TYPE AND SCOPE OF TRUST RELATIONSHIPS IN
COLLABORATIVE INTERACTIONS IN DISTRIBUTED
ENVIRONMENTS
Weiliang Zhao
Vijay Varadharajan
George Bryan
School of Computing and Information technology
University of Western Sydney
NSW 1797, Australia
Department of Computing
Macquarie University
NSW 2109, Australia
Keywords:
Trust Relationship, Classification of Trust, Trust Direction, Trust Symmetry, Trust Scope.
Abstract:
In this paper, we consider the modelling of trust relationships in distributed systems based on a formal math-
ematical structure. We discuss different forms of trust. In particular, we address the base level authentication
trust at the lower layer with a hierarchy of trust relationships at a higher level. Then we define and discuss trust
direction and symmetric characteristics of trust for collaborative interactions in distributed environments. We
define the trust scope label in order to describe the scope and diversity of trust relationship under our taxon-
omy framework. We illustrate the proposed definitions and properties of the trust relationships using example
scenarios. The discussed trust types and properties will form part of an overall trust taxonomy framework
and they can be used in the overall methodology of life cycle of trust relationships in distributed information
systems that is currently in the process of development.
1 INTRODUCTION
Trust is a very important concept in today’s distrib-
uted systems. In the security world, trust was first
used in trusted systems (TCSEC, 1985) and trusted
computing (Landauer et al., 1989). Marsh (Marsh,
1994) has tried to formalize trust as a computational
concept. In reputation-based systems (Wang and Vas-
sileva, 2003; Xiong and Liu, 2003), the community-
based reputation is used to evaluate trust and predict
the future behaviors of involved peers. Recently there
have been several pieces of work on trust negotiations
(Huhns and Buell, 2002; Winsborough et al., 2000;
Winslett et al., 2002). The main concern of trust ne-
gotiations is how to establish trust between entities in
different security domains by means of cryptographi-
cally signed credentials. Another issue that has been
addressed over the recent years is trust management.
Several automated trust management systems have
been proposed and implemented. Typical trust man-
agement systems include PolicyMaker (Blaze et al.,
1996), KeyNote (Blaze et al., 1999) and REFEREE
(Chu et al., 1997). For XML based web services,
two versions of WS-Trust have been proposed (Della-
Libera and et al, 2002; Anderson and et al, 2004).
Our main objective is to develop a sound under-
standing of trust relationships and to create a power-
ful set of tools to model the trust relationships for col-
laboration in distributed systems. In our earlier work
(Zhao et al., 2004), we have outlined a formal def-
inition of trust relationship and a set of definitions
for modelling trust relationships. Our previous re-
search only provides a starting point for the analysis
and design of trust relationships. The classification
of trust has not been addressed and the properties of
trust have not been discussed. In this paper, we de-
scribe different forms of trust relationships under the
taxonomy framework. We categorize the trust types
into two layers and discuss the hierarchy of trust rela-
tionships under the taxonomy framework. We provide
several definitions about properties of trust direction
and symmetry between involved entities. In order to
describe and analyze the scope and diversity of trust
relationships, we provide a definition of trust scope
label. The definitions in this paper provide new ele-
ments of the whole taxonomy framework. We believe
that our definitions of the properties of trust relation-
ships can provide the accurate terminologies and can
be used in scenarios where the trust relationships are
analyzed and modelled.
331
Zhao W., Varadharajan V. and Bryan G. (2005).
TYPE AND SCOPE OF TRUST RELATIONSHIPS IN COLLABORATIVE INTERACTIONS IN DISTRIBUTED ENVIRONMENTS.
In Proceedings of the Seventh Inter national Conference on Enterprise Information Systems, pages 331-336
DOI: 10.5220/0002520003310336
Copyright
c
SciTePress
The remainder of the paper is organized as follows.
