Predictive Coding Do’s and Don’ts

Posted in: Media

What is predictive coding – definition

Predictive coding is an instrument learning method that uses software combined with human guidance to train document review to identify relevant documents and reduce the number of irrelevant and non responsive documents that need to be reviewed manually.

In e-discovery, predictive coding is primarily used to quickly and accurately locate relevant document in order to expedite the review process.

How predictive coding works

The software is a keyword search designed to look at broad patterns of language to establish what is relevant, while creating a profile of both relevant and irrelevant documents.

This can assist in faster decision making, saving time and money, and is more reliable in finding documents which a human reviewer might have missed.

Do’s and Don’ts

Use more than one technique/algorithm
Start with familiar data in order to view its success and functionality, then expand into unfamiliar data
Don’t use the default model accuracy metric
Don’t leave it all up to the software – clear definition of what is needed

Keyword search – How do Private investigators work

When it comes to private investigations, we can use either a manual search of keywords that change from case to case, an automated system that allows the investigator to define keywords to narrow the search.

For example, TA9 IntSight is a data analysis platform that is designed to integrate with a variety of data sources and channels all the data collected to one source. One of the options is to predefine keywords relevant to our search.