What are the free expert systems

Expert systems in tax advice using the example of Section 8a (6) KStG

Table of Contents

List of abbreviations

List of figures

List of tables

1 Introduction
1.1. Development of an expert system
1.2. Action

2. Shareholder outside financing
2.1. Reasons for the law
2.2. Legal development on shareholder debt financing
2.3. Legal development of § 8a KStG
2.4. Legal wording of § 8a Para. 1-5 KStG
2.4.1. Section 8a (1) of the KStG
2.4.2. Section 8a (2) KStG
2.4.3. Section 8a (3) KStG
2.4.4. Section 8a (4) of the KStG
2.4.5. Section 8a (5) KStG
2.5. Legal consequence vGA
2.5.1. At the level of the borrowed corporation
2.5.2. At the level of the lending shareholder, related party, third party entitled to recourse

3. External shareholder financing in the group of companies
3.1. Reasons for the law
3.2. Intra-group acquisition of shares
3.2.1. Seller of the stake and lender
3.2.1.1. Shareholders
3.2.1.2. Related person
3.2.1.3. Third party entitled to recourse
3.2.2. Subsequent change in the quality of the seller of the stake
3.2.3. Subsequent change in the quality of the lender with ongoing financing ..
3.3. Occurrence context
3.3.1. Finality of the loan amount
3.3.2. Indirect acquisition of shares
3.3.3. Change in funding context
3.3.4. Mixed loan as a special case
3.4. Acquisition of the participation by a partnership as a special case
3.5. Temporal application
3.6. Legal consequences

4. Expert systems
4.1. Definitions and handling of knowledge
4.2. Expert system architecture
4.2.1. Knowledge base
4.2.2. Inference component
4.2.3. Dialogue interface: knowledge acquisition and interview component
4.2.4. Explanatory component
4.3. Involved people
4.3.1. The expert
4.3.2. The knowledge engineer
4.3.3. The user
4.4. Existing expert systems and shells: an overview.
4.5. System of choice: d3web

5. Development of an expert system
5.1. Selection of a suitable subject
5.2. Selection of experts
5.3. Methodical development of expert systems
5.3.1. Phase models
5.3.2. Prototyping

6. Expert system for Section 8a (6) KStG: The development environment

7. Summary outlook

attachment

bibliography

List of legal sources

Register of jurisdictions

Other sources

List of abbreviations

Figure not included in this excerpt

List of figures

Figure 1: Relevant equity within the meaning of Section 8a (2) KStG for calculating the safe haven within the meaning of Section 8a (1) No. 2 KStG.

Figure 2: Downstream partnership 11

Figure 3: Basic case of Section 8a (6) KStG

Figure 4: Intermediate partnership, case group 1.

Figure 5: The architecture of an expert system and representation of the dependencies between the people involved

Figure 6: Example of an uncertain rule represented with the help of object-attribute-value triples

Figure 7: Knowledge engineering process

Figure 8: Screenshot of the development environment

Figure 9: Screenshot of the knowledge base ("rule overview")

Figure 10: Screenshot of the inference control ("decision tree")

Figure 11: Screenshot of the dialog interface

Figure 12: Example of the “calculation” of indirect participation rates

Figure 13: Dependence of the need for a knowledge engineer on the size of the expert system

Figure 14: Employee typification according to Ritti: local or cosmopolitan

Figure 15: The dependency of the motivation plan on the classification and typing of the "unwilling expert"

Figure 16: The rule editor

Figure 17: Editing heuristic rules taking probabilities into account

List of tables

Table 1: Examples of object-attribute-value triples with terms from tax theory

Table 2: Evolution of an expert system

1 Introduction

In many areas of business administration, the search is on for possible applications for expert systems,1 because they are assigned a lot of potential. It is predicted that these systems will become increasingly important, especially in the banking / insurance and corporate / tax consultancy sectors.2

Any software that generates solutions based on expert knowledge for specific, precisely delimited problems is referred to as an expert system.3 Expert systems are originally based on research in the field of artificial intelligence and the general problem solver4 back. They are particularly suitable for solving problems that are characterized by high complexity, uncertainty or incomplete information.5

Their use is primarily intended to promote productivity, preserve knowledge and generally contribute to transferring information into structured knowledge.6

1.1. Development of an expert system

The creation of an expert system begins with the selection of a suitable expert system shell7.8 Once you have decided on an expert system shell, you need to create the knowledge base. This process is specifically dependent on the type of knowledge representation of the selected expert system shell. In general, the “hard” rule knowledge of a human expert must first be brought into digital form; this is usually not very problematic. The mapping of "soft" case knowledge, e.g. the knowledge of the interpretation of norms, is difficult, however, in some cases it is even impossible9. However, it is crucial for the accuracy of the expert system.

1.2. Action

The present work is intended to demonstrate the creation of an expert system in tax consulting using the example of Section 8a (6) KStG. However, due to the large number of articles published on this topic, an extremely detailed analysis of Section 8a (6) KStG is not carried out.10 The fundamentals of the legal regulation of the external financing of shareholders (§ 8a Para. 1-5 KStG), as well as the external financing of shareholders in the group of companies (§ 8a Abs. 6 KStG) are presented. The ongoing discussion on the EC and DBA compatibility of the new legal formulation is not discussed in more detail.11

The process at hand is based on the phase model for building an expert system: problem analysis, requirements definition, rough draft, detailed draft, implementation, test.12

In the further course of this work it is necessary to distinguish two strands of thought:

1. the technical processing and systematisation of the standard content and the legal consequences of Section 8a (6) KStG and

2. The definition of expert systems and the elaboration of an expert system for Section 8a (6) of the KStG, as well as the solution of the questions of practical implementation that are known in advance and that arise in the course of the investigation.

The explanations on the technical basis are based both in their process and in their elaboration on the later implementation in the expert system to be created. This means that the wording is as concise as possible and that the in-depth analysis of the discussion in the literature on individual questions of doubt is omitted, because the knowledge resulting from such an analysis cannot be taken into account in the expert system prototype to be created .

2. Shareholder outside financing

A corporation can be provided by its shareholders with equity or debt capital (including shareholder loans).13 There is a fundamental freedom of financing.14 As business expenses, remuneration for borrowed capital reduces economic efficiency and thus taxable income.15 However, remuneration for equity is allocated to the sphere of the use of income and therefore does not reduce the taxable income.16 As a result, shareholders could be tempted to suck up profits by means of shareholder loans.17

2.1. Reasons for the law

According to the legislator, the regulation of § 8a KStG should "[limit] the shareholder external financing in corporations with unlimited tax liability"18, whereby it should be avoided "that profits of corporations operating in Germany are withdrawn from German taxation."19

Expenses for outside capital are accordingly only within the scope of the exemption limit granted in Section 8a (1) of the KStG20 removable. If the expenses exceed this limit, they will be treated in full as VGA.21 The term vGA is not defined in the tax law, but the tax authorities have adopted the definition of the BFH in the KStR 2004: “Eine vGA i. See Section 8 (3) sentence 2 of the KStG is a reduction in assets or a prevented increase in assets, which is caused by the corporate relationship, ... and is not based on a profit distribution resolution in accordance with company law.22

2.2. Legal development on shareholder debt financing

From this the question can be derived for the legislature of how the attempt to circumvent or minimize income and corporation tax through undercapitalization can be countered.23 For example, a government draft from 1979 provided for the one-off taxation of profit-related remuneration for loaned capital in Germany by reclassifying interest payments to a shareholder not entitled to credit in vGA. In addition to this draft law, the Federal Council draft 1979, the ministerial drafts 1982, 1986 and the government draft 1988 failed. With the BMF letter of 16.3.1987 the term “hidden nominal capital” was introduced, with the aim of limiting the partner's debt financing potential. if the addition of equity would have been mandatory. In a judgment of February 5, 1992, the BFH denied the necessary legal basis for the relevant administrative instruction.24

2.3. Legal development of § 8a KStG

With § 8a KStG (old version) it was intended, for the first time, to prevent non-resident taxpayers from undermining the taxable income of German subsidiaries through a high debt financing rate and converting actually taxable profit distributions into deductible interest expenses.25

