How is the doable done?
- Predictive coding 101.
Predictive coding is a type of learning software that can be trained to classify documents into various categories. A subject matter expert trains the system using sample documents for the target category. The application uses these samples to identify the attributes that distinguish documents that belong to the category from those that don’t.
- Doing the un-doable.
Records retention has become virtually un-doable. Defining a retention schedule is one thing, but implementing it has proven more or less impossible. Manual tagging of retention categories, usually at the end user level, involves too much effort, requires an unrealistic level of compliance, and is highly prone to error. But predictive coding makes records retention doable, enabling consistent classification with reasonable effort. It does by propagating the record classification decisions of a subject matter expert, on a small number of sample documents, across the entire repository.
- Predictive coding for records retention.
Zoom’s predictive coding application enables systematic records retention via a controlled, repeatable process. Users train Zoom to classify documents by retention and remediation categories. The QA mechanism verifies quality and statistically quantifies results. Zoom’s decision-support zone then allows the user to decide on a retention policy that meets organization’s cost and risk guidelines. Based on this policy, Zoom assigns documents to the various retention buckets and flags junk documents for deletion.
- Human-bsaed classification of business records has become un-doable
- Organizations are trying to tag the retention categories manually, usually at the end user level, where each employee is required to classify their own documents and emails
- As a result, the manual approach is failing:
- Involves too much effort
- Requires an unrealistic level of compliance
- Is highly prone to error due to multiple and divergent interpretations of retention schedule criteria
From un-doable to doable
- With the recent emergence of predictive coding, the erstwhile un-doable has become eminently doable
- Predictive coding is a type of learning software that can be trained to classify documents
- A subject matter expert trains the system using sample documents for each target category
- The application uses these samples to identify the attributes that distinguish documents that belong to the category from those that don’t
- In records retention, predictive coding can be used to identify junk, and to classify documents in various retention categories, per the organization’s retention schedule
How it’s done
Zoom’s predictive coding application enables systematic records retention via a controlled, repeatable process:
- Users train Zoom to classify documents by retention categories
- The system is trained by a central core team of subject matter experts, with a shared interpretation of the retention schedule, to ensure consistency
- The QA mechanism verifies quality and statistically quantifies results
- Zoom’s decision-support zone then allows the user to decide on a retention policy that meets organization’s cost and risk guidelines
- Based on this policy, Zoom assigns documents to the various retention and junk categories
More than just doable
Doable means more than just “doing records retention”. Doable also means a process that is defensible. And doable also demands that project risk be minimal.
Zoom’s predictive coding enables the transition of records retention compliance and legacy data remediation from mission-almost-impossible to mission-very-doable.
- Doable & Defensible.
Zoom forms the basis of a highly defensible records retention approach:
- Zoom is adopted from e-discovery, where it has been proven over several years and approved by the courts
- Zoom facilitates a standardized, repeatable, closely-monitored process
- Zoom enables a centralized training process in which the organization’s best-informed SME’s train the system via a focused effort and in a consistent and nuanced manner
- Zoom includes built-in systematic quality control and statistically-valid quantification of outcomes
- Doable & Defensible & Low risk.
- Zoom is an add-on to your existing information governance environment
- It does not replicate or replace existing repositories or archives
- As an add-on, Zoom integrates with existing archives to determine retention periods for each document
- This approach minimizes project cost and risk, while protecting the organization’s sunk investment in existing systems
Doing more than records retention
In addition to predictive coding for records retention, Zoom offers an all-round analytics suite for information governance. Capabilities include:
- Search and complex search
- Data profiling
- Pivot analysis
- Email threads
- Similar documents
- Predictive coding
- Metadata faceting
- Language detection
- Natural expressions
These capabilities are used to support a wide range of information governance use scenarios, including records retention, compliance, legacy data remediation, data leakage prevention, behavior alerts and internal investigations.