IMD-RT - IMD Redaction Tool Questions_TIS Responses.xlsx
XLSX spreadsheet 18 KB Posted
- Attached to
- IMD Redaction Tool
- Solicitation number
- Not on record
About this file
The document is a Questions and Answers (Q&A) Excel sheet for the FBI's Intelligent Multi-Layer Document Redaction (IMD-RT) tool procurement, revealing detailed requirements for a next-generation redaction software solution. The FBI seeks a sophisticated redaction tool with capabilities including multi-layer redaction architecture, support for 400 concurrent users, integration with case management systems like Sentinel, and ability to handle documents up to 1000 pages with language support for Spanish, Arabic, French, Russian, and German. Key technical requirements include FedRAMP Medium (potentially High) certification, contextual recognition using AI/NLP technologies, metadata scrubbing, custom pattern training, and the ability to produce both "redacted" and "see-through" document versions.
The Q&A sheet provides extensive insights into the FBI's current redaction workflow, using HighView 4, and outlines specific needs such as distinct FOIA and classification redaction layers, email thread deduplication, integration with workflow management systems, and support for various document formats. Storage and scaling expectations include 1 TB of annual data ingestion with 20-30% yearly increase, 8 TB of file storage, and 2 TB of database storage. The solution must support configurable alert systems, selective layer toggling, and accommodate various review processes including FOIA, prepublication review, discovery classification review, and ad-hoc redaction projects, with a strong emphasis on maintaining precise user control over sensitive information removal.
View the file
Other files
| File | Type | Posted |
|---|---|---|
| Redaction RFI_FINAL.pdf |
On GovTribe
Work with this file
- Download the original file
- Contacts named in this file
- Similar government files
- Ask GovTribe AI about this file
Text version
Sheet1
| Question # | Section | Page # | Question | TIS Response |
| 1 | Contractor Qualifications, #2 | 7 | This section requests "..and experience"; however in the "What Should Not Be Included in Your Response" section, item #2 (page 9) instructs "Do not submit past performance information or general descriptions of corporate experience." Please clarify if we should include experience and if so, the parameters of said experience. | You should include experience, but with specific parameters focused on the redaction tool itself rather than past performance documentation. |
You should include information about the product's maturity and market experience, such as how many years the COTS redaction tool has been available, its version history, and update frequency. Include details about platform compatibility and integration experience, covering supported file formats, operating systems, and document management systems.
You should not include CPAR reports, past performance ratings, formal corporate experience write-ups, or past performance questionnaires. Do not submit detailed case studies or narratives about specific client implementations.
We are seeking information about the product itself - capabilities, maturity, and technical specifications - not your company's past performance record or formal experience documentation.
| 1 | General | How is the FBI currently handling redactions and which tool(s) are currently being used? | The FBI currently uses HighVIew 4 for all redaction activities. Active redactions include highlight, redaction and exemption blocks, annotation, freehand, line, bracket, X out, security markings, polygonal redaction, stamps, circle redactions, sticky notes, arrows and polygonal brackets. | |
| 2 | Section 2 - Additional Core Functionality | Page #4 | Can you please describe if and how the NSCOU process of handling and redacting classified information would require special workflows or additional security requirements? | NSCOU is a separate group outside the disclosure units, so a workflow is required that allows disclosure analysts to send work items to an NSCOU analyst for classification review. NSCOU then identifies what information is classified, if any, and marks the classified information in work items with a highlighted box/overlay which can only be removed by NSCOU (not disclosure). The work item is then returned to disclosure for redaction of the classified information by a redaction box over the highlighted box. The current security requirement that locks the highlighted box so that only NSCOU may remove or modify it ensures that classified national security information is properly protected and the boxes are not removed by anyone other than a classification specialist. |
| 3 | Section 6 - Security and Compliance | Page #6 | Does the solution require a FedRAMP high certification, or would FedRAMP medium suffice? | FedRAMP medium - double check this. |
| 4 | Section 7 - Integration and Deployment | Page #6 | What assumptions should we be aware of regarding deployment, integration, or usage? | Do not make any assumptions as this potential requirement can be potentially redefined in the future |
| 5 | Section 7 - Integration and Deployment | Page #6 | Can you please list the document management systems, case management systems, and other enterprise software that would require APIs and compatibility for the redaction tool? | We are currently working to define our next generation solution and we don't know exactly which solutions will be utilized for the specific components. In saying that, the redaction tool would need to integrate with a workflow management system, a case management system such as Sentinel. |
| 6 | Section 7 - Integration and Deployment | Page #6 | Are there requirements around where FBI documents should be stored (on-prem v. private cloud environment)? | C2E |
| 7 | Contractor Qualifications | Page #7 | In regards to the requested cost estimate, can you please provide volume assumptions so that we can estimate costs? | 1 TB a year with a 20-30% annual increase |
| 1 | Section 7, p. 6 | Is the team looking for a cloud-based, on-premise, air-gapped, or hybrid for system requirements? | C2E | |
| 2 | Section 11, p. 7 | What is the maximum number of concurrent users that should be supported? | 400 | |
| 3 | Section 8, p. 6 | How would the team prefer to receive system alerts, through the software itself or by email? | A configurable solution that would allow users to chose. | |
| 4 | Section 1, p. 3 | Based on the description of the multi-layer redaction architecture, can the FBI clarify whether “layers” refer to distinct categories of redactions (e.g., FOIA exemptions, classification markings, etc) that can be selectively toggled on or off during export? | ||
| The FOIA and Classification layers of redactions are distinct (currently salmon and pink color distinctions). When thinking toward the future, we use the system to process under other authorities such as prepublication review, discovery classification review, and records project reviews (or other ad hoc redaction projects). |
Also, the various layers have previously represented other features such as a highlight used in the review process perhaps. A “redacted”/ “see thru” version and a “sealed”/ “solid” version of each file/page processed is needed. When thinking ahead to other uses of toggling layers, layers identifying other specific pieces of information in the FOIA exemptions/classification code/PRO exemptions (such as application of each exemption related to a code and a way to recall that in a report form) would be beneficial.