The problematic arrangements - from the point of view of the financial administration - were first introduced by a law to restrict tax-harmless shareholder external financing in the form of § 8a in the current version of the StandOG26 contained. It essentially mirrored the wording of the working group's report Hearty and Debatin27 contrary.28

With the StSenkG of July 14th, 200029 the content of the regulations has tightened in essential points. So the safe haven became30 for non-performance-related remuneration reduced from 3: 1 to 1.5: 1. With the entry into force of the so-called Korb II law31 the special regulation of a safe haven of 3: 1 for holding companies that still exists in Section 8a (4) sentence 1 HS 2 KStG in the version of the StSenkG has been deleted.32 The safe haven was also no longer available for income-related interest-bearing borrowed capital.33

On the occasion of the redesign in accordance with European law34 of § 8a KStG, the legislator has with § 8a Abs. 6 KStG35 a ban on financing costs for the acquisition of outside-financed intragroup shareholdings was introduced, with the intention of preventing arrangements which, within the framework of holding constructions, are intended to improve equity through tax-free share sales in accordance with Section 8b (2) KStG, in order to maintain the safe haven to increase.36 Section 8a (6) KStG new version stands “as a special regulation for the sale of shares within the group… full reclassification of remuneration in VGA (ie no exemption limit, no safe haven, no possibility of third-party comparison), if the FK is willing to acquire a capital Participation serves and the seller and FK-giver is an essentially involved AE, a related person within the meaning of Section 1 (2) AStG or a third party entitled to recourse. "37 as a foreign body38 in § 8a KStG new version Due to the non-granted safe haven, exemption limit or the possibility of a third-party comparison, § 8a (6) KStG fundamentally interferes with the freedom of financing.39

2.4. Legal wording of § 8a Para. 1-5 KStG

The following explanations are intended to classify the later, more comprehensive analysis of Section 8a (6) KStG in the overall context of Section 8a KStG, and are therefore limited in scope. In the literature there are more extensive preparations of the legal content of § 8a Paragraph 1-5 KStG.40

Section 8a of the KStG is linked to the principle of the irrelevance of the appropriation of profits for the determination of the tax base, as specified in Section 8 (3) KStG for corporation tax.41

2.4.1. Section 8a (1) of the KStG

Section 8a (1) of the KStG reflects the elementary core of the standard. For example, remuneration for loaned capital is vGA if:42

- the outside capital is not only made available for a short period of time, i.e. 12 months,
- the remuneration the exemption limit in the amount of Exceed EUR 250,000,
- the borrower is a corporation, and
- the lender is a significant shareholder at one point in the financial year.

Section 8a (1) sentence 1 no. 1, 2 KStG regulates the reinterpretation of income-related (no. 1) and income-independent (no. 2) remuneration, respectively. No safe haven is granted for income-related debt capital.43 In the case of income-independent capital, i.e. debt to be paid at a fixed interest rate, a safe haven in the amount of granted by 1.5 times the proportionate equity.44

An expansion of the personal scope of the lender in Section 8a, Paragraph 1, Sentence 2 of the KStG is intended to limit the scope for circumventing the norm. In addition to the above-described shareholders,

- Persons closely related to the shareholder within the meaning of Section 1 (2) AStG and
- third parties entitled to recourse

recorded as a harmful lender.45

2.4.2. Section 8a (2) KStG

The proportionate equity serves as the basis for determining the safe haven. How the equity of the corporation is calculated and how it is assigned to the shareholder in accordance with his participation in the subscribed capital is defined in Section 8a (2) of the KStG. The equity calculation is shown in Figure 1.46

Figure 1: Relevant equity within the meaning of Section 8a (2) KStG for calculating the safe haven within the meaning of Section 8a (1) No. 2 KStG.

Figure not included in this excerpt

Source: Pung / Dötsch in: Dötsch / Eversberg / Jost et al., KStG, § 8a KStG new version, margin no. 323.

2.4.3. Section 8a (3) of the KStG

In § 8a para. 1 KStG, the substantial participation of the loan-granting shareholder in the borrowing corporation is a prerequisite for the reinterpretation of debt remuneration in VGA. The concept of substantial participation is in § 8a para. 3 KStG for the Person of the shareholder regulated.47 A participation is essential within the meaning of Section 8a (1) KStG if the shareholder holds more than 25% of the share in the corporation, either directly or indirectly.48 In the case of indirect participations through a corporation or partnership, the materiality of each individual participation is not important; the “calculated” participation rate is then considered.49

2.4.4. Section 8a (4) of the KStG

The main activity of a corporation is holding shares in corporations, while at the same time50 Financing of these corporations, this corporation is considered a holding company (qualitative qualification)51.52 Complementary53 For this qualification, a corporation can also be considered a holding company if its holdings in corporations account for more than 75% of its total assets (quantitative qualification)54.55 The holding privilege is codified in Section 8a (4) sentence 1 HS 2 KStG. It prevents the equity of the holding parent from being reduced by the book values ​​of the holdings. Such a reduction would regularly lead to cf. low equity as the basis for calculating the safe haven and making the safe haven practically meaningless.56 For this reason, the special holding regulation within shareholder external financing is of considerable importance for Germany as a holding location.57

2.4.5. Section 8a (5) KStG

The old version of Section 8a KStG was prone to abuse and made it easy to circumvent the content of Section 8a KStG.58 By interposing a partnership, as shown in Figure 2, the old version of Section 8a (1-4) of the KStG could easily be overridden. Section 8a (5) KStG (new version) extends the scope of application of section 8a (1-4) KStG (new version) to all cases in which “the debt capital is left to a partnership in which the corporation alone or together with persons closely related to it Within the meaning of Section 1 (2) of the External Tax Act is directly or indirectly involved in more than a quarter. "59 A loan is provided between the lender and the corporation.60Dötsch / Pung identify some of the questions arising from this fiction61which would be too extensive to reproduce at this point.

Figure 2: Downstream partnership.

Figure not included in this excerpt

Source: Dötsch / Pung, DB 2004, 91, 98.

2.5. Legal consequence vGA

In § 8a KStG (old version), remuneration for outside capital such as vGA was dealt with.The means for this treatment of the interest expenses was the fiction of the VGA in the sense of a legal consequence reference to § 8 Abs. 3 S. 2 KStG. As a result, the debt capital payments within the meaning of Section 8a KStG old version were not a VGA, but merely triggered the legal consequences of a VGA.62 With § 8a KStG new version, the legislature changed the legal consequence reference into a legal reference.63 Shareholder loan capital payments are no longer just a hidden profit distribution, they are now too. As a result, in addition to the factual requirements of Section 8a KStG new version, the general requirements of a VGA64 be fulfilled. In addition, this innovation justifies the applicability of all legal consequences of a vGA for all persons involved in the matter.65

2.5.1. At the level of the borrowed capital company

Remuneration for outside capital that qualifies as a vGA does not reduce the profit. The VGA deducted as operating expenses is added to the balance sheet. With the elimination of Section 9 No. 10 GewStG (old version), these remuneration also do not reduce the commercial income66.67 The corporation may have to pay a corporation tax increase in accordance with § 38 KStG.68 If the capital gains tax liability for VGA was still controversial for Section 8a of the old version of the KStG, the remuneration is now subject to capital gains tax.69

The loan itself remains as debt for tax purposes and is not reclassified as equity.70

2.5.2. At the level of the lending shareholder, related party, third party entitled to recourse

The legal consequences of Section 8a KStG for shareholders, related parties and third parties entitled to recourse are largely disputed.71

In the case of natural persons, payments received for loans in accordance with Section 3 No. 40 EStG are halved72 taxed as investment income within the meaning of Section 20 Paragraph 1 No. 1 Sentence 2 EStG.

For corporations, pursuant to Section 8b (5) sentence 1 in conjunction with Section 8b (1) KStG, there is a prohibition on deduction in the amount of 5%. This means that 95% of the remuneration for loans that a corporation receives remains tax-free.73 A prohibition on deducting expenses related to the loan is explicitly excluded from Section 8b (5) sentence 2 of the KStG.