In section 2, we provide the definition of trust rela-
tionship. In section 3, we discuss different forms of
trust and the hierarchy of trust relationships under our
taxonomy framework. In section 4, we provide a set
of definitions for trust direction and symmetric prop-
erties of trust relationships. We employee the Mi-
crosoft’s domain trust as a regressive scenario exam-
ple to illustrate the definitions in this section. In sec-
tion 5, the scope and diversity of trust relationships
are discussed. The definition of trust scope label and
comparison rules between trust scope labels are pro-
posed and a scenario example is provided. Finally
section 6 provides some concluding remarks.
2 DEFINITION OF TRUST
RELATIONSHIP
In our previous work (Zhao et al., 2004), we have pro-
vided a formal definition of trust relationship with a
strict mathematical structure. This definition of trust
relationship is the cornerstone of our trust notion and
the trust taxonomy framework. In this section, we will
provide the details of the definition of trust relation-
ship. The definition of trust relationship is expressed
as:
Definition 1 A trust relationship is a four-tuple
T =< R, E, C, P > where:
R is the set of trusters. It contains all the involved
trusters. It is a non empty set.
E is the set of trustees. It contains all the involved
trustees. It is a non-empty set.
C is the set of conditions. It contains all condi-
tions (requirements) for the current trust relation-
ship. Normally, a trust relationship has some spec-
ified conditions. If there is no condition, the condi-
tion set is empty.
P is the set of properties. The property set de-
scribes the actions or attributes of the trustees. It
is a non-empty set. The property set can be divided
into two sub sets:
Action set: the set of actions that the trusters
trust that trustees will and can perform.
Attribute set: the set of attributes that trusters
trust that trustees have.
The above formal definition of trust relationship has
a strict mathematical structure and a broad expressive
power. It can reflect the commonly used notions of
trust and provides a taxonomy framework. When trust
relationships are used, the full syntax (four-tuple <
R, E, C, P > must be followed. Trust relationship
T means that under the condition set C, truster set R
trust that trustee set E have the properties in set P .
The above definition of the trust relationship is the
basis of all properties of trust discussed in this pa-
per. In our previous work (Zhao et al., 2004), there
are more definitions, propositions and operations for
modelling and analyzing of trust relationships in dis-
tributed systems.
3 CLASSIFICATION OF TRUST
Grandison et al (Grandison and Sloman, 2000) have
given a bottom-up classification and used the terms as
resources access trust, service provision trust, certi-
fication trust, delegation trust and infrastructure trust.
We will illustrate them using our definition of trust re-
lationship. Under our taxonomy framework, we cat-
egorize these trust relationships into two layers and
provide the hierarchy of trust relationships.
Resources Access Trust: Resources access trust
relationship is a kind of trust relationship for the
purpose of accessing resources. The access con-
trol has been the central concern of security for
many decades. The trust relationship can be refined
into authorization policies that specify actions the
trustee can perform on the truster’s resources and
constraints that apply, such as the time periods for
which the access is permitted. With the syntax of
formal definition of trust relationship, resource ac-
cess trust will be like “the trusters trust trustees un-
der some conditions that trustees have the right to
get access to some of trusters’ resources”.
Services Provision Trust: services provision trust
describes trusters’ trust in provided services or re-
sources of trustees. It is related to protection from
maliciously or unreliably provided services or re-
sources. With the syntax of formal definition of
the trust relationship, service provision trust will
be like “the trusters trust trustees under some con-
ditions that trustees will provide the claimed ser-
vices”.
Certification Trust: Certification trust is based on
certification of the trustworthiness of the trustee by
a third party. Certification trust is related to a spe-
cial form of service provision trust. Certification
authority is in fact providing a trust certification
service. With the syntax of formal definition of
trust relationship, certification trust will be like that
“trusters trust trustees if trustees can provide cer-
tificates that trustees have a set of attributes or can
do a set of actions according the certificates”. The
related service provision trust of certification trust
will be like that “trusters trust certification author-
ity under some conditions that the certification au-
thority will only give certificates to suitable enti-
ties”.
ICEIS 2005 - INFORMATION SYSTEMS ANALYSIS AND SPECIFICATION
332
Delegation Trust: Delegation trust is a special
form of service provision trust. With the syntax
of formal definition of trust relationship, delegation
trust will be like that “trusters trust trustees under
some conditions that trustees can make decisions
on trusters’ behalf, with respect to resources or ser-
vices that the trusters own or control” (Ding and
Petersen, 1995).
