Towards a Foundation for Intelligent Contracts

Georgios Stathis

1, 2 a

, Athanasios Trantas

2 b

, Giulia Biagioni

2 c

, Jaap van den Herik

1 d

Bart Custers

1 e

, Laura Daniele

2 f

and Theoﬁlos Katsigiannis

3 g

eLaw - Centre for Law and Digital Technologies, Leiden University, Kamerlingh Onnes Building,

Steenschuur 25, Leiden, The Netherlands

Unit ICT, TNO, New Babylon, Anna van Buerenplein 41a, Den Haag, The Netherlands

Independent Researcher

Keywords:

Artiﬁcial Intelligence, Trustworthy AI, Ontology Engineering, LegalTech, Contract Automation, Intelligent

Contracts, Preventive/Proactive Law.

Abstract:

This article investigates the incorporation of Artiﬁcial Intelligence (AI) within LegalTech. We deﬁne an ontol-

ogy to form the basis for Trustworthy AI processing of contract automation. The value of our research is that

it applies an ontology, as existing tool, in contract automation. Two perspectives are emphasized: communi-

cations analysis and risk analysis. They are explored under a new prism. Our context is Intelligent Contracts

(iContracts), which aim at reducing the time-consuming and often complex contractual process by minimizing

human involvement. Contract communications and risk analysis processes are often neglected in automation.

Therefore, our research investigates to what extent is possible to design an ontology for contract automation

based upon the combination of both. Our methodology is twofolded. First, we concentrate on applying key

word search on an online database to demonstrate the lack of available solutions. Second, we develop an

ontology based upon a case study of a freelancer agreement. Of course, we use existing literature to further

engineer the ontology. Our ﬁnding shows that 9.4 percent of LegalTech solutions deal with contract automa-

tion. From them, 0.7 percent focus on communications and risk automation for contracting. The conceptual

expressiveness of the ontology is validated with research to the use case. A follow-up discussion suggests that

the ontology should be further engineered from a third perspective, trustworthiness, and should be re-validated

experimentally. Our conclusion underlines the need for further innovation in contract automation, especially

in relation to communications and risk data.

1 INTRODUCTION

The promise of a gift is different from the gift of a

promise. Both are attractive. However, the following

questions arise. Which one is better or which one is

always the best? Can we utilize an ontology to guide

us in difﬁcult decisions? And to what extent can Ar-

tiﬁcial Intelligence’s (AI) involvement aid us?

While global media outlets often advance claims

discussing the replacement of humans by robots in the

https://orcid.org/0000-0002-4680-9089

https://orcid.org/0000-0001-7109-9210

https://orcid.org/0000-0002-9005-7945

https://orcid.org/0000-0001-9751-761X

https://orcid.org/0000-0002-3355-8380

https://orcid.org/0000-0002-9267-7160

https://orcid.org/0000-0002-4422-2858

labour market, social confusion often ensues (Larson,

2021). The same holds for the legal world. Univer-

sities, law ﬁrms, and startups from around the world

are often showcasing their innovative activities in the

ﬁeld of AI and LegalTech. However, such marketing

promotions increase the difﬁculty in deciphering what

is the actual social progress in innovation. Therefore,

the aim of our paper is twofold: (1) to clarify the state-

of-the-art innovations in contract automation (i.e. a

particular ﬁeld of AI and Law), and (2) to lay the

foundations of Intelligent Contracts (iContracts).

1.1 Three Innovations

Recently, the ﬁeld of contract automation has expe-

rienced three major innovation. The ﬁrst is the dig-

italisation of contract management (hereinafter digi-

tal contracts), where certain contractual processes are

Stathis, G., Trantas, A., Biagioni, G., van den Herik, J., Custers, B., Daniele, L. and Katsigiannis, T.

Towards a Foundation for Intelligent Contracts.

DOI: 10.5220/0011628200003393

In Proceedings of the 15th International Conference on Agents and Artiﬁcial Intelligence (ICAART 2023) - Volume 2, pages 87-98

ISBN: 978-989-758-623-1; ISSN: 2184-433X

 2023 by SCITEPRESS – Science and Technology Publications, Lda. Under CC license (CC BY-NC-ND 4.0)

digitised, such as signing, drafting, storing, review-

ing, sharing, and analyzing contracts (Timmer, 2019).

The second innovation regards the rise of smart con-

tracts demonstrating that parties can reach and ex-

ecute agreements via programming (Kolvart et al.,

2016). Today we are facing the dawn of the third

innovation, namely iContracts. iContracts introduce

a hybrid approach between human and computer in-

terventions aiming at achieving full automation with

self-executing contracts (Mason, 2017). iContracts

introduce the state-of-the-art innovations in the space

of contact automation due to their compliance with

Hybrid AI principles (Huizing et al., 2020).

1.2 Four Challenges

There are multiple challenges with digital, smart, and

intelligent contracts. They span from usability and

adoption challenges, to the automation of contract ex-

ecution monitoring and dispute resolution. Yet, one

of the major challenges observed is that the commu-

nication process preceding the drafting of contracts is

often neglected. During the communication, contract-

ing parties exchange useful information that affect the

design of contracts. Typically, a legal expert leverages

the information and analyses the risks that may derive,

for example, from relevant legal rules (Stark, 2013) in

order to draft a contract.

The above-described situation can be exempliﬁed

as follows. A client who has a design problem and an

according budget, contacts a freelance designer with

the intention to hire him/her/it. To initiate the project,

the client and the designer (henceforth the contrac-

tors) need to have either a written or verbal agree-

ment. If the agreement is written, both parties are

legally protected to a higher extent, because as the

popular proverb says: verba volant, scripta manent.

In the context of a freelancer agreement, the draft-

ing of the contract is usually entrusted to legal experts;

mostly due to the lack of legal knowledge of contrac-

tors. However, not all freelancer agreements take a

written form. Additionally, practice shows that there

are multiple challenges involved in the process. In the

subsequent part, we discuss four of them.

The ﬁrst challenge concerns the actual implemen-

tation of the freelancer agreement. Contractors in-

volved in freelance projects usually tend to agree ver-

bally. According to the freelancers union, 72 percent

of freelancers do not use written contracts

Second, we examine the issue of pricing. A client

holds certain budget and needs a landing page for a

https://blog.freelancersunion.org/2015/12/21/why-do-

only-28-freelancers-use-contract/

website. The client can contact designers with differ-

ent pricing and quality points. The quality of a de-

liverable may be lower than the client’s expectation,

leading the client to not compensate the designer. If

there is no contract, the designer cannot legally claim

the amount owed.