3. External shareholder financing in the group of companies

With judgment in the Lankhorst-Hohorst case74 § 8a KStG was revised. For the justification of the law, among other things, a75 Redesign, the aggravation of design abuses through the interposition of partnerships76, the prevention of profit extraction by corporations operating domestically by foreign parent companies and, in general, the lower susceptibility to structuring and abuse of § 8a KStG new version.77

Section 8a (6) of the KStG constitutes a special regulation78 for cases of shareholder external financing in the group79. In order for Section 8a (6) of the KStG to apply, the following criteria must be met cumulatively:80

- The borrowed capital is raised to acquire shares in a corporation.
- The seller of the participation as well as the lender must - in relation to the acquiring corporation81 - belong to the group of people of Section 8a, Paragraph 1, Clause 1, 2 KStG, i.e. either a shareholder, a person close to the shareholder, or a third party entitled to recourse.

If the participation is acquired by an intermediary partnership, Section 8a (6) of the KStG applies if the corporation - possibly also together with related parties - holds more than 25% of the partnership.82

The basic case of Section 8a (6) KStG is shown in Figure 3. The parent company (M-AG) finances the acquisition of the T2 stake by T1-GmbH with a loan to T1-GmbH. The resulting interest, which flows from T1-GmbH to M-AG, will be reclassified to vGA in accordance with Section 8a (6) KStG: The debt borrowed by T1-GmbH is used to acquire shares in T2. M-AG is both the seller of the (T2) stake and the lender (to T1). At the same time, M-AG, based on the acquiring corporation (here T1-GmbH), is a shareholder within the meaning of Section 8a, Paragraph 6, Sentence 1, No. 2 KStG. The re-qualification of the interest in vGA affects T1-GmbH in the prohibition of deductions (i.e. off-balance-sheet addition of interest payments to the company profit), with M-AG 95% of the received interest payments are tax-free according to § 8b Abs. 5 S. 1 in conjunction with Section 8b (1) KStG. Since in the following the terms shareholder (or the person closely related to him or third party entitled to recourse), acquirer of shares (or acquiring company or similar) and seller of shares (or seller or similar) are used frequently and without further description For illustration purposes, they are specified in Figure 3 on the basis of the companies involved.

Figure 3: Basic case of Section 8a (6) KStG.

Figure not included in this excerpt

Source: Own illustration based on Bindl, DStR 2005, 1673, 1673.

3.1. Reasons for the law

Section 8a (6) of the KStG83 was enacted with a view to undesirable arrangements in the case of internal, externally financed share sales. The aim is to prevent structures that are tax-exempt under Section 8b (2) of the KStG from the sale of shares within holding structures84 the improvement of equity and, as a result, of the harmless proportion of borrowed capital.85 This increase in equity takes place at the seller of the investment.86 However, the purchaser is harmed by Section 8a (6) of the KStG, who is prohibited from deducting the interest on borrowed capital for the borrowed capital taken out for the purpose of acquiring a stake87, regardless of whether the safe have increased through the acquisition of shares88. This makes it clear that Section 8a (6) of the KStG fails to meet the legislator's objectives. Even the attempt to correct this with a very broad term of abuse does not lead to the goal.89

Particularly in the case of cross-border intra-group acquisitions, there are doubts about the compatibility of Section 8a (6) of the KStG with Article 9 (1) of the OECD-MA.90 A domestic profit adjustment for affiliated companies through application of Section 8a (6) KStG, and the associated reclassification of interest expenses to VGA, is only permissible according to Article 9 (1) OECD-MA if it is based on agreements which cannot withstand an arm's length comparison.91 If the financing of the acquisition of shares in accordance with Section 8a (6) of the KStG is "unconventional, there is likely to be a violation of Article 9 (1) of the OECD-MA"92.

The "rather vaguely formulated factual prerequisites"93 of Section 8a (6) KStG lead to questions of interpretation and doubts. For the purposes of developing an expert system, the analysis of Section 8a (6) KStG is therefore reduced to the points that are relevant in the basic case.

3.2. Intra-group acquisition of shares

In order for the special regulation of § 8a (6) KStG to apply in the case of shareholder external financing, the acquisition of shares must relate to shares belonging to the group (ie the share sold by the seller must correspond to the share financed by the buyer) and financed internally be, maW the seller of the stake or the lender must be interconnected persons in the sense of belonging to a group94 be. The only deviation from this is the case of the third party entitled to recourse.95

The question of the need for a personal identity between the seller and the lender is answered differently in the literature. With a view to the desired effect of the standard, however, the majority of the question answered in the negative.96 If the answer is yes, the rule would only cover the trivial case and could easily be circumvented.97 The BMF draft on Section 8a (6) of the KStG also follows this view.98

According to the majority opinion, the receipt of a participation in the form of a hidden contribution is not recorded as an intra-group acquisition of shares within the meaning of Section 8a (6) of the KStG. By investing shares in a subsidiary instead of selling them, an improvement in equity is achieved that is not covered by Section 8a (6) of the KStG.99

3.2.1. Seller of the stake and lender

In the following, the individual terms shareholder, related person and third party entitled to recourse are explained with a view to the special features of Section 8a (6) KStG.

3.2.1.1. Shareholders

In the opinion of the tax authorities, a shareholder within the meaning of Section 8a KStG is any natural or legal person who has a significant interest in the corporation directly - i.e. without intervening other persons.100 However, there are differing views on this in the literature.101 The domicile of the shareholder in Germany or abroad is not relevant.102 Shareholders can also be a person who only holds a substantial interest in the corporation for a short period of time.103 It should also be noted that the personal scope of Section 8a (6) KStG is not restricted to shareholders who de facto increase their safe haven through the sale of shares.104 It should be noted that cases are also recorded in which the participation was acquired before the introduction of Section 8a or Section 8a (6) of the KStG, i.e. the position as a shareholder was established.105

3.2.1.2. Related person

Section 8a (6) sentence 1 no. 2 KStG refers to Section 1 (2) AStG for the definition of the person closely related to the shareholder. The connection is not entirely clear; the term “taxpayer”, to which the term “related person” refers, must be replaced by the term “shareholder” within the meaning of § 8a KStG.106 The terms “participation” and “controlling influence” are used. This means that natural persons cannot regularly be considered to be closely related to the shareholder, since participation in a natural person cannot exist.107

3.2.1.3. Third party entitled to recourse

A third party within the meaning of Section 8a (1) sentence 2 KStG can be any natural or legal person who is not already a shareholder or a person closely related to him within the meaning of Section 8a (6) sentence 1 No. 2 KStG. A third party entitled to recourse as the seller of the stake does not constitute a harmful intra-group acquisition of shares.108 Regularly there will be a right of recourse109 act on a credit institution equipped with the shareholder or a person closely related to him110which finances the acquisition of shares;111 but it can also be only insignificant112 participating shareholders are deemed to be third parties.113 The possibility of recourse must be documented114 be:115

- through a legal claim, such as a guarantee, a letter of comfort or a surety
- through real security, such as a land charge, security property or a lien.

This makes it clear that there is the possibility of triggering unforeseen VGAs using the concept of a third party entitled to recourse. Particularly problematic, because in practice, the following facts are often found:116

- Damaging purchase of shares with financing by a bank, which becomes a "third party" via the broad definition of the term.
- Sale of shares in which the selling third party finances the purchase price externally and has it secured by the shareholder or a person close to him.

Including the person of the third party entitled to recourse in the group of persons within the meaning of Section 8a (6) of the KStG cannot be justified on the basis of the reasons for the law. In the course of the teleological reduction, only the shareholder or a person close to him should be considered as the seller of the stake.117

3.2.2. Subsequent change in the quality of the seller of the stake

At the time of the sale of the shares, the seller must belong to the aforementioned harmful group of people within the meaning of Section 8a (6) of the KStG.118 If the seller falls out of the group of people at a later point in time, the legal consequences of Section 8a (6) KStG continue to exist.119 If, on the other hand, the seller of the stake does not become part of the group until after the acquisition, this should be detrimental according to the administrative opinion, insofar as the matter follows an overall plan. For a criticism of this view, see Chapter 3.3.2.

