Jump to Sentences TAR Thinks Make a Document Responsive

Electronic Discovery services provider Fronteo has developed an add-in for Relativity that will help expedite technology assisted review and provide a firmer basis to show that it is actually working. KIBIT Automator highlights individual sentences in a document flagged as being responsive, so a reviewer can easily see which parts of the document the TAR software believes make the document relevant.

Fronteo claims that the speed of the human review necessary for TAR can be more than doubled. KIBIT can also generate heat maps to indicate where documents marked as being non-responsive were coded differently than other documents with a high degree of textual similarity. It performs email threading and uses metadata in its machine learning algorithm.

Fronteo claims that its software can perform email threading on more than 2 million documents within 5 hours.

Email threading is performed on emails which are in Asian languages by making use of systems used in Japan. Fronteo is a Japanese company, so it appears as though this is something it does particularly well.

The Obsolescence of Seed Set TAR

Predictive coding or technology assisted review is often regarded as a process which involves a static 'seed set' that is used as a basis on which to categorize a full document set. A group of documents is identified through manual review. The software trains based on that seed set, which should contain documents which represent key concepts. A QC is done to find an acceptable 'overturn' rate - a low percentage of documents that must be re-categorized by a human reviewer - an indication that an effective seed set has been chosen. A report can be prepared to identify which seed set documents lead to the most overturns, and may need to be removed.

The obsolescence of this type of review (known as TAR 1.0), is evident in Relativity's decision to deprecate sample-based learning, a form of seed set based TAR. After September 2021, it will no longer be possible to run sample-based learning projects in Relativity.

After this September, Relativity will direct its clients to use Active Learning, a TAR 2.0 review process, which uses continUous active learning (CAL) to improve machine learning continuously as manual reviewers make coding decisions.

The FTC's Guide to HSR Second Request Productions

The FTC's Hart-Scott-Rodino Premerger Notification Office has published a guide, Model Request for Additional Information and Documentary Material (Second Request), which gives the specifications for responding to a HSR Second Request for information about a proposed merger which the Commission wants to investigate. Litigation support analysts and paralegals should keep in mind the following are among the wide array of documentation and data that is required:

1. Org charts for each department in the company.

2. A data map for the company.

3. The identity of electronic databases for relevant products and services, including the database type, (flat, relational, or enterprise); fields, query forms, and reports available or maintained, and the platform used.

4. Data sets from the database with price, sales, and other data.

5. Minutes and recordings of board meetings.

6. The company's document retention and destruction policy.

7. A list of the federal judicial districts in which the company has an agent to receive service of process. [The company may agree to a stipulation to accept service throughout the United States.]

8. "electronic production tools or software packages utilized by the Company in responding to this Request for: keyword searching, Technology Assisted Review, email threading, de-duplication, and global de-duplication or near-de-duplication."

9. Keywords used in searches for each data custodian.

10. Information on the TAR software, including

    1. how the seed set was selected.

    2. the number of documents reviewed manually.

    3. the number of nonresponsive documents identified without manual review.

    4. how the company validated its TAR results.

11. A designated person to testify about the review process for the electronic production.

In addition to these specifications, the guide also makes the following instructions for the company's response:

1. The guide instructs that the company include all responsive documents obtained up to 45 days prior to the date on which the company complies with the request.

2. PII and PHI, including social security numbers, and financial account numbers, must be removed.

3. A specific form of production is given:

   1. Excel, Access, and PowerPoint files must be in native format.

   2. Emails, their attachments, and other documentation should be in TIFF (Group IV) format, and include extracted text and metadata fields. These metadata fields are required for non-email documents, and other metadata fields are specified for emails and their attachments.

4. Productions over 10GB have to be made on hard drives.

5. NIST FIPS-Compliant encryption is encouraged for all productions.

6. "Each production shall be submitted with a transmittal letter that includes the FTC matter number; production volume name; encryption method/software used; list of custodians and document identification number range for each; total number of documents; and a list of load file fields in the order in which they are organized in the load file."

7. A FTC representative must authorize the use of de-duplication and email threading software.

8. Foreign language documents have to be translated into English.

9. Documents must be produced in color when it is needed in order to make substantive content intelligible.

10. A log must be included listing the custodian for each document range.