Third, a challenge during the communication

phase of contracting is that discussions can be ab-

stract and require speciﬁc legal knowledge. The pres-

ence of a legal expert in the information collection

process narrows down the scope of the discussion.

Moreover, an experienced legal expert connects the

information exchanged with the relevant risks, before

designing the contract (Eggleston et al., 2000).

Fourth, the technological challenge is that despite

the availability of many contract automation technol-

ogy solutions, the legal communication and risk anal-

ysis processes are not often automated. As a result,

today’s legal experts spend signiﬁcant time collecting

information from the communication exchange be-

tween contracting parties and analyzing risks before

drafting a secure contract.

iContracts can greatly beneﬁt each described chal-

lenge, because they introduce practical, data-driven

solutions. Gradually, the solutions can be applied in

more complex contexts, including for example enter-

prise or government contractors. So far the main al-

ternative for freelancers or organisations is mostly re-

liant on physical contract interventions that are hardly

scalable. The focus of our research is sufﬁcient to

demonstrate how iContracts beneﬁts small scale chal-

lenges, for future application in more complex envi-

ronments.

1.3 Two Possible Solutions

Automating the legal communication and risk analy-

sis processes can beneﬁt the contracting parties. The

ultimate beneﬁciary of such automation would be the

non-legal experts, since they can leverage the automa-

tion of the contracting process. However, one should

not rush to avoid the inclusion of a legal expert or at-

tempt to avoid one altogether. At this moment we are

cautious to note that automation cannot be immedi-

ately successful for all types of contracts. Therefore,

our research focuses on a straightforward contracting

case study.

The case study concerns a contract regarding the

provision of freelance services. The scope of this pa-

per will be limited to this case study, as there is of-

ten little contract communication and risk analysis re-

quired before drafting such a contract.

If the automation proves to be successful, grad-

ually this automation can be applied to more com-

ICAART 2023 - 15th International Conference on Agents and Artiﬁcial Intelligence

plex types of contracts. Due to lack of literature for

the speciﬁc type of automation, there are no avail-

able data sources structured accordingly. As a conse-

quence, collecting and building data-sets will be hard.

Hence, as matters now stand, it becomes imper-

ative for AI and LegalTech researchers to structure

the available data for automating a contract in the

present context. The process of structuring data may

involve many diverse challenges such as selecting the

right contract templates, gathering multiple commu-

nication data and analyzing multiple risks in order to

make secure contracts. Yet, the most prominent chal-

lenge is: how the communication and risk analysis

processes can be handled in a harmonised, scientiﬁc

manner?

1.4 Three Research Approaches

To address the issues involved, we are asking the

question: can Machine Learning (ML) help us? There

are three available options. First, a ML approach,

whereas an unsupervised ML algorithm can be ap-

plied to any available contract data. Yet, such data

are hard, if not impossible, to ﬁnd or obtain. Second,

a relational-data model can be developed that demon-

strates how various data sources connect with each

other. Yet, such a model may be limited for the appli-

cation of AI. Third, ontology engineering is an option.

Ontologies demonstrate how it is possible to handle

interconnected data sources with a great variety of

data types. They are able to support the structuring

of data in a scientiﬁc manner (Duan et al., 2017). The

beneﬁts of developing an ontology relate to interoper-

ability, standardisation, conceptualisation, inferential

reasoning and information retrieval.

Hence, we choose an ontology to serve as the

backbone of an explainable “intelligent” platform

where various modern technologies can be incorpo-

rated and tested (Sarker et al., 2020). The core mod-

ule of this platform will utilise modern models and

techniques in the ﬁeld of AI.

The value of the ontology stems from its potential

to support the implementation of communication and

risk management in contract automation. It is extend-

able, can support additional solutions and clarify lim-

itations. In essence, an ontology can be the method

that helps the systematic study of contract automa-

tion, towards achieving the goal of iContracts. Hence,

following the discussion, it logically follows to be-

gin by creating an ontology that increases the trans-

parency of the communication and risk processes in

contract automation.

The above-mentioned information lead us to the

following Research Question (RQ): To what extent is

it possible to deﬁne an ontology to sufﬁciently au-

tomate contracts based on communications and risk

data?

To answer the RQ, we structured the paper as fol-

lows. In Section 2, the literature is described. Section

3 presents the method of research. Then, Section 4

states the results of our research and Section 5 dis-

cusses those results. Finally, Section 6 answers the

RQ and provides our conclusion.

2 LITERATURE

The literature Section is structured as follows. Sub-

section 2.1 introduces contract law. Subsection 2.2

presents sources on contract automation solutions.

Subsection 2.3 introduces contract communication

literature. Subsection 2.4 discusses the literature on

legal risk. Then, Subsection 2.5 introduces the state-

of-the-art literature on smart contracts. Subsection

2.6 discusses iContracts. Subsection 2.7 states forth

the literature of ontology engineering. Subsection 2.8

presents the relevant ontology literature on contract

automation.

2.1 Contract Law

In most jurisdictions around the world, contracts are

deﬁned as follows.

• Deﬁnition 1: A contract is a legally binding

agreement, verbal or written. (Smits, 2017)

For an agreement to be binding, certain requirements

must be met. Those requirements are usually laid

out in the contract law of the relevant state, which

typically also ensures that conﬂicts can be resolved

through the court system In general, contracts are

governed by private law and in each jurisdiction there

are certain rules for contracting. Generally, those

rules may be substantively divergent. In this research,

this divergence is limited to the lowest extent possible

as one of the simplest forms of contracts is used in the

case study, in other words, the freelance agreement.

2.2 Contract Automation Solutions

The two largest online databases on available contract

automation solutions are (1) Stanford University’s

Legaltechlist

and (2) Legalpioneer.org

. Legaltech-

list, is a strictly curated database while the Legalpi-

oneer list is a more extensive database. At the time

https://techindex.law.stanford.edu

https://www.legalpioneer.org/

Towards a Foundation for Intelligent Contracts

this research was conducted (September 2022), Stan-

ford’s website has a total of 1,962 results and Legalpi-

oneer’s website has 9,608 business cases archived. In

these databases, the amount of available contract au-

tomation solutions that relate with this research will

be identiﬁed. We decided to focus on identifying

companies in Legalpioneer due to the larger amount

of available data. These data is expected to support

the importance of our research scope.

2.3 Contract Communication

Most of the available literature on contract commu-

nication is focused on contract negotiations. The

word “communication” concerns to a larger extent

how contracting parties should talk with each other, in

order to gain a negotiation advantage, reach an agree-

ment, and avoid the escalation of conﬂicts. Here we

remark that in our research, the word “communica-

tion” refers to the substantive information that is di-

rectly relevant for the design of a contract.