3.2.3. Subsequent change in the quality of the lender with ongoing financing

If the lender has held the position as a shareholder or a person closely related to him for at least one point in time in the financial year, this is sufficient to belong to the above-mentioned harmful group of people within the meaning of Section 8a (6) KStG.120 As a result of this fact, debt financing originally outside the group could turn into damaging internal debt financing if the lender acquires a significant stake in the acquiring company in a later financial year.121 To what extent this is sufficient to establish a causal connection is questionable. At least a close temporal relationship between the raising of outside capital and the acquisition of shares within the Group is necessary in order to suspect such a connection.122 From the consideration of the selected tenses in the legal text123 shut down Wood apples / Köplinthat the position of the persons involved at the time of the interest expenses must meet the requirements of Section 8a (6) sentence 1 no. 2 KStG. If the quality of the lender changes over time, this justifies or ends a reclassification of interest expenses in vGA.124

3.3. Occurrence context

3.3.1. Finality of the loan amount

According to Section 8a, Paragraph 6, Sentence 1, No. 1, the borrowed capital must serve the purpose of acquiring a stake in the share capital of a corporation.125 According to the prevailing opinion, it depends on the "specific purpose at the time the funds are raised"126 at.127Wood apples / Köplin call for a "final connection between the acquisition of a stake and the taking out of a loan"128, whereby the context must be an economic one. The BMF draft regards the cause as given if "the loaned capital is actually used to acquire an equity stake"129. In the case of multi-level acquisition structures in the sense of a "passed on" loan within the group130 It can be concluded from this that not every loan issued in this chain is “infected”, but only the acquisition at the end or the loan issued for the purpose of acquiring shares is detrimental within the meaning of Section 8a (6) of the KStG.131

In addition to the economic context, there must also be a temporal relationship.132 In the BMF draft, for a "period of less than one year"133 assume a detrimental temporal relationship between taking out a loan and acquiring a stake.

Who is responsible for proving or refuting a possible connection between taking out a loan and acquiring a share? In practice, for reasons of practicability, a "rebuttable presumption"134 to be going out.135 Likewise the majority opinion,136 according to which the duty to prove lies with the tax authorities, since "the wording of the law does not impose any additional obligations to provide evidence on the taxpayer"137.

3.3.2. Indirect acquisition of shares

If no direct causal connection is recognizable, for example in the case of an only indirect acquisition of shares, the tax authorities will justify the causal connection by means of the so-called overall plan. Then there would be a "harmful causal connection (...) even if the transfer of the borrowed capital took place according to the recognizable will of the parties involved with the aim of indirectly acquiring a capital stake"138. In the context of the overall plan, it is irrelevant whether the capital is provided in the form of equity or debt.139 However, this opinion of the tax authorities cannot be followed with a literal interpretation of Section 8a (6) No. 1 KStG140, since the standard expressly only covers the transfer of outside capital.141 Furthermore, the borrowed capital is not taken out directly for the purpose of acquiring a stake; as a result, no harmful causal connection is established.142

3.3.3. Change in the funding context

The past tense “was borrowed” in Section 8a, Paragraph 6, Sentence 1, No. 1 indicates that the purpose of borrowing must be fulfilled at the time of this.143 Subsequent changes in the facts, such as the sale of the participation and possibly the associated reallocation of the borrowed capital, have no effect on the original legal effect of Section 8a (6) KStG.144. However, the continuing harmful connection between taking out a loan and, for example, a shareholding that has been sold can be broken simply by repaying the harmful loan and taking up new outside capital145, insofar as the new borrowed capital does not again meet the requirements of Section 8a (6) KStG and the legal consequence of the reclassification of interest expenses in VGA occurs again.146

The question of what effect a later reallocation of borrowed capital will have can be answered in a similar manner. Since the application requirements of Section 8a (6) sentence 1 of the KStG must be cumulatively met at the time of borrowing, the legal consequences of Section 8a (6) of the KStG do not apply if one or both of the prerequisites at a later point in time - in other words post-acquisition - are fulfilled.147

3.3.4. Mixed loan as a special case

If the borrowed capital exceeds the amount used for the detrimental acquisition of shares, the pro-rata "loan part and the corresponding FK remuneration associated with the acquisition of shares" must be deducted.148 and to be reclassified as a VGA within the framework of Section 8a (6) KStG, insofar as the provision of outside capital meets its requirements149. If, in addition to a harmful acquisition of shares, a loan is also used to finance material goods, for example150, the interest expense for said borrowed capital is allocated proportionally to the amount used to finance the harmful acquisition of the stake.151

3.4. Acquisition of the participation by a partnership as a special case

A tax-free increase in equity within the meaning of the legal justification for Section 8a (6) KStG can also be achieved by selling shares in partnerships.152 Therefore, Section 8a (6) sentence 2 KStG extends the personal scope of application of Section 8a (6) sentence 1 KStG to acquiring partnerships if the corporation:153

- in the partnership alone or together with persons closely related to it
- has a direct or indirect interest of more than 25%.

They are therefore recorded - analogous to Section 8a (5) KStG154 - Partnerships downstream from the corporation.155Wood apples / Köplin distinguish between four case groups, two of which are to be presented here:156

- Case group 1: The partnership finances the acquisition of shares with equity. Version A: The corporation is a co-entrepreneur in the partnership and finances the acquisition of shares from outside capital within the meaning of Section 8a, Paragraph 6, Sentence 1, No. 1 of the Corporation Tax Act. Variation B: The partnership acquires an in-house share that is self-financed. Subsequently, the partnership in the partnership is established with outside financing by another corporation belonging to the group.

Figure 4: Intermediate partnership, case group 1.

Figure not included in this excerpt

Source: Holzaepfel / Köplin in: Erle / Sauter, Heidelberger Commentary on external financing by shareholders, § 8a KStG. Margin no. 866f.

Dötsch / Pung, among others, has doubts whether the constellation of variant B falls under Section 8a (6) sentence 2 KStG.157

- Case group 2: The partnership finances the acquisition of shares by the shareholder, a related person or a third party with recourse. The corporation does not raise any outside capital. The borrowed capital provided for the purchase of shares is then deemed to have been provided to the corporation in accordance with Section 8a (6) sentence 3 of the KStG158which leads to a vGA at the corporation level159. By fiction160 In the case of the transfer of capital to the corporation, interest expenses are added in the case of the partnership; there is a de facto prohibition of deduction for interest expenses.161

The above-mentioned increase in safe haven can also be achieved by selling a stake in a foreign partnership.162 The question arises to what extent Section 8a, Paragraph 6, Clause 2, 3 of the KStG is suitable for penetration by foreign partnerships.163Wood apples / Köplin explain that a teleological reduction would be necessary here.164 Insofar as interest expenses within the meaning of Section 8a (6) KStG are incurred by the foreign partnership, the foreign regulations apply to the deduction as business expenses.165 In the event that the fiction of attribution within the meaning of Section 8a (6) sentence 3 of the KStG fakes foreign operating expenses as incurred in Germany, taxing these company building expenses, which are then re-qualified as VGA, would be equivalent to taxing foreign tax volumes.166

3.5. Temporal application

The special regulation of Section 8a (6) KStG applies for the first time to financial years beginning after December 31, 2003167if the financial year of the debt-financed share purchaser corresponds to the calendar year168. In the event of a financial year deviating from the calendar year, Section 8a (6) of the KStG applies to the 2004/2005 financial year.169 Such intra-group sales of shares that took place before December 31, 2003 are also recorded if interest expenses for the loan are for periods after December 31, 2003.170Grains sees it as a violation of the "constitutional prohibition of retroactive effect"171. According to a majority opinion in the literature, however, it is a bogus, constitutionally unproblematic retroactive effect.172

3.6. Legal consequences

The legal consequences of Section 8a (6) KStG differ from those of the basic case within the meaning of Section 8a (1) KStG173 by their sharpness. In contrast to the basic case174

- no exemption limit is granted (in the basic case: exemption limit of € 250,000),
- there is no harmless equity / debt capital ratio (safe have) and
- There is no way to avert the legal consequences through a third party settlement.

For further legal consequences see Chapter 2.5 and Wood apples / Köplin175.

