This opportunity is closed for applications

The deadline was Thursday 7 March 2019

Home Office

Natural Language Processing (NLP) Analysis Platform

13 Incomplete applications

8 SME, 5 large

28 Completed applications

20 SME, 8 large

• Published: Thursday 21 February 2019
• Deadline for asking questions: Thursday 28 February 2019 at 11:59pm GMT
• Closing date for applications: Thursday 7 March 2019 at 11:59pm GMT

Overview

• Summary of the work: The Home Office requires the development of an innovative and cutting edge Natural Language Processing (NLP) analysis platform that enables the ingest, indexing and analysis of large text-based data sets of terrorist and extremist propaganda.
• Latest start date: Friday 29 March 2019
• Location: No specific location, eg they can work remotely
• Organisation the work is for: Home Office
• Budget range: First statement of work has a limit of £200-250K

About the work

• Why the work is being done: The Home Office is responsible for delivering analysis on terrorist and extremist communications and online behaviour that shapes HMG’s policy and the delivery of counter-terrorism & counter-extremism communications.; ; We have identified a need to better track and recognise trends in terrorist and extremist propaganda over time at scale and compare this data to other terrorism and extremism datasets to understand the relationship between propaganda, attitudes and behaviours.; ; This capability will help improve the speed, quality and variety of the Home Office's analysis and provide us with valuable data for the shaping and targeting of counter and alternative narrative communications campaigns.
• Problem to be solved: We are seeking the development and delivery of a NLP analysis platform that enables the ingest and indexation of large text-based propaganda, allowing users to search within this platform for content by time, producer, title or keywords. In addition, it will assess propaganda against the following metrics:; ; 1. Consistency of rhetorical/thematic focus over time.; ; 2. Alignment to other terrorist/extremist rhetoric.; ; 3. Influence of message/messenger (supplier to propose measures of influence).; ; 4. Relationship between propaganda and behaviours (using open source data on terrorist activity).; ; Initial data ingest for this project will be in the form of the Home Office's text data.
• Who the users are and what they need to do: Home Office analysts need to be able to better understand macro and micro trends in terrorist and extremist propaganda and messaging over time in order to understand the threat this content poses and provide relevant policy and operational advice. Analysts would work with the supplier to ensure that the functionality of the solution meets their needs.
• Any work that’s already been done: The Home Office has carried out some concept work with a supplier on the application of Natural Language Processing for the detection of terrorist propaganda. We intend to use the results of this work, code, training data and the method adopted to inform the development of this project. This will enable the Home Office and the winning supplier of the procurement, to bring to market a tool based on proven learnings.
• Existing team: The supplier would be working with an internal team of Home Office analysts.
• Current phase: Alpha

Work setup

• Address where the work will take place: It is anticipated that the successful bidder will be able to work from their own location.
• Working arrangements: The Authority would look to hold an initial workshop followed by meeting at milestones and touchpoints, subject to further definition and input from the bidders proposal.
• Security clearance: SC level clearance is required for non-development staff working under the proposed contract.; ; The supplier must have a location and technology that supports development at Official Sensitive. The successful supplier will also need to confirm they have Cyber Essentials accreditation or have it prior to receiving any data.

Additional information

• Additional terms and conditions: 1. Payment will be via the Home Office's electronic payment system ; 2. Any travel and expenses will be in accordance with Home Office policy.; 3. Key Supplier staff included in proposals will be expected to work on the contract.; 4. Supplier will be required to agree to the Authority's security requirements including having Cyber Essentials; 5. The Supplier shall (when designing, implementing and delivering the Services) ensure that it and its employees or sub-contractors work in alignment to the Government Commercial Function Supplier Partner Code of Conduct (September 2017) https://www.gov.uk/government/publications/supplier-code-of-conduct throughout the term of the contract.

Skills and experience

Buyers will use the essential and nice-to-have skills and experience to help them evaluate suppliers’ technical competence.

• Essential skills and experience:
  • Demonstrate experience of developing tooling for the purpose of data analysis.
  • Demonstrate experience of applying a range of natural language processing techniques to support the analysis of text-based data.
• Nice-to-have skills and experience:
  • Demonstrate experience of working in counter-terrorism & extremism and/or on the analysis of terrorist or extremist propaganda.
  • Provide evidence of Government domain expertise and Government Digital Service
  • Demonstrate experience of developing dashboards for data analysis.
  • Experience of designing services to best practice industry standards.
  • Provide evidence of providing a scalable, resource pool whilst maintaining governance and delivery.
  • Demonstrate experience of supporting software development using natural language processing techniques
  • Have a suitable location and technology to develop, manage and store Official-Sensitive data.