2.4 Legal Risk

The ﬁrst framework for the management of contrac-

tual risks emerged in 1950 with the introduction of

Preventive Law by the lawyer and attorney Louis M.

Brown. (Brown and Rubin, 1950) Brown believed that

preventive law concerns the cost difference between

entering into and avoiding legal costs. He thought

that legal problems arise because of legal risks. At

the end of the century, his student, Eduard A. Dauer,

started the development of a systematic analysis for

the management of legal risks (Dauer, 1987). In 2002,

the academic Thomas D. Barton took an interest in

continuing this line of research by advancing Dauer’s

analysis further with his own method (Barton, 2002)

Around the same time, in 1998, the lawyer and

academic Helena Haapio introduced the concept of

Proactive Law (Haapio and Varjonen, 1998). Proac-

tive law is a future-oriented approach to law and legal

agreements, placing an emphasis on legal knowledge

to be applied before things go awry

. The difference

between preventive and proactive law is that the lat-

ter, except from the preventive dimension, adds the

promotive dimension in terms of good and desirable

behavior (Berger-Walliser, 2012). Haapio is mostly

concerned with the application of proactive law in

contracts.

In 2010, she created a synergy between proac-

tive law and the United States school of law as a

competitive advantage with the academic George J.

http://www.juridicum.su.se/proactivelaw/

Siedel (Haapio and Siedel, 2010). As a consequence

of this synergy, in 2013 they published the book A

Short Guide to Contract Risk where they analyze con-

tractual legal risks (Haapio and J, 2013). At around

the same time in 2010, Haapio introduced the theory

of legal design, which advances the theory of Preven-

tive/Proactive Law (PPL) by translating complex legal

language into clear language expressions and visual-

izations, so that contracts can be understood by ev-

eryone before legal problems arise (Berger-Walliser

et al., 2017).

In 2004, the academic Jon Iversen introduced Le-

gal Risk Management (Iversen, 2004). Then, in 2007,

the academic Tobias Mahler discovered the difﬁculty

in deﬁning legal risk and how diverse it is (Mahler,

2007). Following the introduction of a compliance

risk management by ISO (International Standardiza-

tion Organization) in 2014 (Bleker and Hortensius,

2014), Mahler along with the academic Samson Es-

sayas, set out to systematically analyze and model

compliance risk in 2015 (Esayas and Mahler, 2015).

Recently in 2020, ISO introduced the ﬁrst Legal Risk

Management (LRM) standard focused exclusively on

legal risk for organizations and deﬁnes legal risk as

follows(ISO, 2020).

• Deﬁnition 2: Legal risk is risk (effect of un-

certainty on objectives) related to legal, reg-

ulatory and contractual matters, and from

non—contractual rights and obligations.

2.5 Smart Contracts

Today the research of PPL focuses on smart con-

tracts (Corrales et al., 2019b).

• Deﬁnition 3: A smart contract is a computerized

transaction protocol that executes the terms of a

contract.(Szabo et al., 1994)

Essentially smart contracts refer to the programming

functionalities of legal contracts (Kolvart et al., 2016).

Smart contracts were introduced to the public initially

via blockchain technologies. Bitcoin introduced the

ﬁrst peer-to-peer electronic payment system, which

manages transactions without any need for interme-

diaries

. Then, Ethereum, expanded upon this tech-

nology by managing to codify the necessities for the

satisfaction of contracts by adding certain milestones,

specifying which work could be used as insurance for

payments, creating smart contracts

. Even though

smart contracts have been popularised via blockchain-

based decentralized applications, theoretically they

www.bitcoin.org

www.ethereum.org

ICAART 2023 - 15th International Conference on Agents and Artiﬁcial Intelligence

can also be applied on centralized applications

The novelty from the use of smart contracts in

LegalTech stems from the adoption of computer

code instead of human language for managing con-

tracts (Kozlova and Aleksandrina, 2020). It is from

this perspective that the school of legal visualisation

under PPL is conducting research on smart contracts,

so that the smart contract rules are more understand-

able and accessible for contractors (Corrales et al.,

2019a) (Barton et al., 2019). Haapio often empha-

sises the importance of design for contracts, but that

it is particular so for smart contracts (Hazard and Haa-

pio, 2017).

2.6 Intelligent Contracts

In essence, smart contracts work when one is able to

manage and prove milestones. On a macro level, ap-

plying this technology in a complex legal situation

unfolding in, for example, an energy project would

require higher sophistication. This higher sophis-

tication is examined under the aegis of intelligent

contracts or iContracts (McNamara and Sepasgozar,

2021). So far iContracts are deﬁned as follows.

• Deﬁnition 4: An intelligent contract or iCon-

tract is a contract that is fully executable without

human intervention

This ﬁeld introduces a hybrid contract automation ap-

proach and considers the need for contract automation

that correspond to the complexities of reality, aim-

ing for the transition into full self-executing automa-

tion, with minimal human intervention or without it,

if possible (Mason, 2017). Motivated by the devel-

opments in Industry 4.0, this ﬁeld is mostly evident

in construction (McNamara and Sepasgozar, 2018),

where a high level of complexity drives the need

for such innovation (McNamara, 2020); the same

can be said for smart factories (Aimin and Yunfeng,

2019). Despite the large academic call for the need

of iContracts and the developing frameworks for its

adoption (Pillai and Adavi, 2013), many acceptance

challenges are evident in practice (McNamara and

Sepasgozar, 2020). The key value of iContracts is

that they can leverage information from smart Inter-

net of Things (IoT) sensors for automated monitor-

ing of contracts (McNamara and Sepasgozar, 2020).

It should be noted here that iContracts can be im-

plemented in both centralised and decentralised sys-

tems (Zheng et al., 2020).

https://contractbook.com/blog/smart-contracts-withou

t-blockchain

https://bravenewcoin.com/insights/pamela-morgan-

at-bitcoin-south-innovating-legal-systems-through-blockc

hain-technology

The iContract developments prove that the moni-

toring and execution of contracts is more related with

project management. However, the level of project

management with respect to technological readiness

is diverse. For example, in a freelance agreement,

where there is a lack of IoT sensors, it would be harder

to monitor a contract.

