1232SA23Q0118 Sole Source Justification_Redacted.pdf

The Sources Sought Notice yielded capability statements from The technical POC reviewed the contractors’ capabilities and determined none were capable of fulling this requirement. The technical POC’s specific rationale is outlined below:

was found to be technically incapable as this vendor, describes experience with search technology, which is not required. The company does not meet requirements in skills, resources, or experience specific to the state-of-the art consultation on controlled vocabulary and the international and other related semantic web standards or AI appropriate for subject indexing, nor do they have experience or resources related to their applicability to NAL's specific interoperability scenarios, infrastructure, and technical and communications support required in the SOW.

was found to be technically incapable as this vendor, does not meet requirements in skills, resources, or experience specific to the state-of-the-art consultation on controlled vocabulary and the international and other related semantic web standards or AI appropriate for subject indexing, nor do they have experience or resources related to their applicability to NAL's specific interoperability scenarios, infrastructure, and technical and communications support required in the SOW.

was found to be technically incapable as this vendor, has familiarity with some NAL IT technology and programs, however this does not meet requirements in skills, resources, or experience specific to the state-of the-art consultation on controlled vocabulary and the international and other related semantic web standards or AI appropriate for subject indexing (not search indexing), which underpins our requirements for subject indexing quality assurance and related research, testing, and quality assurance and workflow development, nor do they mention any experience or resources related to their applicability to NAL's specific interoperability scenarios, infrastructure, and technical and communications support for applying NALT that are required in the SOW.

As there are zero technically capable responses provided, it is determined that only one source is available to meet this requirement. Market Research Memo is attached to end of this document.

6. Is this a new requirement or a follow-on to an expiring award?

New requirement Follow-on to an expiring award

7. Describe the requirement. Please note that the requirement is the minimum acceptable specification or capability, and is not necessarily the best in the industry. Please note whether any additional follow-on work is anticipated in response to the current proposed effort.

ARS is focused on developing support and infrastructure for Artificial Intelligence (AI) and Machine Learning (ML), along with SCINet to empower the Agricultural Research Service (ARS). Even ARS management is developing AI and Natural Language Processing (NLP) to synergize budget and other constraint analyses to support ARS research and partnering efforts around the world. The National Agricultural Library (NAL) Indexing and Informatics Branch (IIB) can be a great resource for this work by developing and supporting AI to improve NAL indexing quality and production, and to develop resources that researchers and others can and will use for AI research and development.

A recent development in the NALT for the Machine Age Initiative is an internal partnership with ARS Partners for Data Innovations, which will require continuing support for this contract as the resources developed and tested in the prior contracts are expanded for wider application across the agency, and the USDA. We can meet this challenge by continuing the optimization of the National Agricultural Library Thesaurus Concept Space (NALT Concept Space) for AI, and enhancing its online presence. An optimized NALT Concept Space can be used by NAL for Knowledge Services Division or other projects, and to improve the indexing quality of AGRICOLA, PubAG, Ag Data Commons, as well as improving the search results for NAL online products and tools. ARS researchers or managers will be engaged in NALT Concept Space curation activities, and use the optimized NALT Concept Space for their own AI research and development. This approach will provide NAL and ARS Administrators with exciting new stories to inform about

SOLE SOURCE JUSTIFICATION

MANDATORY TEMPLATE FOR FEDERAL PROCUREMENTS UNDER THE SIMPLIFIED ACQUISITION THRESHOLD results, data, and other advances in AI, NLP, and ML because of our research and development, or by those in ARS using these new NAL resources.

The project integrates assessment to recommend or apply changes to IIB overall business operations and workflows, and state of the art technology. The expected outcomes are to 1) develop quality metrics and increase automation of NAL subject indexing using NALT Concept Space, and 2) support the curation of NALT Concept Space itself through automation, and 3) through the engagement of subject matter experts, and 4) allow NAL to provide relevant, new state-of-the-art tools and services for the agricultural research and library communities.

Osuuskunta Sange’s principle, Dr. Osma Suominen will lead integration of the vocabulary and indexing tool, Annif, and other state-of-the-art technology for indexing, and provide technical support and consultation for design and curation of NALT Concept Space and overall business process review.

8. List the proposed vendor and contact information.

OSUUSKUNTA SANGE

SAM: FEMELKHULLT6

CAGE: A705G

DUNS: 369485946

Aurorankatu 13B 16 Helsinki, Uusimaa 00100, Finland Osma.suominen@sange.fi

9. Describe why the proposed vendor is the only vendor from which the requirement can be obtained. Examples of acceptable justifications include but are not limited to: sole manufacturer and seller, adherence to warranty provisions, urgency, etc. If other vendors can provide similar products or services, describe why they cannot meet our particular requirement.

