RFI 22.23-36, Fuzzy Data Matching .pdf

Solution for Fuzzy Matching Data (Business Record Linkage)

State of Florida

Department of Revenue - General Tax Administration Request for Information

FDOR Solution for Fuzzy Matching Data (Business Record Linkage)

RFI # 22/23-36

1. Introduction The Florida Department of Revenue (the “Department” or “FDOR”), hereby issues this Request for Information (“RFI”) seeking information from interested parties who can provide services/software to the Department for the purposes of identifying common business and individual entities across different datasets in the absence of common disambiguated identifiers herby hereinafter in this document referred to as “record linkage”.

Record linkage (also known as data matching, entity resolution, and many other terms) is the task of finding records in a data set that refer to the same entity across different data sources.

The Department administers more than 30 taxes and fees, including sales and use tax, reemployment tax, communications services tax, corporate income tax, and fuel tax. The Department also has access and receives large amounts of third-party public, private, and administrative data which contains non-unique elements (such as last name, company name, Doing Business As name, fictious name or street address).

The quantity of business-related databases has grown exponentially over the past few years significantly and presents taxing authorities the opportunities to assimilate information through record-linkage to support its taxing and oversight mission. Record linkage necessity continues to grow in significance as a fundamental activity in tax authoritative agencies.

With the ever-evolving computer science technology agencies must leverage multiple string comparators algorithms to assess the similarity of records using state of the art machine learning algorithms to disambiguate these ‘similar” matching relationships.

The goal of this RFI is to provide the Department with a solution which can be definitively demonstrated to find the probability of records being “similar” between disparate datasets for the express purpose to discover what persons or entities are not registered with the Department that should be; or that identifies/provides proof of unreported taxable activities/unpaid tax through these newly discovered record linkages, for persons or entities registered with the Department.

https://en.wikipedia.org/wiki/Record_(database) https://en.wiktionary.org/wiki/entity

2. Purpose of an RFI

2.1. Rule 60A-1.042, Florida Administrative Code, provides that an agency may request information by issuing a written RFI. Agencies are authorized to use an RFI in circumstances including, but not limited to, determining whether or not to competitively procure a commodity or contractual services, determining what solicitation process to use for a particular need, or researching general, special, and/or technical specifications for a solicitation. A vendor’s response to a RFI is not an offer and the agency may not use the vendor’s submission to justify a contract with that vendor without otherwise complying with Chapter 287, Florida Statutes (F.S.) and Rule 60A-1.042, F.A.C.

Vendors submitting a response to an agency’s RFI are not prohibited from responding to any related subsequent solicitation.

2.2. Any future purchase of the services will be conducted in accordance with Chapter 287, F.S.

2.3. The Department may use responses to this RFI to prepare one or more competitive solicitations.

2.4. The Department may use this RFI as the basis for subsequent vendor meetings and product demonstrations for those firms that meet the requirements specified in Section 3 of this document.

2.5. If demonstrations are requested, then they will be scheduled directly with only those vendors who respond to all questions in section 4 of this document.

2.6. If the Department receives responses from multiple vendors on the same product, the Department will contact the manufacturer of the product and ask them to present the demonstration.

3. Requirements

3.1. Current Process - The Florida Department of Revenue – Lead Development Process receives files from various other agencies or third-party data providers. These files are imported into an internal application and the file records are matched to existing records in the database. Many records contain an FEI or SSN that are easily matched to existing records. Our challenge is when the imported records do not contain an FEI or

SSN and we are forced to perform “fuzzy matching” based on name and address. Our current process:

Third Party Flat File

• Flat files are received via FTP from various organizations outside of the Department. These flat files can be in a variety of formats. They contain data fields such as name, address, FEI and SSN.

Microsoft SSIS and Microsoft SQL Server Database

• Microsoft SSIS is used to open the file, pull the data, and write that data to a

Microsoft SQL Server Database. Current versions are in the 2015-2016 range but will be upgraded at some point in the future.

• SSIS is then used again to pull the data from the database to be address corrected via an SAP Remote Function Call and then update that data in the SQL database.

• SSIS is used for its native Matching and Fuzzy Matching to match on FEI/SSN, Name and Address to other records (from internal DOR sources or other Third- Party sources) all of which reside in the same SQL Server database.

We would like to see improvements in matching accuracy including finding more matches and lowering number of false matches.

3.2. General requirements –

The solution for “record linkage” must include:

• A demonstratable analysis of why the solution/software is superior to others.

• Ability to integrate with Systems Applications and Products in SQL Server.

• Should be easy to use, flexible, scalable, produce extremely accurate matches.

• Ability to process vast amounts of data in short time frames.

• Must be able to link multiple data sets (rather than a single pair) and are ideal in circumstances where no set of data is a superset of another.