2.7 Ontology Engineering

To advance the ﬁeld of contract automation, we need

annotated data (deterministic data for the application

of expert system algorithms, and probabilistic data

for the application of Machine Learning (ML) algo-

rithms) (Macmullen, 2005) which must be encoded

into a machine-friendly structure. This is vital for

AI applications, which include not exclusively (1)

game playing, (2) speech recognition, (3) understand-

ing natural language, (4) computer vision, (5) expert

systems, and (6) heuristics classiﬁcation (McCarthy,

2004). We believe that a new structure heavily re-

lies on the development of an ontology. An ontol-

ogy is used in many different ways in literature but its

original meaning comes from Philosophy. It concerns

the study of being or being existent (Gruber, 1993a)

(Guarino et al., 2009) and includes things that exist

in the real world. In our work, we deﬁne ontology as

follows.

• Deﬁnition 5: Ontology is a formal, explicit spec-

iﬁcation of a shared conceptualization that is

characterized by high semantic expressiveness re-

quired for increased complexity (Feilmayr and

oß, 2016).

We should note here that ontologies are closely inter-

connected with Knowledge Graphs (KGs). Ontolo-

gies represent the context (t-box as in tool box) while

KGs are the tool used to utilize them (a-box as in al-

gorithmic box).

• Deﬁnition 6: A Knowledge Graph is deﬁned as

a 4-tuple G = (N, E, L, f) being a directed labeled

graph, where N is a set of nodes, E⊆N×N is a set

of edges, L is a set of labels, and f: E→L is an

assignment function from edges to labels.

The assignment of a label B to an edge E=(A,C) can

be viewed as a triple (A,B,C).The triple A, B, C is re-

ferred to as the subject, the predicate, and the object of

the triple, respectively. KGs have been instrumental

in the context of knowledge representation learning

as they can store output data in a standardised format,

simultaneously consuming the data to obtain domain

knowledge (Vinay K. Chaudhri, ). Our interest here

lies in KGs’ property to capture ontologies (Gruber,

1993b). An example of a triple in our case study is

that a (A) legal expert (B) designs a (C) contract.

Towards a Foundation for Intelligent Contracts

In the end, an ontology serves to create a formal

representation of the entities in the graph. The differ-

ence between a KG and an ontology is that the former

includes an ontology as well as data that validate it.

Both ontologies and KGs are based on the Resource

Description Framework (RDF) (Group, b) triples.

They have a similar representation style and tend to

resemble each other in visualizations, although on-

tologies are usually based on a taxonomy (Education,

). It is interesting to see that they can contain multiple

taxonomies, thus having their own deﬁnitions. The

Web Ontology Language (OWL) (Group, a), the Se-

mantic Web Rule Language (SWRL) (Ian Horrocks,

) along with RDF belong to a family of representa-

tion languages standardised by the World Wide Web

Consortium (W3C) (w3c, ) with the goal to capture

knowledge on the internet.

Creating an ontology can be done in many ways

by supplementing manual knowledge engineering

techniques with signiﬁcant automation and crowd-

sourcing. More precisely, ontologies can be learned

from unstructured or semi-structured data sources, as-

sembled from existing ontologies, usually aided by

various semi/fully-automated data validation and in-

tegration mechanisms, or created from scratch by do-

main experts (Paulheim, 2017).

2.8 Ontologies in Contract Automation

So far ontologies have been applied multiple times

in a legal context, but not for the speciﬁc context

of contract automation via communication and risk

data. We provide ﬁve examples: for the structuring

of legal norms and court decisions (Filtz, 2017), for

posing legal questions related to legislative sources

and answering them (Sovrano et al., 2020), for com-

pliance purposes in complex multi-lingual, multi-

jurisdictional environments (Schneider et al., 2022)

(Montiel-Ponsoda et al., 2018), for online case anal-

ysis (Yu et al., 2021), and similarly, for case recom-

mendations (Dhani et al., 2021). In relation to con-

tract automation, ontologies have been used for con-

ceptualizing contracting terms and promoting interop-

erability regarding concepts (Garc

ıa and Gil, 2008) as

well as on infological- and datalogical-level data ex-

changes for blockchain-based smart contracts (Krui-

jff and Weigand, 2017). Moreover, they have been

exploited more generally for blockchain-based smart

contracts (Zhou et al., 2020) and other research con-

cerning contracts (Kaltenboeck et al., 2022) or con-

tract risk management (Wu, 2020), albeit at a higher

level of abstraction. The closest research to date on

our subject is that of Legislate, where they use an on-

tology for drafting and negotiating contracts as well

as representing rights and obligations, which happens

behind closed doors as their KGs

are protected by a

patent on semantic document generation

. In ad-

dition to all the applications, our research is apply-

ing ontologies from the perspective of communica-

tion and risk data automation which makes it unique

in the available body of literature. The potential of ap-

plying ontologies in the legal domain may even reach

the level of developing industry-wide interoperability

standards, in the same way that it occurred in the ﬁ-

nancial industry via the Financial Industry Business

Ontology (FIBO)

3 RESEARCH METHOD

The methodology initially concerns the analysis of

Legalpioneer data which is based on Key Word

Search (3.1). Then, for ontology engineering, the

methodology follows two stages. Determining the

Case Study (3.2), and deﬁning of the ontology (3.3).

3.1 Key Word Search

To gather data in Legalpioneer we had to conduct key

word search, with the expectation of identifying the

available contract automation solutions today. Af-

ter typing in the key words “contract automation”,

around 905 results appeared. We manually analyzed

every single result presented after a search, includ-

ing their respective websites, and conducted anala-

ysis about their relevance and connection to our re-

search. We identiﬁed two numbers: (1) the percent-

age of the total contract automation solutions in all

available LegalTech solutions, and (2) the percentage

of contract automation solutions based on communi-

cation and risk management in all available contract

automation solutions.

3.2 Case Study

The scope of this case study concentrates on a free-

lancer agreement. To get this agreement, we down-

loaded an NDA agreement from the open-source le-

gal documentation database of Capital Waters

and

we adjusted it to ﬁt the needs of our case study.

https://www.legislate.tech/post/knowledge-graphs-

know-more-about-your-contracts

United States Patent 11087219

https://edmcouncil.org/page/ﬁnancialindustrybusiness

ontology

https://www.capitalwaters.nl

ICAART 2023 - 15th International Conference on Agents and Artiﬁcial Intelligence

3.3 Ontology Engineering

To design an ontology, requirements need to be gath-

ered. The requirements collected for the design of the

ontology are based on the case study and a literature

review. Taking into consideration the requirements,

we arrived at the Onassis Ontology, which is visible

in Figure 1 and accessible via Github

Figure 1: A visualization of Onassis Ontology.

The ontology that we designed retraces the inter-

active process of asking questions and giving answers

between a legal expert and a contractor. The pro-

cess that we aim to frame for automated methods (and

consequently bind it to a formal contract) can be pre-

sented through the following ontological conceptual-

ization.

The starting points of the above-mentioned inter-

active process are the legal expert and the contractor.