How suppliers will be evaluated

• How many suppliers to evaluate: 5
• Proposal criteria:
  • Approach and methodology - clear and detailed timeframes, considering all factors from develop through to delivery, future proofing, ongoing maintenance and upgrades (18%)
  • Technical infrastructure - evidence that the organisation has the technical ability to develop, build, future and maintain the platform including upgrades (12%)
  • How the approach or solution meets user needs - how the solution will be tailored to meeting the problem statement and how this might be managed during the process. (12%)
  • Risks and dependencies and possible mitigation - evidence of consideration and understanding of risks (6%)
  • Team structure and composition (including work histories for core staff) (6%)
  • Value for money - demonstrates a clear method in ensuring that costs provided are managed effectively (6%)
• Cultural fit criteria:
  • Provide evidence of working with UK Government, local authorities or large organisations with a range of different stakeholders (2%).
  • Provide evidence of sharing knowledge, experience and collaborating as part of a wider group (2%).
  • Have the ability to communicate effectively and be responsive to client needs (1%).
  • Has respect and understanding of client's internal governance and procedures (1%).
  • Be innovative, cutting edge and not afraid to try new things (2%)
  • Have security at the heart of its organisation (2%)
• Payment approach: Capped time and materials
• Assessment methods:
  • Written proposal
  • Case study
  • Work history
  • Reference
  • Presentation
• Evaluation weighting:

  Technical competence

  60%

  Cultural fit

  10%

  Price

  30%

Questions asked by suppliers

• 1. Any work that’s already been done' – who was the supplier involved and can you share their work to date?: This information will be shared at the next stage, following shortlisting.
• 2. What's Home Office's full tech stack in relation to this development e.g. what other technologies will this development need to interact with?: This information will be shared at the next stage, following shortlisting.
• 3. Any work that’s already been done' – what was the tech stack used by the supplier?: This information will be shared at the next stage, following shortlisting.
• 4. SC level clearance is required for non-development staff working under the proposed contract."Could you please clarify what is classed as non-development staff?And would the development staff require any other type of security clearance?: Non-development staff are those that would oversee delivery of the project and liaise with the Home Office on their requirements. Information about the Home Office’s analytical objectives and intended outcomes of this project can only be discussed at SC.; ; Development staff are those responsible for building the technology. These staff will not require SC. While they will be able to access the data the Home Office shares with the supplier, they will not be party to details on the technology’s deployment. All Staff should be aligned the minimum specification of HMG Baseline Personnel Security Standard found at: https://www.gov.uk/government/publications/hmg-personnel-security-controls.
• 5. The opportunity specifies: SC level clearance is required for non-development staff working under the proposed contract.Could you expand on the reason for the difference in clearance, the type of roles you expect to be involved that are classed as non-development and the level of clearance you expect for development staff.: Please see above.
• 6. Will you open to consider a proposal with onsite(within UK, London) -offshore (Remote, outside of UK) delivery mechanism?: the acceptability and viability of a suppliers proposal being based overseas may be subject to the countries stated and their acceptability within Home Office’s security policy.
• 7. What languages are you expecting to process for topic modelling, entity identification, etc?: English is required. Arabic and major European languages are desirable.
• 8. Please could you confirm that the only language you expect to be supported as part of this solution is English? If not, what other languages need to be in scope?: Please see above.
• 9. If initial data ingest for this project will be in the form of the Home Office's text data, please could you describe the dataset (e.g. format, size, context etc) ?: We are unable to disclose details of the dataset at this stage, however you should expect that the totality of this dataset will amount to millions of words (including repetition). Tooling that could ingest a variety of file formats would be preferable.
• 10. Is SC clearance required only for resources that have visibility of the data or for everyone involved in the project?: Please see above SC Clearance response.
• 11. Would you consider using a pre-built, out of the box solution (e.g. NLP APIs on cloud), rather than developing one from the ground up?: Our requirement is for a bespoke solution to which the Home Office could own the IPR.
• 12. Please can you confirm if you are planning to re-deploy the same tech stack used in the previous concept work?: This work does not necessarily need to be developed using the tech stack we previously used. Design on a different tech stack could be informed from the output of our previous work.
• 13. Can the NLP analysis platform be hosted in a public cloud environment?: We will consider both cloud-hosted and Home Office hosted proposals.
• 14. Will it be permitted to propose resources who are currently in the process of attaining HO SC and have SC Clearance from other government bodies (e.g. Sellafield and MBDA)?: Yes
• 15. We intend to use the results of this work (the code, training data and the method adopted) to inform the development of this project. If, upon review, other methods are discovered that provide better results, would we be permitted to utilise such methods?: The work already undertaken will be provided to inform the development of the digital outcome, however will not define it. If other methods are discovered or introduced that contribute to the digital outcome, then this will be permitted, following consultation and acceptance with the Home Office.
• 16. Please could you confirm what Home Office IT Business Portfolio this requirement is originating from i.e. OSCT, NCDS, or other?: OSCT.
• 17. Would you please provide information on how the source data can accessed and estimate on the amount of data?: Access would be through hard drive transfer. We anticipate that the total dataset amounts to millions of words (including repeated terms).