4. Expert systems

4.1. Definitions and handling of knowledge

Expert systems come from the research area of ​​artificial intelligence176 emerged. This is a software class that depicts the problem-solving behavior of human experts by finding a solution to a specific problem through interaction with the user177. Due to the developments in the field of artificial intelligence, an expert system relates to a narrowly defined subject area. Artificial intelligence, and thus expert systems as an application of artificial intelligence, are after Rich "The study of how to make computer do things at which, for the moment, people are better."178

So-called expert system shells are used to develop expert systems. These are empty expert systems, consisting of an inference machine suitable (for the respective project), a dialog interface, an explanatory component and an empty knowledge base.179 The expert system shell is therefore not equipped with knowledge, but with predetermined types of knowledge representation and an application to add rules and knowledge to the knowledge base.180

Expert systems differ from standard software in one decisive point181. To run standard software, all of the data requested by the program must be entered in advance. Expert systems, on the other hand, can also answer questions for which not all of the requested information has (can) be entered in full.182 So you can act under uncertainty, as it largely corresponds to the behavior of a human expert.

Since expert systems belong to the knowledge-based systems183, the handling of knowledge plays an important role in explaining the theoretical concepts relevant to the development of expert systems. For a better understanding of the profound problems in dealing with knowledge in expert system development, these are discussed in the respective sections in a discourse.

The challenge is primarily to gain the expert's knowledge with the help of suitable tools: "Knowledge elicitation, which implies obtaining knowledge by means of interaction of analyst and domain expert with the aim of building the subject area model and revealing expert reasoning strategies, remains a bottleneck in designing the intelligence systems. "184 In order to overcome this limitation, several methods for acquiring knowledge are available in the literature.185

Expert knowledge can be divided into two basic types of knowledge: explicit and implicit knowledge. While explicit knowledge can be formalized relatively easily by the expert as knowledge of facts, there are some special features to consider when gaining implicit - i.e. not or difficult to articulate - knowledge.186

- The problem-solving ability of an expert does not necessarily result in the ability to formalize the problem-solving strategy. Consequently, such tacit knowledge cannot be obtained by interviewing the experts.
- The ability to formalize the problem-solving strategy decreases to the same extent as the scope of the problem-solving strategy increases.187

The acquisition of tacit knowledge is therefore made more difficult, together with the inability to formalize by the expert, due to a lack of understanding of the relevance of individual information. When solving problems, experts take note of information and process it further in some cases without being able to perceive these processes and formalize them accordingly.

When defining the term expert system precisely, it is helpful to break it down into two parts. In the first part, the typical architecture of an expert system is presented. The group of people involved in an expert system is then explained. This is especially necessary with a view to a clear understanding of the distinction between the term expert and that of the knowledge engineer.

4.2. Expert system architecture

Since no generally recognized definition of the term software architecture has emerged in the literature, a very catchy explanation of the term is given here: "A software architecture is a structured or hierarchical arrangement of system components and a description of their relationships."188

The relationships between the components of an expert system described below are presented in Figure 5.189

Figure 5: The architecture of an expert system and representation of the dependencies between the people involved.

Figure not included in this excerpt

Source: Own illustration based on Doll, 1991, 13; Harmon / King, 1989, 40.

Before the architecture (points 1 - 6) is explained, the relationships between the people involved in the development of an expert system apply190, namely to the expert and the knowledge engineer. The knowledge engineer acquires knowledge (facts) and problem-solving strategies in the form of rules (including heuristics) from the expert (point A). The knowledge engineer asks the expert for questions, problems of understanding and necessary but not yet delivered knowledge (point B). If the comparison between experts and knowledge engineers has taken place, the knowledge engineer transfers the collected expert knowledge as well as the problem-solving strategies in formalized, coded form via the knowledge acquisition component into the knowledge base (point C). This clearly shows the interface function of the knowledge engineer between experts and the expert system. Points 1 - 6 represent the dependency relationships between the individual components of an expert system, which together describe the architecture. At the same time, an attempt is made to show a loosely chronological sequence of relationships when using the expert system. Double arrows indicate frequent, opposing data throughput, which is triggered by just one user action. M.a.W. the interview, explanation, inference components and the knowledge base are closely linked. If the expert system is used, the user gives data191 via the interview component (point 1). The data is forwarded from the interview component to the inference component (point 2). The relevant facts and problem-solving strategies stored in the knowledge base are selected (point 3) and implemented. This creates interim solutions which, together with an explanation of the solution and the plausibility probability, are output to the user via the explanation component (point 4) (point 5). This process is repeated until a final solution can be generated, i.e. until the expert system is able to provide a (probably) correct answer to the question. Point 6 represents the possibility for the user to impart new knowledge to the expert system. The connecting line is drawn in dashed lines because, on the one hand, not every expert system shell offers this option, and on the other hand, such a transfer of knowledge can only be used meaningfully if the user is also to a certain degree an expert. This means that only a user who has a certain amount of specialist knowledge in the field of the expert system is able to make accurate assessments with regard to the correctness of the (interim) solutions provided by the inference component, and thereby expand the knowledge base.

4.2.1. Knowledge base

The knowledge base is simply the "part of an expert system in which the knowledge is stored."192, thus serves as the source of the information processed in the expert system. The knowledge stored there consists of “facts and heuristics. The 'facts' constitute a body of information that is widely shared, publicly available, and generally agreed upon by experts in a field. The 'heuristics' are mostly private, little-discussed rules of good judgment (rules of plausible reasoning, rules of good guessing) that characterize expert-level decision making in the field. The performance level of an expert system is primarily a function of the size and the quality of a knowledge base it possesses. "193

The two kinds of knowledge after Karst are:

- declarative knowledge (= static factual knowledge, explicit),194 and
- dynamic knowledge (= dynamic problem-solving knowledge, implicit).195

Facts are the smallest unit of knowledge. When facts are processed, information is created.196 If several pieces of information are associated with one another, knowledge arises.197 Factual knowledge is characterized by the fact that it is limited to the description of a state of affairs. In contrast to procedural knowledge, it does not contain any information about how the knowledge is used in the course of the expert system.198 Declarative knowledge (following: facts) is usually used within the knowledge base in the form of object-attribute-value triples199 saved. An object is a physical unit or a conceptual unit, such as debt capital.200 An attribute is assigned to each object per triplet, which describes a specific characteristic or a specific property of this object. Finally, the attribute is assigned a value within a triplet that quantifies or qualifies the attribute of the object.201 The example in Table 1 is used to make the concept of object-attribute-value triples easier to understand. Each line represents an object-attribute-value triplet. In addition, the granularity of the facts stored in such triplets in relation to the entire knowledge base is made clear.

Table 1: Examples of object-attribute-value triples with terms from taxation calendars.

Figure not included in this excerpt

Source: Own illustration based on Doll, 1991, 29.

Declarative knowledge is regularly presented with the help of production rules. Production rules enable problem-solving knowledge to be formalized in the form of if-then links. If the conditions of the if part are met, the then part is executed. Production rules are also formed in combination with other forms of knowledge representation, but mainly with the above-mentioned object-attribute-value triples.202 If the two concepts of knowledge representation shown here are expanded by the factor of uncertainty, the application of these basic tools for creating a knowledge base can be shown in Figure 6 as an example.

Figure 6: Example of an uncertain rule represented with the help of object-attribute-value triples.

Figure not included in this excerpt

Source: Own illustration based on Harmon / King, 1989, 49; Karst, 1991, 36.

The premise consists of four object-attribute-value triples, which are connected with the logical operator “and”203. Each of the four object-attribute-value triples is checked. If all four triples prove to be true, the expert system comes to the conclusion according to the example rule shown: There is a 90% probability of an intra-group sale of shares. The intra-group sale of shares is an attribute of the vGA property. The conclusion that there is a vGA within the meaning of Section 8a (6) in conjunction with Section 8 (3) KStG is again as a rule in the form of IF (vGA, group-internal sale of shares, yes (x)) AND (vGA, eventual context, yes ( y)) THEN (Diagnosis, vGA, Yes (z)) deposited204.

