Workshop

 

Title

Intelligent solutions for monitoring, case assessment and service provision in the tax domain

Intended audience

Tax administration representatives, legal knowledge engineers, administrative scientist, legal professionals and scholars, political scientists, ICT scientists

Organizer

Prof. dr. Tom M. van Engers (vanEngers@uva.nl)

 

Outline

Tax administrations were among the early adaptors of artificial intelligent (AI) solutions. Many examples of successful AI-based solutions have been introduced since the nineteen eighties of the previous century, addressing a wide variety of usages, including tax assessment, monitoring and risk assessment, service provision with regard to complex fiscal matters etc.

 

This workshop is intended to provide a platform for exchanging experiences and best practices in creating intelligent applications in the Tax domain. We will address some innovations that will help tax administrations and their network partners and clients to enhance their efficiency and effectiveness.

 

Recent examples of intelligent solutions for monitoring, case assessment and service provision will be presented and demonstrated, including a recent intelligent frame-base tax system that helped to increase tax returns for 35%, monitoring internet presence of entrepreneurs, design approaches for increasing tax administrations’ agility etc.

 

Participants are invited to contact the organizer if they have innovations they want to share with the participants to this workshop..   

 

Keywords: Intelligent Services, Monitoring, Risk Assessment, Tax


 

Programme

 

09.00h - 09.20h Registration.

09.20h - 09.30h Opening

Block 1 Smart Services in the Tax Domain

09.30h – 10.15h Right from the start. (Kors Kool)

10.15h – 11.00h A model-driven approach to intelligent client handling support (Jeroen van Grondelle)

11.00h – 11.30h Coffee break

Block 2 Innovations in Tax Assessment

11.30h – 12.15h Innovation in the Hungarian Tax Administration (Iván Futó)

12.15h – 13.00h A Frame-based solution for Tax Administrations (Coos van der Togt)

13.00h – 14.00h Lunch

Block 3 Intelligent Monitoring and Risk Assessment

14.00h – 14.45h Monitoring the Internet (Eric Hertogh)

14.45h – 15.15h Coffee break

15.15h – 16.00h Scenario building and Agent-based diagnosis (Giovanni Sileno & Alexander Boer)

16.00h – 16.45h Information-based diagnosis of non-compliance (Bernd Veldman)

Block 4 Challenges and Future Developments

16.45h – 17.15h Recap A generic approach to intelligent client handling (Robert van Doesburg)

17.15h – 18.30h Drinks

 


Block 1 Smart Services in the Tax Domain

09.30h – 10.15h Right from the start. (Kors Kool)

The Dutch Tax and Customs have developed a new strategy aimed at preventing non-compliance by moving upfront and doing it ‘right from the start’. This approach that is also known as ‘horizontal monitoring’ is based on strengthening cooperation between the tax administration, taxpayers and intermediaries. Embedded compliance and working based upon a trust framework are amongst the strategically chosen measurements.

Kors Kool is the responsible manager for the development and implementation of this new way of working.  He is also co-chair of an international cooperation group on tax compliance.

10.15h – 11.00h A model-driven approach to intelligent client handling support (Jeroen van Grondelle)

In this presentation we share experiences in capturing tax legislation and policy declaratively into models and show some of the intelligent services that can be offered based on these models using AI techniques. The end user experience of such model driven user interfaces has been a challenge: we present an approach for declaratively influencing the user experience. User Interaction designers can create engaging end user dialogs, independent of the contextual content of the dialog, derived from domain models. Jeroen van Grondelle is cofounder of Be Informed and one of the creators of its business process platform. Currently, he is principal architect at Be Informed, responsible for innovation and research. Because of its groundbreaking approach to business processes, it is crucial for Be Informed to cooperate closely with the academic world. Van Grondelle develops relations with universities, research institutes and other innovative enterprises. Through research projects, joint publications and case studies by visiting researchers, he is able to present researchers with real world challenges, experiences and is able to adopt the newest research developments early. Van Grondelle regularly speaks at conferences and seminars. He recently published a number of papers that dealt with enabling business experts to use formal representations in capturing their knowledge. He has a masters degree in mathematics from Utrecht University, specializing in algorithm design and large-scale parallel computing.

Block 2 Innovations in Tax Assessment

11.30h – 12.15h Innovation in the Hungarian Tax Administration (Iván Futó)

The Hungarian Tax and Financial Control Administration (APEH) was the most innovative institution of the public administration at the beginning of the 2000’s. The electronic taxation was the first electronic service of the Hungarian government and the Office was at the same time the Certification Authority (CA), the Registration Authority (RA), the application, signature and time stamp provider, as well.

Despite the fact that already in 2001 the use of knowledge based (expert) systems was part of the IT strategy no such systems were implemented in the following years. 

This presentation deals with the difficulties of introducing knowledge-based applications in a Tax Office or in the Public Administration in general. We will try to explain why it is so, based on the presenter‘s experience with the Hungarian Tax and Customs Administration, where he was the IT vice-president for six years.

It is also interesting to see how in a country with a good former AI tradition, the situation changed in the past 10-15 years regarding the acceptance of knowledge-based systems.

Paradoxically in a situation where basic laws change year by year, or even twice a year, and new tax laws are introduced frequently to raise the incomes of the budget by the end of the year, the resistance against knowledge-based applications that facilitate the implementation of such regulations is so strong.

Some of the potential obstacles discussed are: aversion to unknown technology, nobody wants to be the first, missing authentic ”champions”, the problem of selecting a good pilot application, difficulties in integrating the application into the legacy system, the low level of MP’s IT support.