The legal expert writes a question for the contractor

who has previously selected a speciﬁc scope for the

contract. By replying to the question, the contractor

provides an answer. On a “physical level”, the an-

swer occupies a position in (1) a variable of a para-

graph, (2) a paragraph of a section, (3) a section of a

contract, and (4) a contract. The variable, paragraph,

and section follows a numerical order within the con-

stituent parts of the digital document (i.e., the con-

tract). The paragraphs of a section (i.e., the section

itself) are grouped under standardised topics and are

regulated by legal rules.

The contract contains various numbers of agree-

ments. An agreement here is conceived as a consen-

sus involving at least two different parties and regard-

Onassis is accessible at https://github.com/onassison

tology/onassisontology and is protected by the open-source

GNU General Public License https://www.gnu.org/licenses

/gpl-3.0.html

ing an answer and a question. Every time that a ques-

tion is asked by a legal expert and an answer is pro-

vided by a contractor, an agreement takes place. The

contract and agreement are always associated with a

risk that is deﬁned by the legal expert. The risk, as

well as all the additional constituent parts of the con-

tract, can be reviewed by the contractor before sign-

ing the contract. He/she/it is ultimately in charge to

decide whether or not to enter in a legally binding

agreement with the other involved party or parties.

To design the vocabulary terms of the ontology

described above, we used the Resource Description

Framework Schema (RDF/S) and Web Ontology Lan-

guage (OWL). Onassis reuses the following exter-

nal Vocabularies: (1) the Friend of a Friend Ontol-

ogy (FOAF), (2) the Ontology of Units of Measure

(OUM), and (3) the Good Relations Vocabulary (GR).

The logical consistency of the ontology has been

tested by launching the reasoner Hermit 1.4.3.456 on

sample data in Prot

e editor. The use case employed

is presented in the results.

4 RESULTS

The results of our research are twofold. The ﬁrst re-

sult is (4.1) the percentage of available contract au-

tomation solutions relating to the scope of this re-

search based on the Legalpioneer data. The second

result is (4.2) the KG that works as a validation mech-

anism for the ontology.

4.1 Contract Automation Solutions

After the key word search in Legalpioneer the fol-

lowing two results were evident, for (1) contract au-

tomation and (2) contract automation based on com-

munication and risk data. First, given the 905 re-

sults for contract automation out of the total of 9,608

LegalTech solutions, the percentage of contract au-

tomation solutions is 9.4 percent. Second, there are

a total of 7 solutions related to communication and

risk management contract automation, which is 0.7

percent of the total of 905 contract automation so-

lutions. The second ﬁnding is based on the fol-

lowing three results. (A) Six companies are de-

veloping technologies related to contract communi-

cation automation (legislate.tech, lawcus.com, cas-

estatus.com, app4legal.com, lawren.io, and joseﬂe-

gal.com). (B) Two companies are developing contract

risk automation technology (sirionlabs.com and icer-

tis.com). (C) Two companies are merging commu-

nications and risk automation but not for contracting

(cognitiveview.com and smarsh.com). The results in-

Towards a Foundation for Intelligent Contracts

dicate that despite the abundance of contract automa-

tion solutions, there is a signiﬁcant omission for solu-

tions which are focused on communications and risk

data analysis.

4.2 Knowledge Graph

To validate the coherency of the ontology with the do-

main knowledge, we ran competency questions on the

instance data that we structured via the vocabulary

terms of the Onassis ontology. The process demon-

strates the level of expressiveness of the vocabulary.

At present, Onassis can support the following use case

scenario, the visualization of which is presented in

Figure 2 and can be accessed via the GitHub page.

Figure 2: Visualization of the use case scenario modeled

with Onassis’s ontological expressiveness. In the ﬁgure, in-

dividuals are represented as rectangles. Their associated

datatype values are highlighted in blue. Relationships are

represented as arrows.

In the use case scenario, Laura, who is seen as a le-

gal expert, deﬁnes both the scope of the agreement to

be a freelance design agreement and the risk (which in

this case is a payment risk). Successively, she writes

the question “what is the budget?” and waits for an

answer. Atanasio, as a contractor, selects the scope

of the agreement to be a freelance design agreement.

Subsequently, he indicates his budget (500 euros) as

an answer. The answer provided by Atanasio up-

dates the variable uniquely identiﬁed by the number

124590456, related to paragraph 19078956, of sec-

tion 1234509, of contract 345698. The variable, para-

graph and section have an order number in their re-

https://github.com/onassisontology/onassisontology

lated parts. The question asked by Laura and the an-

swer provided by Atanasio are involved in the agree-

ment number 123456, which can have a maximum

of one question and one answer. In fact, for every

question asked and answer given, a uniquely identi-

ﬁed agreement is created. Multiple agreements can

be part of a same uniquely identiﬁed contract. Once

Atanasio has reviewed the legal document he can sign

it by adding his signature on the contract. This action

will legally bind the parties involved in the various

agreements connected to the same scope within a sole

contract. The data shows that our ontology structure

is robust for the requirements of a freelance agree-

ment.

5 DISCUSSION

The discussion concentrates on analyzing (A) the

database ﬁndings, (B) the ontology engineering, (C)

the AI applications, (D) iContracts, and (E) the added

beneﬁts.

5.1 Database Findings

The ﬁnding that 9.4 percent of the total LegalTech so-

lutions focus on contract automation proves how sub-

stantial innovation in contracting exists. Yet, the ﬁnd-

ing that only 0.7 percent of those contract automation

solutions focus on the automation of communication

and risk data, proves how far we still are from adopt-

ing mature intelligent contracting solutions.

A general comment on the inspected data is that

most of the technologies investigated address the legal

experts as users and not the contractors as users. Re-

garding the contract communication automation solu-

tions, none of the them generates the contracts as an

output automatically. As for the contract risk automa-

tion solutions, even though legal risk is part of ev-

ery contracting process, such solutions are not widely

available. Most risk-related solutions identiﬁed relate

to compliance automation.

5.2 Ontology Engineering

Looking back at the pricing example of our research

introduction, the ontology can help contractors spec-

ify an optimum pricing in balance with a normative

speciﬁcation of the qualitative expectation for both

parties. As presented, this can be done by ﬁnding a

middle ground between two sets of answers that the

contractors have provided. In this way, the risk be-

tween two parties for a dispute is minimized, as well

as the potential consequential costs for both of them.

ICAART 2023 - 15th International Conference on Agents and Artiﬁcial Intelligence

The ontology is novel on a legal level as it rep-

resents how a mature legal expert handles contracts.