Infrastructure Trust: Infrastructure trust is a kind
of trust that trusters trust some base infrastructure
under some conditions (Abrams, 1995; Abrams
and Joyce, 1995). With the syntax of formal def-
inition of trust relationship, infrastructure trust will
be like that “trusters trust base infrastructure under
some conditions for a set of properties of the in-
frastructure (some actions and attributes)”.
All the above trust types must build on a more ba-
sic trust relationship which is the authentication trust
or identity trust. Authentication trust is “trusters trust
trustees under some condition that trustees are what
they are claimed”. Authentication trust belongs to a
separate layer and all other trust types belong to an-
other layer above it. This is illustrated in Figure 1.
Note that trust types of layer two may not be neces-
sarily specified in terms of an identity. Anonymous
authorization belongs to access trust and it is an ex-
ample that there is no specified identity. Anonymous
authorization can be implemented using certificates
with capabilities. The real identity of the involved
trustee will not be revealed. For example, a customer
has a certificate for accessing some resources on the
Internet. The customer’s behaviors of accessing the
resources can be recorded. If it is desirable that the
customer cannot be identified, the related access trust
is a kind of anonymous access trust. Particularly for
the resource access trust and service provision trust,
the anonymous authentication is desirable in some
cases. In such a situation, the layer of authentica-
tion still needs to provide a mechanism to deal with
the same entity as the trustee in the whole scope of
the trust process. Normally, there is a temporary and
dynamic identification which will be uniquely con-
nected with the involved trustee in the scope of the
trust process. At layer two, trust relationships can
Figure 1: Trust Layers
be classified in different ways. In the following, we
will give another kind of classification which is dif-
ferent from the bottom-up classification of Grandison
et al. Based on strict definition of trust relationship,
trust relationships at layer two can be classified ac-
cording to the nature of the trustees in trust relation-
ship < R, E, C, P >. If E is an infrastructure,
the trust relationship belongs to infrastructure trust. If
E is not an infrastructure, the trust relationship be-
longs to non-infrastructure trust. Non-infrastructure
trust relationships can be classified based on the own-
ership of the property set. If the trusters have the
ownership of the property set, the trust relationship
belongs to access trust. If the trustees have the own-
ership of the property set, the trust relationship be-
longs to provision trust. If some properties are owned
by trustees and some other properties are owned by
trusters, then the trust relationship belongs to mixture
(A&P) trust. The hierarchy of trust relationships at
layer two is illustrated in Figure 2. In such a classifi-
cation, delegation trust and certification trust are not
independent types. As we have discussed, the delega-
tion trust is a special form of provision trust, trustees
are the providers of delegated decisions on behalves
of trusters. A certification trust can be any subtype
of non-infrastructure trust based on the nature of its
property set.
Figure 2: Trust Hierarchy
4 TRUST DIRECTION AND
TRUST SYMMETRY
The properties of trust direction and trust symmetry
play an important role in the collaborative interactions
between involved entities in distributed information
systems. In this section, we will provide a general de-
scription of the properties of trust direction and trust
symmetry by a set of definitions and an scenario ex-
ample. We believe that these definitions can cover
most situations in the real world and can be used as
standard scenarios for analyzing and modelling direc-
tion and symmetric characteristics of trust. The set
of definitions about trust direction and trust symme-
try are related to each other and they should be coop-
eratively used to analyze and model the properties of
direction and symmetry of trust in distributed environ-
ments. One-way trust relationship, two-way trust re-
lationship and reflexive trust relationship are defined
for the properties of trust direction. For the prop-
TYPE AND SCOPE OF TRUST RELATIONSHIPS IN COLLABORATIVE INTERACTIONS IN DISTRIBUTED
ENVIRONMENTS
333
erties of symmetry of trust relationships, definitions
of symmetric trust relationships, symmetric two-way
trust relationship, and the whole set of trust relation-
ships are provided. The details of the definitions are
described as follows.