12.15h – 13.00h A Frame-based solution for Tax Administrations (Coos van der Togt)

Tax Administrations needs to shorten their time to implement changes in tax laws. Local Governmental Administrations seek cooperation in executing local tax laws. Civilians and companies expects services based on their specific situation.

Lower budgets forces administrations to reduce costs. Large projects often fail to meet the expectations. Delays and higher expenses are more common than exception. Of the shelf solutions based on best-practices avoids large development projects.

 

A frame-based solution for Tax Administration gives the answer. Coos van der Togt presents the business approach based on more than 20 years of Tax Administration experience (Dutch Tax Office, Dutch Caribean tax implementation). Understanding the tax business is the necessary for successful create frame-based tax solution in which rules, processes and data are separated for changes and coupled for execution.

Coos shows us his approach which resulted in a generic tax solution based on the Be Informed business platform.

13.00h – 14.00h Lunch

Block 3 Intelligent Monitoring and Risk Assessment

14.00h – 14.45h Monitoring the Internet (Eric Hertogh)

The Internet Service Centre of the Dutch tax administration collects internet data for different purposes and processes, such as processes that deal with services, supervision, investigation and webcare. Different systems and processes are activated to collect this data. Collecting and refining such data comes with not only technical challenges, but also legal and social ones. Eric explains in his presentation how these issues are handled in practise. Eric will introduce various tools that have been developed for monitoring the Internet, including Xenon (by the Dutch Tax administration developed and in use in many European and non-European countries), QWMS (type of sample analysis), Coosto (social media analysis), European News Monitor (news monitor) and the identification analysis of websites.

Eric Hertogh is head of the Internet Service Centre of the Dutch tax authorities. Eric started as a clerk at a local department 27 years ago and later became tax auditor. Eric became project advisor for the Xenon project, an Internet Monitoring system which development started at the mid nineties of the 20th century. The current Xenon tool however has not yet landed in the primary process. Therefore Eric founded the Internet Service Centre. Starting with three people, this center has now grown to 25 people. These people often give training session for both national as well as international Tax agencies. Amonst Eric’s roles are stimulating cooperation between tax authorities.

14.45h – 15.15h Coffee break

15.15h – 16.00h Scenario building and Agent-based diagnosis (Giovanni Sileno & Alexander Boer)

In the Agile project, a cooperation between The Leibniz Center for Law, TU Delft, the Dutch tax administration, and the immigration service, we developed a knowledge acquisition methodology for legal knowledge in service-oriented architecture, developed to address the agility of  public administration in responding to change of the law, and processes of change in society that affect its implementation and interpretation of the law. An important result is to make the case law, and legal expert knowledge of critical legal incidents in organizations, two diagnostic knowledge sources underutilized in modern management and business process engineering, more accessible as a resource for design of social structures and intelligent systems.

Part of the methodology is a conceptual framework for abstract cases of conforming and non-conforming scenarios, collected for animation in an agent-based platform for design purposes, or used as reference base for model-based diagnosis tool. To reduce complexity of the system and its environment, we do not focus on autonomous, arbitrarily complex “intelligence” of involved agents, but instead concentrate attention on story-based descriptions and on institutional agent role abstractions as addressed by court cases and domain experts, that capture the beliefs and intentions ascribed to these story-based or institutional agent roles. The framework has been tested with a number of real estate tax fraud scenarios, supplied by Bernd Veldman.

16.00h – 16.45h Information-based diagnosis of non-compliance (Bernd Veldman)

The Dutch Tax Office has a thematic approach on real estate. Starting point for the Dutch tax administration is the wish to create the best possible support to the taxpayer resulting in correct fiscal statements.  In order to shift the focus to supporting taxpayers rather than assessing every individual one, a more efficient risks-oriented strategy is needed. The tax office works in a way where the costs of auditing are agreeable low and were it becomes possible to accept certain risks that come with real estate transactions.
We combine a micro-theory (top down) based risk analysis with (bottom up) data mining approach, which enables us amongst others to determine the total tax gap in real estate transactions in the Dutch pension fund branche, the social housing corporation branche en transactions between connected people. In this presentation we’ll present a micro-theory based audit approach where we audited a representative set of selected transactions to validate certain hypothesis in expected real estate transactions risks. This combination is called micro-theory based information driven auditing.
Theory driven data mining also plays a role when developing new fiscal laws. The most recent fiscal law on the taxation of real estate for example includes services for possessors of more than 10 pieces of real estate, who will have their tax returns partly pre filled in in advance. Also the choice for the definition of the tax-object and its features has been influenced by existing data collections (information driven tax laws). This makes it possible lower the administrative burden for the taxpayer as well as the cost of auditing for the tax administration.
drs. B.L.A.M. Veldman RA is a forensic accountant and working at the Dutch competence center of real estate. At this moment he is working on a PhD at the University of Amsterdam. His subject is the development of a scientific approach for the creation of knowledge concerning risk-indicators in real estate transactions. In this field it is very important tot work with valid indicators that measure the correct items.

Block 4 Challenges and Future Developments

16.45h – 17.15h An innovative generic approach to intelligent client handling (Robert van Doesburg)

The Dutch Immigration Service recently implemented implemented a rule based Information system for client handling. This system creates a treatment plan for every individual client based on personal en case related information.

The knowledge base needed for this operation is based on the law. The Dutch Immigration Service is developing a generic approach for modelling law based client handling in co-operation with the Dutch Tax Administration.

 

Robert van Doesburg is working on this project and is a project manager for the Direction  Information Systems of the Information Service. He will present a framework for intelligent client handling and its consequences for public organisations.

17.15h – 18.30h Drinks