This has not done before according to literature on

ontological representation level. In that respect, the

ontology is novel as it provides a practical tool for the

legal world rather than providing just another theory.

With the rise of LegalTech, the production of tools

in academic research are becoming more common.

This research further demonstrates the need for such

practical tools. Moreover, the ontology adds value

for both smart contracts and iContracts, as it demon-

strates the extent to which certain processes can be

programmed and those which cannot.

The novelty of the ontology is that it has conceptu-

alized a new domain, which beneﬁts the world of on-

tology at a vocabulary level. In extension, this novelty

is relevant for semantics, as it demonstrates how the

semantics of contract automation work at this level

of conceptualisation. The end-value of the ontology

should be examined by an experimental view in fu-

ture research, in particular from two perspectives. The

ﬁrst perspective needs to ensure that the activities in-

volving the legal expert are designed in a trustworthy

manner. The second perspective needs to ensure that

the activities involving the contractors are designed

in a trustworthy manner. Only then it is possible to

validate the design of the ontology and justify its ap-

plication in real-life experimental use cases.

By making the ontology publicly available we

achieve two levels of innovation. First, we can stay

connected with state-of-the-art developments as the

feedback based on iterations increases as opposed to

if this ontology would stay behind closed doors. It

also becomes more accessible to the public and its de-

velopment becomes more academically trustworthy.

Second, the open source model goes against the mo-

nopolistic power that some corporations follow in in-

novation, which slows general innovation and social

progress.

The present research is relevant for three reasons.

First, the Onassis ontology can provide a framework

for managing risks in contract automation in a trust-

worthy manner as well as preventing contract dis-

putes. Second, it can pave the way for demonstrating

how it is possible to standardise contract drafting lan-

guages in order for contracting to become more inter-

operable. Third, it can maximise the value contractors

extract from contract automation via the application

of AI in a more trustworthy manner than the available

technologies, due to making explicit an analysis pro-

cess which is usually implicit.

After the ontology has been validated, its potential

value can inﬂuence contract automation signiﬁcantly.

Legal experts may no longer be involved physically

in contracting processes between two or more parties;

their interaction may only occur using a computer.

The contractors will be able to: (1) obtain a contract

more rapidly, and (2) trust its content, without having

to enter into extensive discussions.

5.3 AI Applications

The value of the ontology for AI is that it reduces

complexity and helps clarify how advanced ML algo-

rithms can be applied. Moreover it can help make the

algorithmic results explainable. Still, in some cases

involving data, such as risk data, bias is present and

should be addressed.

At this point, three relevant AI applications that

can be implemented with the ontology to achieve a

higher degree of automation are presented. (A) Text

extraction can be used to automatically extract the an-

swers of the contracts from the questions. (B) Data

extraction can also be used to automatically extract

risks for a speciﬁc contract. (C) Text generation can

be used to draft a contract based on extracted risk

data. Based upon this, we can see how ML algorithms

can also applied for classiﬁcation or prediction.

As a result of AI applications, certain analytic

beneﬁts may arise as well. In general, an ontology

can be used to analyse qualitative theory quantita-

tively. Moreover, following the same pricing exam-

ple, we can quantify what is more precise or faster:

the traditional contracting process, the programmed

smart contract process, or the hybrid intelligent con-

tract process? Also, nuanced data analytics can pro-

vide insights that can be used in order to quickly prove

which party is at fault in case of a payment dispute re-

garding quality.

In addition, there are also beneﬁts at the level of

logical reasoning. We can deﬁne a set of rules for re-

curring entities, so as to examine which classes are the

best candidates for codependency inﬂuencing such re-

lations, and apply advanced reasoning to uncover hid-

den data or further relations. The value of this infer-

ential reasoning is that it can support automated rea-

soning for automated dispute resolution.

5.4 Intelligent Contracts

In the same way that smart contracts began with cryp-

tocurrencies and are now applied in more use cases,

the iContract literature should gradually expand into

more directions as well. In the present research, we

make a ﬁrst attempt in demonstrating how iContracts

apply in a freelance project. Moreover, further sci-

entiﬁc examination of the iContracts concept should

occur to help specify their general value for Legal-

Towards a Foundation for Intelligent Contracts

Tech, as well as for AI. A valuable addition that iCon-

tracts bring in contract automation is that they point in

the direction of monitoring during the contract execu-

tion stage. It is in fact vital that for a higher degree

of automation to be achieved, project data should be

connected with iContracts. By expanding the scope

of contract automation during the execution phase,

the management of risks also improves. By connect-

ing iContracts with realistic project execution, this

higher effectiveness in contract execution can also be

achieved. This is where PPL, and in particular le-

gal visualisation, can help due to the lack of sufﬁ-

cient frameworks. Adding on that, the ontology can

help by standardising data classes while a more har-

monised approach can be taken for the classiﬁcation

and collection of such data. To that end, more re-

search in the ﬁeld of how iContracts can beneﬁt from

IoT devices, as well as how they connect more gener-

ally with project management, would be useful.

5.5 Added Beneﬁts

One of the main beneﬁts for iContracts automation for

complex projects relates with insurance premiums. In

general, by having better risk predictions, insurances

can be provided with more accuracy and the premi-

ums calculated more realistically. This has a direct

effect on the operational expenses of organisations. It

also has an effect on the policy choices they make,

as by being able to better measure contract risk, im-

proved policy-making is possible. Moreover, in such

projects where there is often complex contractor and

sub-contractor relationships, the main contract ends

up bearing the major risks; by improving iContracts

from the perspective of risk management, there are

added beneﬁts for the main contractors.

Last but not least, risk frameworks are not stan-

dardised in legal practice as they are, for example,

in the energy or ﬁnance sectors. That is potentially

because the underlying legal practice is already suf-

ﬁciently complex. iContracts help as they can cre-

ate the space for responsible risk management based

on validated frameworks, and by reducing workload.

Currently the proactiveness of contracts is not mea-

sured, so iContracts can also help with quantifying

risk.

6 CONCLUSION

The below Subsections provide the answer to the RQ

(6.1), further research suggestions (6.2) and the re-

search novelty (6.3).

6.1 Answer to the RQ

The paper progresses the state-of-the-art in ontologies

for the legal domain by providing an approach for

contract automation based on communications and

risk data. The RQ of this research is To what extent

is it possible to deﬁne an ontology to sufﬁciently au-

tomate contracts based on communications and risk

data?.

The answer to the RQ is that deﬁning an ontology

for this purpose is possible for sufﬁcient automation.

However, it remains essential to test extensively its

validity, and to conduct further research to ensure an

adequate level of trustworthiness for any action the

legal expert and contractors will be involved in.

