algorithm learns from reviewer input and adjusts its decisions accordingly. This is a key difference from TAR 1.0 (sometimes referred to as Simple Active Learning). TAR 1.0 requires manual re-training after each document review cycle is complete.

With Tar 2.0, organizations can conduct faster and more accurate reviews. This is because the algorithm will become increasingly better at recognizing relevant documents over time. Often times, clients will use CAL/TAR 2.0 for prioritization and getting eyes on likely to be responsive documents first. If the ESI protocol allows for it, clients using TAR 1.0 and TAR 2.0 have the option of stopping review and conducting an elusion sample. The elusion sample can prove that reviewing the remaining documents would prove overly burdensome and not fruitful.  


No matter which eDiscovery method your organization chooses – whether it be Simple Active Learning (TAR 1.0) or Continuous Active Learning (TAR 2.0) – in any review process, accuracy and efficiency should always be the top priority. While both TAR 1.0 and TAR 2.0 can help streamline document reviews and save time, CAL/TAR 2.0 provides an even greater level of accuracy and efficiency due to its ability to continuously learn based on reviewer input while also minimizing manual re-training efforts. Ultimately, understanding the differences between these two methods will enable you to make informed decisions when it comes time to choose an eDiscovery solution for your organization’s needs. Sandline’s analytics experts are happy to help discuss the right solution for your project.