The knowledge of the connections and the interplay of the declarative knowledge is what is known as procedural knowledge. A common form of representation of procedural knowledge are heuristics. These are rules of thumb with the help of which the search space for problem solutions can be limited to a considerable extent. As a result, heuristics can produce incorrect results, but at the same time they speed up the search process for a (probably) correct result.205

4.2.2. Inference component

Most expert system shells come with an intelligence machine by default206so that the knowledge engineer (as the person entrusted with the development of an expert system) or the expert system developer does not have to program this part of the system. Lenz provides a compact description of the task of an inference machine: "The inference component processes the case-specific information entered into the system according to different strategies in order to derive intermediate results and problem solutions by using the domain-specific knowledge presented in the knowledge base."207 The knowledge base, together with the inference engine, is the core of an expert system.208 Together, these two components do the problem-solving work.209

Lenz however, only describes the first - and more obvious - function of the inference engine. In addition to inferring (as the first function), the inference engine controls (as the second function) "the order in which the conclusions are drawn"210 the dialog with the user. "211

Inferring happens regularly212 about the mode ponens. This type of knowledge derivation is also used, for example, in the example in Figure 6 (the reader applies this inference method intuitively). The modus ponens can be generalized as follows. The rule is: “if A, then B”. If A is true, it can be concluded that B is also true.213 In general terms: If the premise is fulfilled, the actions described in the end (part B) are carried out.

However, in some cases the modus ponens cannot infer a conclusion. The rule applies again: “if A, then B”. If B is wrong, the modus ponens cannot draw a conclusion. The Tollens mode, on the other hand, then deduces that A is also wrong.214

4.2.3. Dialogue interface: knowledge acquisition and interview component

The knowledge acquisition component, together with the interview and explanation component, forms the dialogue interface between the expert system and the user.215 The main task of the dialog interface is so Willow head, "To prepare the system-internal representation of the knowledge in such a way that it is accessible to the user in an understandable manner."216crank extends this description to include the criterion of user-friendly communication.217 This includes taking into account the divergent level of knowledge of the various user groups.218 Accordingly, they are given by the basic requirements for the dialog interface219

- a high interaction speed,
- short training period and
- high fault tolerance.

The knowledge acquisition component serves three purposes: the introduction of new knowledge, the maintenance of existing knowledge and the removal of obsolete knowledge.220 The introduction of new knowledge is called knowledge acquisition.221 This includes the "identification and processing of expert knowledge for an expert system"222 to understand. As a rule, knowledge is taken from a human expert, which is then transformed from human know-how (i.e. unstructured knowledge) into machine-compatible know-how (i.e. structured knowledge) and ultimately transferred to the knowledge base.223 However, expert system shells regularly do not have any knowledge acquisition components that can be seen as a separate program part; the knowledge acquisition takes place via the rule editor or similar.224 This helps to consolidate the bottleneck in knowledge acquisition, which can be seen in the difficulty of transforming unstructured into structured knowledge, because only the knowledge engineer is then able to further develop the knowledge base.

To deal with the above-mentioned bottleneck, four approaches are discussed in the literature.225 The two classic approaches are the direct (the expert enters his knowledge directly into the expert system) and the indirect (the knowledge engineer collects the knowledge from the expert) knowledge acquisition.226 Regardless of whether the expert himself or the knowledge engineer tries to formalize the knowledge as an intermediary, the restrictions and difficulties in dealing with knowledge apply, as already explained in Chapter 4.1. In addition to the two classic approaches, there is automatic knowledge acquisition. With this approach, the expert system learns independently227without the intervention of an expert or knowledge engineer.228 The automatic knowledge acquisition takes place either by solving problems and an independent expansion of the knowledge base by the expert system, or by entering cases and corresponding solutions into the expert system, from which it then independently induces rules.229 This last approach with rules induced by the expert system itself is problematic and suffers from a lack of acceptance: the induced rules are of good quality, but often cannot be understood by human experts.230 The most current approach is model-based knowledge acquisition. The aim of this approach is to develop conceptual models to avoid loss of knowledge due to the early orientation of the knowledge acquisition to a specific implementation tool.231

The queries to the expert system are made via the interview component. The user does not necessarily have to enter the data. It is also possible to obtain data from measuring devices or sensors.232Doll distinguishes between three dialog forms of data entry by the user:

- Problem specification with batch processing: all data are entered in advance so that the expert system can solve the problem without further user intervention.
- Passive dialogue: the expert system asks the user questions that the user answers.
- Dialogue with mutual initiative: in addition to passive dialogue, the user can also add data to the system himself.233

The d3web used for this work relies on passive dialogue. Doll describes further techniques for designing the interview component.234

4.2.4. Explanatory component

The explanatory component fulfills two main tasks.235 For the user, it serves as an orientation at which point in the inference process the expert system is at any given point in time. In addition, it explains to the user why he is asked certain questions and how (interim) results come about.236 For the expert or knowledge engineer, the explanatory component is used to validate the system, i.e. to check the inference strategies selected by the system for correctness.237

4.3. Involved people

Several people or roles are necessary to bring the individual components of an expert system into a meaningful context. These are mainly the expert, the knowledge engineer and the user.238 The role of the expert system shell developer is negligible for most operational applications of expert systems.239 The expert makes his or her expert knowledge available. The knowledge engineer takes the knowledge from the expert and structures it according to the expert system shell. The user ultimately uses the expert system to solve practical problems from the area of ​​application of the expert system.240 There are two typical special cases to consider, which are characterized by the union of two people / roles in knowledge acquisition:241

- The expert will later be the user of the expert system.
- The expert is his own knowledge engineer.

In the latter case, one person combines the role of a technical expert and that of a knowledge engineer. This means that this person must meet the requirement profiles of both roles. This constellation offers the particular advantage of a relatively short period of time between the start of the project and the creation of a prototype. This is due to the elimination of time-consuming coordination between experts and knowledge engineers.242 The problem with this constellation is already fixed in the premise: the expert must have the skills of a knowledge engineer243 have or acquire them. The former is likely to be seldom the case, while the latter usually at least compensates for the time saved in the development process, even when using relatively easy-to-use expert system shells. This is especially true when the expert system is to map a subject very extensively and accordingly very powerful expert system shells are used.244

4.3.1. The expert

An expert is a person who has a knowledge advantage over an average person in a subject area245, more precisely, "an individual who is widely recognized as capable of solving a certain type of problem that most other people cannot solve nearly as efficiently or effectively"246. The most decisive characteristic of the expert is to have experience that is derived from long work in the specialist area and that is not regularly available in concrete, written form.247 This type of knowledge - implied knowledge - is difficult to articulate. Nevertheless, a suitable expert must be able to formalize all of his knowledge and thereby reveal his problem-solving process.248

While it is unlikely that every expert will have all of the skills that describe a suitable expert, it is useful to have an overview of those same desirable skills and characteristics. With the help of such an overview, the most suitable experts can then be identified.249Prerau In addition to the above-mentioned ideal case of complete formalizability of one's own knowledge, lists the following features as desirable:250

- Far above average expertise in the field. This is crucial for the quality of the content of the expert system.
- As long as possible practical experience. The expert can only formalize his expertise with the help of heuristics, among other things, with sufficiently long practical experience in the subject.251
- An excellent reputation as an expert in the field. This minimizes doubts about the quality of the statements made by the expert system. The reputation or recognition of expert status among specialist staff is therefore often the decisive determinant for the success and acceptance of the expert system.