Definition 2 One-way trust relationship is the trust
relationship with a unique trust direction from the
trusters to trustees.
One-way is the default feature of a trust relationship
if there is no further description.
Two-way trust relationship can be defined and used
in information systems such as Microsoft’s domain
trust. Actually, two-way trust relationship is the result
of binding two one-way trust relationships together.
We define two-way trust relationship as follows:
Definition 3 Two-way trust relationship T T
is the
binding of two one-way trust relationships T =<
R, E, C, P > and T
=< R
, E
, C
, P
> with
R
= E and E
= R. T and T
are the reflective trust
relationships with each other in the two-way trust re-
lationship.
In the above definition, “binding” is the key word.
If there are two one-way trust relationships between
R and E but they are not bound with each other,
then they are only two one-way trust relationships and
there is no two-way trust relationship. When two one-
way trust relationships are bound together, there is
a two-way trust relationship and these two one-way
trust relationships can be called reflective trust rela-
tionships with each other.
If the trusters and the trustees are the same, the trust
relationship is reflexive. The reflexive trust relation-
ship is defined as follows:
Definition 4 Trust relationships T =<
R, E, C, P > is an reflexive trust relationship
when R = E.
The symmetry of two trust relationships could be an
important concern in the analysis or modelling of trust
relationships in distributed information systems. The
symmetry of two trust relationships is defined as the
follows:
Definition 5 If there is trust relationship T
=<
R
, E
, C
, P
> which is the result of swap-
ping trusters and trustees in another trust relation-
ship T =< R, E, C, P > (the swapping includes
all possible ownerships in condition set and property
set), there is symmetry between T and T
, T and T
are symmetric trust relationships with each other.
In the above definition, the swapping of trusters and
trustees includes all possible ownerships in condi-
tion set and property set. The two trust relation-
ships have the same condition set and property set
except the possible ownerships in them. The symmet-
ric/asymmetric two-way trust relationship is defined
as follows:
Definition 6 A two-way trust relationship T T
is
symmetric two-way trust relationship if there is sym-
metry between T and T
; otherwise T T
is an asym-
metric two-way trust relationship.
Sometimes it is necessary to discuss the symmetry
of all trust relationships between a truster set and a
trustee set, we have the following definition:
Definition 7 WTR(R,E) is the whole set of trust rela-
tionships with same truster set R and trustee set E.
Definition 8 If every trust relationship in WTR(R,E)
has a symmetric trust relationship in WTR(E,R) and
every trust relationship in WTR(E,R) has a symmetric
trust relationship in WTR(R,E), the trust between R
and E are symmetric.
Scenario Example : Here we use Microsoft’s domain
trust as a regressive scenario example to discuss the
properties of trust direction and trust symmetry de-
fined in this section. Domain trust allows users to
authenticate to resources in another domain. Also,
an administrator is able to administer user rights for
users in the other domain. Our general definitions
for the properties of direction and symmetry of trust
relationships have general expressive power and can
cover broad range of commonly used notations. The
related concepts in domain trust can be viewed as spe-
cific cases of these general definitions. In the follow-
ing, we will use our terms defined in this paper to re-
view some concepts in domain trust.
Based on definition 1 in section 2, the domain trust
can be expressed as “entities in domain A trust enti-
ties in domain B without any condition that entities
in domain B have the right to get access of the set
of resources in domain A”.
From our view point of trust classification in sec-
tion 3, Microsoft’s domain trust belongs to re-
sources access trust. Domain trust binds the au-
thentication and authorization together and has a
standard two layer structure described in section 3.
Microsoft’s domain trust includes both one-way
trust and two-way trust. In Microsoft’s domain
trust, one-way trust is defined as a unidirectional
authentication path created between two domains.
This means that in a one-way trust between domain
A and domain B, users in domain A can access re-
sources in domain B. However, users in domain B
cannot access resources in domain A. Microsoft’s
one-way trust is an example of one-way trust rela-
tionship in definition 2. In a two-way domain trust,
authentication requests can be passed between the
two domains in both directions. Two-way trust is
an example of two-way trust relationship in defini-
tion 3.