Dr. Suominen is uniquely qualified to provide services for this contract. He has a Doctor of Technology, and is a computer scientist specializing in user interface design, Linked Data and Semantic Web Technologies, combined with extensive experience working with national libraries, and other national and international organizations who maintain controlled vocabularies, and/or perform indexing for agriculture, including past work with NAL on the GACS vocabulary, a key component of this contract work. Dr. Suominen is a creator of the open-source web tool Annif to apply machine learning for automated indexing, which is ongoing, and central to this work.

Annif, while it cannot meet our full indexing production needs yet, Annif will allow us to perform the development work and meet the requirements for content expansion, quality enhancement, and support much better integration with NALT Concept Space, and the engagement and support of research communities that the larger vendors and tools, including Cogito, simply cannot do. Annif has the clear advantage and is uniquely suited for NAL in terms of cost and NAL staff skills. It is also unique from other tools in its ability to deploy NAL created indexing with multiple vocabularies. This is critical to the NALT for the Machine Age service and engagement with research communities.

Annif is open source and was originally designed by Dr. Suominen for the National Library of Finland. As such, it is perfectly suited to the needs of USDA’s National Agricultural Library. Other indexing tools researched are focused on business applications generally without their own vocabulary as NAL has as a major focal point and product, and therefore cannot meet NAL’s requirements.

1. Annif continues to be the foundation of the AI/ML testbed developed in this contract for subject indexing, with training using NALT Core sub-scheme, also a product of this overall project. As more domain specific sub-schemes are added to NALT Concept Space, these will also need to have models developed and tested for indexing an ever-wider variety of content. The testbed results will be compared with the results of NAL’s existing or proposed indexing systems.

a. Annif has the unique capacity to not only store multiple controlled vocabularies, but to also deploy them for subject indexing.

b. Rather than requiring highly specialized computer skills, Annif can provide NAL staff and patrons a user interface allowing for testing state-of-the-art AI and ML approaches to subject indexing with its stored vocabularies.

c. Finally, Annif is exceptional in its ability to allow users to test ‘ensemble’, or combined algorithm approaches with AI.

2. Dr. Suominen is a global leader in understanding and developing vocabulary structures and related tools to manage and visualize them to assess the structure of NALT Concept Space vocabulary and existing workflows for its curation. In addition to Annif, Dr. Suominen is a developer of open-source Skosmos software, which is compatible with the new NALT Concept Space format. Skosmos is a globally recognized vocabulary visualization tool that NAL has recently deployed on the NAL website, and USDA internal websites. NAL has increased the frequency of NALT Concept Space publication to several times a year with a goal of publishing every month.

3. Dr. Suominen has unique experience as a computer scientist with deep knowledge working with a national library, and related development of tools and requirements. For example, developing Finto, a national service for Finland to maintain and publish controlled ontologies or vocabularies, like NALT Concept Space and NALT Concept Space sub-schemes.

4. Dr. Suominen has experience defining metrics and analytical approaches to measure how a national library’s internal and external communities of experts might use, or help maintain specific subsets of vocabularies, like NALT Concept Space and NALT Concept Space sub-schemes.

5. A key component of this work is to define a revolutionary “NALT Core” sub-scheme structure, within the Global Agricultural Concept Space (GACS).

a. Dr. Suominen has extensive firsthand knowledge of GACS as an initial partner, he served in a leadership role along with the three GACS founders, the NAL, the United Nations Food and Agriculture Organization (FAO), and the Center for Agriculture and Bioscience International (CABI).

b. His consultation will be based on the extensive analytical and editorial work jointly completed over a period of four years by the GACS project in the early to mid-2010’s.

Dr. Suominen is uniquely qualified to provide the assessments and recommendations outlined in the SOW. His work has already, and will continue to enhance NALT Concept Space and automated indexing software solutions required by NAL. In addition, NAL will be able to provide new and better resources for NAL Divisions, USDA researchers and other NAL patrons to engage in their own Artificial Intelligence research with these new resources.

Dr. Suominen is the only vendor capable of meeting NAL’s requirements for deploying Annif to help NAL meet requirements for indexing content expansion and quality improvement, and in support of NALT for the Machine Age including NALT Concept Space integration and community engagement. Therefore, abasis for other than full and open competition is warranted. NAL IIB indexing operation is vital to the government’s operation of the NAL and supports many library services the NAL offers. The NAL indexing operation generates approximately $400,000peryear through licensing of indexed content to several large publisher aggregators who provide access to thousands of customers worldwide, primarily college and university libraries. Enhancement of this NAL service could ultimately expand this distribution and support.

10. Certification:

I hereby certify that the information provided above is accurate and complete to the best of my knowledge and belief.