In addition to these features, the presence or absence of which can usually be checked objectively, so-called “soft skills” of the expert also determine the success of the project. He must be able to structure his knowledge clearly and convey it to a knowledge engineer who is usually inexperienced in the field. It is difficult to check such “soft skills”, so that auxiliary variables have to be used. For example, if an expert has already published in his field, he is usually able to structure his knowledge; if he has taught, it can generally be assumed that he can communicate his knowledge to a layperson.252

As discussed in more detail in the following chapter, however, the expert is regularly not fully motivated to pass on his knowledge.253 The degree or the type of unwillingness to pass on knowledge can according to Lightfoot be typed254:

- Type 1: Unintentional knowledge retention. The expert may fail to do so. explain the specific circumstances under which a particular unit of knowledge applies; Knowledge may not be (easily) retrieved and formalized; the expert assumes a higher level of knowledge / understanding on the part of the knowledge engineer than is appropriate. It can also be problematic that experts regularly fill gaps in knowledge with heuristics, but do not make them recognizable as such to the knowledge engineer.
- Type 2: Intentional knowledge retention. The expert consciously withholds knowledge in the hope of preventing the creation of the expert system. The knowledge engineer may not be able to recognize this because - as already mentioned above - his level of knowledge in the expert field is insufficient. In type 2, the knowledge engineer must actively counteract knowledge retention with motivational aids.
- Type 3: Uncooperative expert. Type 3 sees the expert system as a threat to its position (within the company) and partially denies itself to the knowledge engineer completely. On the one hand, dealing with type 3 experts is very difficult; on the other hand, it is easy for the knowledge engineer to identify them as type 3. This means that there is little risk of incorrect or bad knowledge entering the knowledge base by a type 3 expert.

4.3.2. The knowledge engineer

Since the expert, as is regularly assumed, is not able to formalize his knowledge - or only to a limited extent - there is a need for an interface between the expert and the expert system (see point C in Figure 5). The function of the knowledge engineer is to transfer human expertise (expert knowledge) into the knowledge base. This is known as knowledge engineering.255 As shown in Figure 7, the knowledge engineering process begins with knowledge acquisition. Here, the knowledge engineer makes decisions about the relevance of the individual pieces of information from the totality of the expert knowledge. The knowledge acquired is modeled by the knowledge engineer, i.e. a suitable type of representation is selected256. Finally, the knowledge engineer codes the knowledge and transfers it to the knowledge base in the form of formalized knowledge. Within this process - but also for the development of the entire expert system - the process of knowledge acquisition is the most critical. There are a number of possible methods that the knowledge engineer can use to acquire knowledge from the expert, e.g. interviewing the expert257, group discussion, expert observation or brainstorming.258

Figure 7: Knowledge engineering process.

Figure not included in this excerpt

Source: Own illustration.

A phenomenon with which the knowledge engineer often has to deal with the subtask of knowledge acquisition - should the expert system be able to develop its maximum benefit - is the "unwilling expert".259 As a resource, expert knowledge is subject to a relative scarcity and is therefore valued at a relatively high price in a market economy. At the same time, the expert status also means power or recognition for the expert.260 Expert systems now make it possible to replicate the scarce resource.261 Since this means a withdrawal of power or recognition for the expert, experts may defend themselves against the knowledge extraction by the knowledge engineer.262 In this case, it is important that the knowledge engineer has the appropriate tools to motivate the expert to pass on knowledge, since the success of an expert system depends to a large extent on the underlying knowledge base.263 There are two (contrary) approaches to motivating people, which should only be presented here as an overview.264 The from Lightfoot The motivation plan developed is based on both approaches and essentially conveys the knowledge that only the "unwilling experts" have to be dealt with honestly,265 to motivate them to cooperate. This insight appears to be of limited innovation, since honest dealings should of course also be maintained with experts who are willing to cooperate. The developed motivation plan seems suitable to promote any cooperative relationship between expert and knowledge engineer. It consists of three steps:

- 1st step: Protect the reputation of the expert (Maslow Fourth layer). The fear of being suppressed by the expert system should be taken away from the expert. This is done by jointly delimiting the tasks to be carried out by the expert system. As a rule, this is about releasing the experts from routine tasks.266
- 2nd step: Classification of the "unwilling expert" after Ritti267. The aim is to classify the “unwilling expert” as local or cosmopolitan. This enables a clear identification of the motivators (Herzberg).
- 3rd step: Create a motivation plan. The knowledge engineer creates a motivation plan corresponding to the type (local or cosmopolitan). The design of the motivation plan depends on the type of unwillingness to pass on knowledge (unintentional, intentional, uncooperative) and the type of “unwilling expert”.268

This approach enables the greatest possible motivation to build up in the "unwilling" expert.269

From these three subtasks of the knowledge engineer (knowledge acquisition, modeling and coding)270 The requirements for the person of the knowledge engineer can be formulated from this: in addition to the “comprehensive general education” and the “self-evident computer qualifications”, knowledge of system analysis and logic are important.271

4.3.3. The user

The development of the expert system is geared towards the user. He uses the expert system as an aid in solving problems.272 There are two cases to be distinguished. In the first case, the user is identical to the expert who makes his knowledge available during development. This means that the expert system is a supporting tool. In the second case, the user is not an expert in the specialist field of the expert system, which can have a disadvantageous effect, since the expert's knowledge and ways of thinking have been adopted. The expert system thus follows an approach that regularly deviates from that of the user. This circumstance may mean a restriction of the performance of an expert system if it is mainly intended to provide expert advice to non-specialist users.273

4.4. Existing Expert Systems and Shells: An Overview

A very extensive compilation of expert systems in use in German-speaking countries is provided Mertens. 274 More up-to-date but also less comprehensive Arend Applications of expert systems in commercial practice275, as well as the development tools available for creating expert systems (= expert system shells)276. Deliver a compilation of expert system shells in the international area Harmon / King.277 Even if these market overviews are not up-to-date, they make one thing clear: Expert systems will become more and more important - also for operational applications - to the extent that their development (including the development environments available) is standardized.278

4.5. System of choice: d3web

The expert system shell used in Chapter 6 was developed by the Chair of Computer Science 6 at the University of Würzburg279. It is now available as a web-compatible version d3web.280 Several demonstrations were also developed at the above-mentioned chair, which depict expert systems from different departments with the help of d3web.281 Even if the development of d3web began in the medical field, the expert system shell is still suitable for modeling legal (including tax law) expertise. The choice of d3web as a development tool is mainly based on the tidy and logically structured user interface. In addition, the shell offers some powerful tools such as real-time test runs of the previous state of development, validation of the knowledge base and visualization of the decision trees without forcing the knowledge engineer to learn an independent programming language.

5. Development of an expert system

If tax advice is understood as a subgroup of legal advice, the statements of the relevant literature on the use of legal expert systems can be transferred. The knowledge of a tax expert is generated from legal texts282 and case collections.283 This means that this knowledge is very "strongly defined by established facts and rules"284 is. However, a legal expert system will not always produce correct solutions285, just as little as legal experts can. In this sense, it is necessary to precisely delimit the problem or the subject area in which the expert system is to produce solutions.

5.1. Selection of a suitable subject

Expert system software should be used when the problem to be solved cannot be sensibly solved using standard software286, and the necessary expert knowledge can only be compiled with considerable effort. If it is easy to obtain, the cost of developing an expert system is too high relative to the benefit.287 Suitable problems are characterized by the fact that concepts must be dealt with in order to solve them. This means that whenever logical processes lead to a solution, the use of expert systems is preferable to standard software. Problems whose solution is mainly defined by numerical calculations are not suitable.288 The strength of expert systems in comparison to standard software lies precisely in the fact that they can handle heuristic data and thus also with concepts in which uncertainty, incomplete information or qualitative differences may also play a role.289 At Prerau there is a checklist for selecting and checking the suitability of a subject.290

5.2. Selection of experts

Following the precise delimitation and description of the specialist area, suitable experts must be found.291 This process can be broken down into the identification and acquisition of experts for the project. For identification see the checklist of Prerau,292 For the acquisition of, especially unwilling, experts see Sections 4.3.1 and 4.3.2. It should also be emphasized that companies should not be afraid to make their top experts available for expert system projects.293 The one-time effort is offset by the benefit of the permanently preserved expert knowledge. In addition, the allocation of such top experts to the expert system project in particular should not be canceled before the end of the project.294 Not only would the new expert have to familiarize himself with the project - the conditions of the expert system shell, selected forms of knowledge representation, etc.; he would have to learn the solution strategies of the old expert himself, which were already recorded in the knowledge base, in order to guarantee the consistency of the knowledge base.295