• Match probabilities can be adjusted for any uncertainties in the underlying matched data.

• Transparent: the process and the procedure for generating match probabilities are clearly documented and are provided to users

• Ability to Scale: To be set up as ongoing jobs or easily created in short fashion for quick ad hoc runs with little technical expertise.

3.3. Reporting

• Provide the ability to define key metrics

• Ease of use/access/update

• Provide matching statistical reports

3.4. Security

• Must allow access to only authorized users

• Must allow access to create/change/display data based on user authorization

4. Response Content

Please answer each of the following. Please use the corresponding number, e.g. 4.1, 4.2 and then your response.

4.1 Please respond to your ability to meet the requirements (circle yes or no) and provide an explanation regarding how the requirement is met. Do NOT include pricing in your response.

A. Ability to integrate with SQL Server .

Yes No B. Ability to provide demonstratable analysis of why the solution/software is superior to others.

Yes No

C. Ability to provide proof to be easy to use, flexible, scalable and produce extremely accurate matches.

Yes No

D. Ability to provide demonstratable evidence to process vast amounts of data in short time frames.

Yes No

E. Ability to link multiple data sets (rather than a single pair) and are ideal in circumstances where no set of data is a superset of another.

Yes No

F. Ability to provide demonstratable match probabilities can be adjusted for any uncertainties in the underlying matched data.

Yes No

G. Ability to provide transparency: the process and the procedure for generating match probabilities are clearly documented and are provided to users.

Yes No

H. Ability to Scale: To be set up as ongoing jobs or easily created in short fashion for quick ad hoc runs with little technical expertise.

Yes No

I. Ability to provide reports and identify key metrics.

Yes No

J. Reporting is easy to use/access/update.

Yes No

K. System provides security access to only authorized users Yes No

L. System allows access to create/change/display data based on user authorization Yes No

Additional comments:

4.2 Please provide your offerings on implementation, customization, migration, and training for the recommendation.

4.3 Please provide your pricing methodology for your recommendation (i.e., your basis for charging), but do NOT include actual pricing in your response.

4.4 Are you currently on any of the following contracts or any other governmental contract that permits cooperative purchasing: State of Florida State Term Contract with the Florida Department of Management Services, U.S. Communities Contract, GSA Schedule 70 Contract with a cooperative agreement, etc. as described on the DMS web page:

https://www.dms.myflorida.com/business_operations/state_purchasing/state_contracts_and _agreements

If so, please provide the applicable contract number.

5. Written Questions & Responses

Written questions concerning this RFI shall be submitted via email to the Purchasing Agent, Ed Valla, edwin.valla@floridarevenue.com no later than 12:00 p.m. Eastern Time, Friday, February 10, 2023.

Please label your questions RFI #22/23-36 in the subject line.

FDOR will provide written responses to all timely submitted written questions via the Vendor Information Portal on or about Monday, February 13, 2023.

6. Response Submissions All responses are due by Noon Monday, February 27, 2023.

Please send responses via email to Ed Valla, edwin.valla@floridarevenue.com no later than 12:00 p.m. Eastern Time, Friday, February 27, 2023.

Please label your response RFI #22/23-36 in the subject line.

7. Additional Information

7.1. This is a Request for Information (RFI) and should not be construed as intent, commitment or promise to acquire or purchase the proposed services presented by vendors.

7.2. The Florida Department of Revenue will not be obligated to any vendor as a result of this RFI. The Department is not obligated for any cost incurred by vendors in the preparation of their response to this RFI. The Department will not pay for any information herein requested nor be liable for any costs incurred by the vendor.

https://www.dms.myflorida.com/business_operations/state_purchasing/state_contracts_and_agreements https://www.dms.myflorida.com/business_operations/state_purchasing/state_contracts_and_agreements mailto:edwin.valla@floridarevenue.com mailto:edwin.valla@floridarevenue.com

7.3. This RFI is being submitted strictly for the purpose of gaining knowledge on services available on the market for the provision of these services, related services, and options available.

7.4. From the information collected through this RFI, the Department will review all information and options related to the services, related services, and desirable options.

7.5. At a later time, it is anticipated that the Department will release our requirement through the formal bidding process or through a certified alternate contract source.

7.6. All information obtained shall become the property of the Department upon receipt and will not be returned. The Department cannot guarantee that it will not be compelled to disclose all or part of any public record under the Florida Sunshine Law.

7.7. In the RFI, the Department has addressed a series of questions to vendors, and request that vendors reply to the Department in the same sequence and format.

7.8. The Department requests that all vendors submit responses that are short, clear, concise and complete.

7.9. Submitting a response to this RFI does not exclude any vendor from submitting a response to a solicitation as a result of the information collected from this RFI.