| 5 | Section 2, p. 4 | NSCOU Process - is this a watermark? | No | |
| 6 | Section 2, p. 5 | Are you interested in PST advanced filtering, remove duplicate threads, and PST redaction? | Yes - better technology is needed to identify email threads and improve deduplication | |
| 7 | Section 3, p.5 | Do you need race/ethnicity redacted as part of your PII categories? | Case by case basis but that is not a normal reason for redacting | |
| 8 | Section 1, p. 3 | What is the average length (pages) of your files? Do you anticipate working with files over 1000 pages in length? | Average is 90 pages, but yes we do work with sizes over 1000. | |
| 9 | - | Do you need the ability to password protect your files? | No | |
| 10 | - | Should the system support integration with FBI's internal case-management system? | Yes | |
| 11 | - | Do you need functionality that would allow you to highlight text or apply sticky notes to different pages? | Yes | |
| 12 | - | Do you need the ability to replace text, rather than redact it? Ex: Replace John Doe with Suspect 1 | No | |
| 13 | - | Do you anticipate redacting files exclusively in English, or in other languages? If others, can you please expand on which languages you commonly see? | If the system could translate that would be preferred but not required Spanish, Arabic, French, Russian, German | |
| 1 | Contractor Qualifications #5 (Page 7) | For the estimated cost, can the Government please provide any estimated parameters such as, estimated user count, number of gigabytes (GBs) in legacy data, proposed GB ingested per year? | 400 users, 8TB of file storage, 2TB of database and a guestimate of 1 TB ingested per year. | |
| 2 | RFI Preparation Checklist Item #4 (Page 11) | Can the FBI clarify which section is being referenced in “All items for Section ?? have been included for Industry Day eligibility determination”? | ||
| Under comtemplated requirement | ||||
| 1 | 2 | 4 | The RFI mentions "Contextual Recognition" beyond pattern matching. Is the government seeking a solution that utilizes Large Language Models (LLMs) or NLP for semantic analysis? If so, are there constraints on using cloud-based inference models due to data classification levels? | Yes to both questions. |
| 2 | 2 | 4 | Regarding the "Master Document" requirement, does the government require this multi-layer file to be a proprietary format unique to the vendor, or does the FBI prefer a non-proprietary open standard to ensure long-term data interoperability and archive retrieval? | Latter |
| 3 | 2 | 4 | Regarding migration, does the legacy system currently utilize a "burned-in" redaction method, or will the vendor be responsible for migrating existing un-flattened redaction coordinates/metadata from the legacy system into the new multi-layer architecture? | We currently utilize a burned in sealed product as well as a layered working product. The migration strategy and the responsible entity for establishing that strategy is TBD. The focus of this RFI is to determine the most suitable product to meet our users document explotation needs. |
| 4 | 2 | 4 | Specifically regarding "Metadata Handling", does the requirement extend to "scrubbing" dirty metadata (e.g., author, creation time, version history) automatically upon export, or must this be a manual review step within the layer interface? | Yes |
| 5 | 2 | 4 | The RFI lists "SaaS" as a deployment option but also references "NSCOU process" for classified information. Can the government confirm if the intended deployment environment is an air-gapped (Top Secret/Secret) network, or if a FedRAMP High cloud environment is sufficient for the highest classification level of data expected? | FedRamp High |
| 6 | 2 | 4 | Regarding "Custom Patterns", does the FBI require the capability to train custom AI models for agency-specific identifiers (e.g., operation code names), and must this model fine-tuning occur exclusively within the agency's secure environment? | Yes |
| 7 | 2 | 4 | Does the Government require the system to display confidence scores or contextual reasoning (e.g., highlighting the trigger context) for AI-generated suggestions to assist the mandatory human verification process? | This is preferred. |
| 8 | 2 | 4 | Are there specific Case Management Systems or Document Management Systems that the solution must provide native API connectors for, or is a standard REST API capability sufficient for compliance? | A standard REST API capability is sufficient. We are currently working to define our next generation solution and we don't know exactly which solutions will be utilized for the specific components. In saying that, the redaction tool would need to integrate with a workflow management system, a case management system such as Sentinel. |
| Which business applications, if any, require integration with the redaction solution? | Many different microservices will need to interact with the redaction tool. Example: Content Manager/Sentinel | |||
| How many users will require access to the redaction system? Please specify both the total number of anticipated users and the expected number of concurrent users. | Estimated 400 total | |||
| Does the Bureau have a preferred deployment model for the solution (e.g., on-premises, cloud-hosted, hybrid, or SaaS)? | C2E | |||
| Regarding the cost estimate requested under the Contractor Qualifications section: Will the cost estimate be counted toward the page limit for the proposal response, or should it be submitted as a separate attachment? | The potential estimated cost will NOT be counted toward the page limit for this RFI |
File details come from the government source that posted it. Updated .