5.3. Methodical development of expert systems

Part of the task of this thesis is the creation of an expert system for intragroup shareholder external financing within the meaning of Section 8a (6) KStG. Since the development of an expert system is a deductive process, the development process is very well suited for being divided into several phases296, whereby the general is concretized into the specific. The main goals of the development are applicability, problem adequacy, plannability and validation.297 The systematic implementation of these goals is known as a phase model.298 In almost each of the common phase models formulated for expert system development there is so-called prototyping299 anchored.300

5.3.1. Phase models

The use of phase models is an approach known from the development of standard software. In accordance with the nature of standard software, the classic phase models are organized linearly.301 This implies that they cannot be transferred to the development process of expert systems, since in the linear models each phase is usually only run through once. The creation of expert systems, however, involves a high degree of iterative processes302which make it necessary to go through the individual phases several times. Several phase models are described in the literature that take this peculiarity of expert system development into account.303

5.3.2. Prototyping

Prototyping is to be understood as the “iterative process of creating, running, evaluating and correcting software prototypes”.304 A prototyping cycle consists of the phases of choosing or refining the problem, developing a prototype, evaluating, integrating and maintaining the system.305 A prototype serves as a "working model of ... an information system"306. M.a.W. prototypes of expert systems serve to gain additional knowledge - which is detailed in the requirement profile for the expert system.307 Go to the prototype development phase Waterman of an evolutionary origin,308 as shown in Table 2. This prototyping cycle is repeated until a fully functional application system can be implemented.309

Table 2: Evolution of an expert system.

Figure not included in this excerpt

Source: Own translation by Waterman, 1986, 140, table 12.1.

This schedule - also known as rapid prototyping - provides for ongoing development310 of the first prototype up to commercial usability. The particular advantage lies in the fact that a functioning expert system is already available at the beginning of the development work. Further advantages of using prototyping are:311

- Fast verifiability of the knowledge engineering process for correctness.312
- The suitability of the chosen method for knowledge representation and the inference strategies can be checked at an early stage.
- The expert can continuously test the problem-solving behavior of the system and, if necessary, make the knowledge engineer aware of deficiencies.
- It may become clear early on in the development that the problem can be solved better by a conventional program. There are then no further (unnecessary) costs for the expert system development.

The advantages are offset by a few disadvantages:313

- Due to the phasing of the development process, it is not possible to define milestones.314
- The knowledge representation orients itself too strongly on the implementation process, i.e. the form of the representation is chosen according to the point of view of the quick implementability. Possibly the knowledge formalized in the knowledge base is no longer manageable for experts because it no longer corresponds to their terminology.
- Overall, the knowledge engineer is more focused on the details of the technical implementation than on the knowledge engineering process.
- The specific design of the system architecture315 will be determined in good time. This may lead to later in the development process on difficulties with extensions.

When it comes to tax advice, the advantages seem to be far more important than the disadvantages. The dynamics inherent in prototyping are clearly advantageous, since the knowledge on which the expert system is based - tax law - also has a certain fast pace in its creation and further development. Establishing a specific system architecture and forms of knowledge representation at an early stage is unproblematic, since the solution strategies and data on which tax law is based are in turn constant. The only disadvantage that cannot be clearly eliminated remains the focus of the knowledge engineer on the implementation, instead of on the actual knowledge engineering process.

6. Expert system for Section 8a (6) KStG: The development environment

The expert system developed is to be understood as a demonstration prototype and serves as a proof-of-concept316. To Waterman A period of one to three months should be allowed for the development of a demonstration prototype.317 This corresponds to the time frame of the present work. As a demonstration prototype, the expert system only solves part of the problem posed. As a result, it does not recognize every special or doubtful case. This limitation is due to the fact that the prototype contains relatively little facts and knowledge. The possibility of skipping questions, i.e. setting the answer to "unknown", is hidden in the prototype. This means that the prototype cannot deal with uncertainty. The problem can only be solved by providing all the requested input data. To the extent that the system is further developed and, for example, should also be able to interpret questions of doubt in a meaningful way, this option can be activated within d3web.318 Regardless of this, the prototype enables the analysis of whether the problem can be captured efficiently by expert system technology and whether the exact problem definition, the defined scope of the problem and the chosen form of knowledge representation are expedient.319

Figure 8 shows the d3web development environment, ie the expert system shell. The question sets and diagnoses are managed on the left in the "Navigator". The desired editing mode of the object selected in the "Navigator" can be selected in the middle.320 In the figure, the “attribute table” is activated: The object “seller of the investment” is selected in the “navigator”, the assigned attributes are stored with their respective values ​​in the “attribute table”. Each line therefore represents an object-attribute-value triple.321

Figure 8: Screenshot of the development environment.

Figure not included in this excerpt

The rules or links between object-attribute-value triples created with the help of the rule editor or the decision tree are stored in the knowledge base and can be displayed and edited via the "rule overview"322 as shown in Figure 9. The rule editor can, among other things, record heuristic rules and the probabilities assigned to them. For example, a rule can say: IF x and y and (v or w), THEN z (0.4). So if x and y as well as v or w are given, then z is diagnosed as 40% likely.

Figure 9: Screenshot of the knowledge base (“rule overview”).

Figure not included in this excerpt

The decision tree shown in Figure 10 is a central component in addition to the “rule overview”. It visualizes the process control of the survey of the user. It enables a quick and robust representation of procedural knowledge, i.e. how questions are interdependent and in what chronological order they are asked to the user. The previously created object-attribute-value triples can be chained together in this editing mode.

Figure 10: Screenshot of the inference control ("decision tree").

Figure not included in this excerpt

D3web is ideally suited for the development of an expert system with recourse to (rapid) prototyping. At any point in time, the knowledge engineer can start the dialog interface dynamically using the "Start case" option, i.e. all changes to the knowledge base are taken into account.323 The dialog interface is opened in the web browser, as shown in Figure 11, and is technically implemented using standardized web technologies, which ensures extensive compatibility on all common platforms. In addition to the option of calling up the dialog interface in real time, it can also be exported as a CD-ROM. All necessary server applications that control the course of the dialog are included so that integration is possible on desktop workstations as well as on network servers.

Figure 11: Screenshot of the dialog interface.

Figure not included in this excerpt

On the enclosed CD-ROM324 is the expert system developed in the course of this work for Section 8a (6) KStG. The reader can thus get an impression of the dialog interface.

7. Summary outlook

The aim of this work was to provide practical evidence of the usefulness and feasibility of expert system development in tax theory and consulting. For this purpose, a previously delimited topic was analyzed and - in the process of knowledge engineering, so to speak - applied to the given problem. The theoretical constructs relevant to the operational application of expert systems were then explained. Finally, a functional expert system was created as a demonstration prototype. So it has been possible to prove the feasibility of expert systems in tax consulting. Naturally, the sense of purpose lies in the subjective approach of each individual; but at least for the topic to be worked on it is given. It remains to be seen whether expert systems will prevail in German tax advisory practice. Relevant research projects could drive development forward. In addition to further developments in the technical implementation of expert systems in tax consulting, it will be of overriding importance to develop programs for expert motivation in order to mobilize the existing knowledge for the purposes of expert system development. The twofold benefit from expert systems could be decisive for the support provided by the companies: in addition to the core benefit of standardized advice based on validated expert knowledge, expert systems could also be used as a training tool.325 It would then be conceivable to use such a system to roll out expert knowledge on new or changed legislation as quickly as possible,

The trends that can be identified in the field of expert systems are also positive, some of which are particularly suitable for the use of such systems in tax consulting. In connection with the knowledge problem, research on the formalization problem of implicit knowledge and the integration of so-called information retrieval systems are particularly important326 to call.327 Although the knowledge engineering process is the most important sub-task in the creation of expert systems, few innovative approaches to knowledge acquisition can be found in the more recent literature. There is clearly a need for research328 ; especially for knowledge acquisition in the business environment of tax consulting.

attachment

Figure 12: Example of the “calculation” of indirect participation rates.

Figure not included in this excerpt

Source: Own representation of the example in Frotscher, 2004, margin no. 452

Figure 13: Dependence of the need for a knowledge engineer on the size of the expert system.

Figure not included in this excerpt

Wagner, 1992, 69, Figure 17.

Figure 14: Employee typification according to Ritti: local or cosmopolitan.

Figure not included in this excerpt