Moreover, the ﬁnding that only 0.7 percent of

the total contract automation solutions in the Legal-

pioneer database concentrate on automating contract

communications and risk data demonstrate a signif-

icant omission in existing solutions. This omission

justiﬁes our scientiﬁc attention to the subject. The

aims for this research was to bridge the gap be-

tween smart contracts and iContracts and to clarify

our stance. All in all, we may conclude that automat-

ing a contract based on communications and risk pro-

cesses, which have long been neglected, can prove

to be the missing link in realising both self-executing

contracts and iContracts.

6.2 Further Research

The key question at this point is how to best move

forward from here. An essential step in conducting

further research is to understand the correlation of

the ontology classes. By selecting certain correlated

classes, we may conduct speciﬁc quantitative or qual-

itative experiments to further our research. Based on

the aforementioned discussion, the communications

and risk processes need to be examined more deeply.

Therefore, our follow-up research will focus (1) on

the legal expert-based inputs and outputs , (2) on the

contractor-based inputs and outputs (responsible con-

tract communication management), and (3) on the ex-

perimental validity of the KG. Through this exami-

nation and validation, our ontology may be improved

upon and expanded.

6.3 Research Novelty

An intriguing question that arose while conducting

this research was: to what extent is our research in-

novative? And further, to what extent is it likely for

iContracts to become a separate scientiﬁc discipline

for study? Let us use as a benchmark the discussion

ICAART 2023 - 15th International Conference on Agents and Artiﬁcial Intelligence

around bits and atoms and say that innovation at a

bit level is incremental, while innovation at an atomic

level is fundamental. Is this paper innovative in bits

or atoms? So far none of the examined technologies

points to any atom, and therefore we have a collec-

tion of bits. But, what if most innovation is based

upon bits and not upon atoms? Matt Ridley discusses

this topic in his book How Innovation Works. He con-

cludes that innovation in atoms is usually a collection

of multiple innovations in bits. We hope that our work

follows along this line of reasoning and it can set the

stage for a paradigm shift as to how the AI and Legal-

Tech world can jointly work on contract automation.

To conclude, this research article began by giving

you a promise but by the end of it, we hope to have

provided a real gift: a systematic way to study con-

tract automation and to achieve the goal of iContracts.

ACKNOWLEDGEMENTS

Georgios is the main author. Athanasios contributed

in the literature; and Giulia in the method and result.

Jaap and Bart are the main supervisors. Laura and

Theoﬁlos provided expert feedback.

REFERENCES

World wide web consortium (w3c). https://www.w3.org/C

onsortium/. Accessed: 1-8-2022.

Aimin, D. and Yunfeng, L. (2019). “intelligent factor-

ing” business model and game analysis in the sup-

ply chain based on block chain. Management Review,

31(9):231.

Barton, T. D. (2002). Preventive Law: A Methodology for

Preventing Problems. National Centre fro Preventive

Law.

Barton, T. D., Haapio, H., Passera, S., and Hazard, J. G.

(2019). Successful contracts: integrating design and

technology. In Legal Tech, Smart Contracts and

Blockchain, pages 63–91. Springer.

Berger-Walliser, G. (2012). The past and future of proactive

law: an overview of the development of the proactive

law movement. Proactive Law in a Business Environ-

ment, Gerlinde Berger-Walliser and Kim Østergaard

(eds.), DJØF Publishing, pages 13–31.

Berger-Walliser, G., Barton, T. D., and Haapio, H. (2017).

From visualization to legal design: a collaborative and

creative process. Am. Bus. LJ, 54:347.

Bleker, S. and Hortensius, D. (2014). Iso 19600: The devel-

opment of a global standard on compliance manage-

ment. Business compliance, 2:1–12.

Brown, L. M. and Rubin, E. (1950). Manual of preventive

law. Prentice-Hall.

Corrales, M., Fenwick, M., and Haapio, H. (2019a). Dig-

ital technologies, legal design and the future of the

legal profession. In Legal tech, smart contracts and

blockchain, pages 1–15. Springer.

Corrales, M., Fenwick, M., and Haapio, H. (2019b). Legal

Tech, Smart Contracts and Blockchain. Springer.

Dauer, E. A. (1987). Preventive law: Above all else, predict

what people will do. Preventive L. Rep., 6:9.

Dhani, J. S., Bhatt, R., Ganesan, B., Sirohi, P., and

Bhatnagar, V. (2021). Similar cases recommenda-

tion using legal knowledge graphs. arXiv preprint

arXiv:2107.04771.

Duan, Y., Shao, L., Hu, G., Zhou, Z., Zou, Q., and Lin, Z.

(2017). Specifying architecture of knowledge graph

with data graph, information graph, knowledge graph

and wisdom graph. In 2017 IEEE 15th International

Conference on Software Engineering Research, Man-

agement and Applications (SERA), pages 327–332.

IEEE.

Education, I. C. Knowledge graph. https://www.ibm.com/

cloud/learn/knowledge-graph. Accessed: 1-8-2022.

Eggleston, K., Posner, E. A., and Zeckhauser, R. (2000).

The design and interpretation of contracts: why com-

plexity matters. Nw. UL Rev., 95:91.

Esayas, S. and Mahler, T. (2015). Modelling compliance

risk: a structured approach. Artiﬁcial Intelligence and

Law, 23(3):271–300.

Feilmayr, C. and W

oß, W. (2016). An analysis of ontologies

and their success factors for application to business.

Data & Knowledge Engineering, 101:1–23.

Filtz, E. (2017). Building and processing a knowledge-

graph for legal data. In European Semantic Web Con-

ference, pages 184–194. Springer.

Garc

ıa, R. and Gil, R. (2008). A web ontology for copy-

right contract management. International Journal of

Electronic Commerce, 12(4):99–114.

Group, O. W. Web ontology language (owl). https://www.

w3.org/OWL/. Accessed: 1-8-2022.

Group, R. W. Resource description framework (rdf). https:

//www.w3.org/RDF/. Accessed: 1-8-2022.

Gruber, T. R. (1993a). A translation approach to portable

ontology speciﬁcations. Knowledge acquisition,

5(2):199–220.

Gruber, T. R. (1993b). A translation approach to portable

ontology speciﬁcations. Knowledge Acquisition,

5:199–220.

Guarino, N., Oberle, D., and Staab, S. (2009). What is an

ontology? In Handbook on ontologies, pages 1–17.

Springer.

Haapio, H. and J, Siedel, G. (2013). A short guide to con-

tract risk.

Haapio, H. and Siedel, G. J. (2010). Using proactive law

for competitive advantage. American Business Law

Journal, Volume 47, Issue 4, pages 641–686.

Haapio, H. and Varjonen, A. (1998). Quality improvement

through proactive contracting: contracts are too im-

portant to be left to lawyers! In ASQ World Confer-

ence on Quality and Improvement Proceedings, page

243. American Society for Quality.

Towards a Foundation for Intelligent Contracts

Hazard, J. and Haapio, H. (2017). Wise contracts: smart

contracts that work for people and machines. In

Trends and communities of legal informatics. Pro-

ceedings of the 20th international legal informatics

symposium IRIS, pages 425–432.

Huizing, A., Veenman, C., Neerincx, M., and Dijk, J.

(2020). Hybrid ai: The way forward in ai by develop-

ing four dimensions. In International Workshop on the

Foundations of Trustworthy AI Integrating Learning,

Optimization and Reasoning, pages 71–76. Springer.

Ian Horrocks, Peter F. Paterl-Scnheider, H. B. S. T. B. G.

M. D. Swrl: A semantic web rule language combining

owl and ruleml. https://www.w3.org/Submission/SW

RL/. Accessed: 1-8-2022.

ISO (2020). Iso 31022 risk management – guidelines for

the management of legal risk. http://www.iso.org.

Iversen, J. (2004). Legal risk management. Thomson Gad

Jura.

Kaltenboeck, M., Boil, P., Verhoeven, P., Sageder, C.,

Montiel-Ponsoda, E., and Calleja-Ib

nez, P. (2022).

Using a legal knowledge graph for multilingual com-

pliance services in labor law, contract management,

and geothermal energy. In Technologies and Applica-

tions for Big Data Value, pages 253–271. Springer.

Kolvart, M., Poola, M., and Rull, A. (2016). Smart con-

tracts. In The Future of Law and etechnologies, pages

133–147. Springer.

Kozlova, M. Y. and Aleksandrina, M. (2020). “smart

contracts” vs legal technology in contract practice.

In Institute of Scientiﬁc Communications Conference,

pages 1204–1212. Springer.

Kruijff, J. d. and Weigand, H. (2017). Ontologies for

commitment-based smart contracts. In OTM Con-

federated International Conferences” On the Move

to Meaningful Internet Systems”, pages 383–398.

Springer.

Larson, E. J. (2021). The myth of artiﬁcial intelligence. In

The Myth of Artiﬁcial Intelligence. Harvard University

Press.

Macmullen, W. J. (2005). Annotation as process, thing, and

knowledge: Multi-domain studies of structured data

annotation. In ASIST Annual Meeting. Citeseer.

Mahler, T. (2007). Deﬁning legal risk. SSRN, page 28.

Mason, J. (2017). Intelligent contracts and the construction

industry. Journal of Legal Affairs and Dispute Resolu-

tion in Engineering and Construction, 9(3):04517012.

McCarthy, J. (2004). What is artiﬁcial intelligence. URL:

http://www-formal. stanford. edu/jmc/whatisai. html.

McNamara, A. (2020). Automating the chaos: Intelligent

construction contracts. In Smart Cities and Construc-

tion Technologies. IntechOpen.

McNamara, A. and Sepasgozar, S. (2018). Barriers and

drivers of intelligent contract implementation in con-

struction. Management, 143:02517006.

McNamara, A. J. and Sepasgozar, S. M. (2020). Devel-

oping a theoretical framework for intelligent contract

acceptance. Construction innovation.

McNamara, A. J. and Sepasgozar, S. M. (2021). Intelli-

gent contract adoption in the construction industry:

Concept development. Automation in construction,

122:103452.

Montiel-Ponsoda, E., Gracia, J., and Rodr

ıguez-Doncel, V.

(2018). Building the legal knowledge graph for smart

compliance services in multilingual europe. In CEUR

workshop proc., number ART-2018-105821.

Paulheim, H. (2017). Knowledge graph reﬁnement: A sur-

vey of approaches and evaluation methods. Semantic

web, 8(3):489–508.

Pillai, M. and Adavi, P. (2013). Intelligent contract manage-

ment. International Journal of Scientiﬁc and Research

Publications, 3(1).

Sarker, M. K., Schwartz, J., Hitzler, P., Zhou, L., Nadella,

S., Minnery, B., Juvina, I., Raymer, M. L., and

Aue, W. R. (2020). Wikipedia knowledge graph

for explainable ai. In Iberoamerican Knowledge

Graphs and Semantic Web Conference, pages 72–87.

Springer.

Schneider, J. M., Rehm, G., Montiel-Ponsoda, E.,

Rodr

ıguez-Doncel, V., Mart

ın-Chozas, P., Navas-

Loro, M., Kaltenb

ock, M., Revenko, A., Karam-

patakis, S., Sageder, C., et al. (2022). Lynx: A

knowledge-based ai service platform for content pro-

cessing, enrichment and analysis for the legal domain.

Information Systems, 106:101966.

Smits, J. M. (2017). Contract law: a comparative introduc-

tion. Edward Elgar Publishing.

Sovrano, F., Palmirani, M., and Vitali, F. (2020). Le-

gal knowledge extraction for knowledge graph based

question-answering. In Legal Knowledge and Infor-

mation Systems, pages 143–153. IOS Press.

Stark, T. L. (2013). Drafting contracts: How and why

lawyers do what they do. Wolters Kluwer.

Szabo, N. et al. (1994). Smart contracts.

Timmer, I. (2019). Contract automation: Experiences from

dutch legal practice. In Legal Tech, Smart Contracts

and Blockchain, pages 147–171. Springer.

Vinay K. Chaudhri, Naren Chittar, M. G. An introduction

to knowledge graphs. http://ai.stanford.edu/blog/intro

duction-to-knowledge-graphs/. Accessed: 6-7-2022.

Wu, Y. (2020). Summary of research on contract risk man-

agement of epc general contracting project—based on

vosviewer knowledge graph analysis. In International

Symposium on Advancement of Construction Manage-

ment and Real Estate, pages 1043–1058. Springer.

Yu, H., Li, H., et al. (2021). A knowledge graph con-

struction approach for legal domain. Tehni

cki vjesnik,

28(2):357–362.

Zheng, K., Zhang, Z., and Gauthier, J. (2020). Blockchain-

based intelligent contract for factoring business in

supply chains. Annals of Operations Research, pages

1–21.

Zhou, X., Lim, M. Q., and Kraft, M. (2020). A

smart contract-based agent marketplace for the j-park

simulator-a knowledge graph for the process industry.

Computers & Chemical Engineering, 139:106896.

ICAART 2023 - 15th International Conference on Agents and Artiﬁcial Intelligence