The entities in same domain trust each other with-
out any condition that entities have the right to get
ICEIS 2005 - INFORMATION SYSTEMS ANALYSIS AND SPECIFICATION
334
access of the set of resources in the same domain.
This is an example of reflexive trust relationship in
definition 4.
There is symmetry in the two-way domain trust.
The two one-way trust relationships bound in the
two-way trust relationship are “entities in domain
A trust entities in domain B without any condition
that entities in domain B have the right to get access
of the set of resources in domain A and “entities
in domain B trust entities in domain A without any
condition that entities in domain A have the right
to get access of the set of resources in domain B”.
These two one-way trust relationships are symmet-
ric trust relationships with each other in definition
5. Microsoft’s two-way trust is symmetric two-way
trust relationship in definition 6.
In domain trust, the W T R(A, B) based on defin-
ition 7 has only one trust relationship from truster
domain A to trustee domain B. For two-way do-
main trust, the trust between domain A and domain
B is symmetric based on definition 8.
The properties of direction and symmetry of trust re-
lationships play an important role in the modelling of
the trust relationships in collaborative interactions in
distributed environments. These definitions are new
elements of the taxonomy framework about trust. We
believe that they can cover most situations related
with direction and symmetry of trust relationship in
the real world.
5 SCOPE AND DIVERSITY OF
TRUST RELATIONSHIP
Scope and diversity are two other aspects related to
the trust relationship. The diversity of trust has been
discussed by Jøsang (Jøsang, 1996) who expresses
trust in three diversity dimensions. The first dimen-
sion represents trusters or trust originators, the second
represents the trust purpose, and the third represents
trustees. Jøsang uses the term trust purpose based on
the observation that trust is relative to a domain of ac-
tions. In our formal definition of trust relationship,
trusters and trustees are two tuples and they are simi-
lar to the terms of Jøsang. The origin diversity about
trusters and target diversity about trustees are straight-
forward and have been described clearly by Jøsang
(Jøsang, 1996). Jøsang’s term of trust purpose is re-
lated to a domain of actions. In this section, we will
define trust scope label to take the place of the trust
purpose. The benefits of trust scope label will be dis-
cussed later in this section. The trust scope label is
based on the four tuples of a trust relationship and it
is the binding of the condition set and property set.
The trust scope label is a new element of our taxon-
omy framework defined as follows:
Definition 9 A trust scope label is a two-tuple
T SL =< C, P > where C is a set of conditions
and P is a set of properties.
The details of condition set C and property set P can
be found in the formal definition of trust relationship
in section 2. Actually, trust scope label provides a
new layer of abstraction under the trust relationship
and it defines the properties of the trust and its asso-
ciated conditions. To compare two trust scope labels
T SL
1
=< C
1
, P
1
> and T SL
2
=< C
2
, P
2
>, we
have the following rules:
1. C
1
C
2
and P
1
P
2
T SL
1
T SL
2
;
2. C
1
= C
2
and P
1
= P
2
T SL
1
= T SL
2
;
3. C
1
C
2
and P
1
P
2
T SL
1
T SL
2
.
4. In other cases, T SL
1
and T SL
2
can not be com-
pared with each other.
The trust scope label is beyond the trust purpose
in several aspects. Trust scope label composes of
a subspace of trust relationships (two tuples out of
four tuples) and describes the characteristics of the
combination of condition set C and property set P .
Trust scope labels could be treated as an independent
subspace of trust relationships in the analysis and
design of overall information systems. The property
set in trust scope label covers not only actions but
also attributes of trustees. Trust scope labels can be
embedded in all the trust types described in section
3 and two trust scope labels could be compared with
each other based on the rules given above.
Scenario Example: Consider an online soft-
ware shop. We assume that anybody who wants to
enter the online shop must register as a member of
the online shop first. For describing the condition
set and property set in possible trust relationships
between the shop and possible customers, we use the
following notations:
p1 stands for that customers can read the documen-
tation of the software.
p2 stands for that customers can download the soft-
ware.
c1 stands for certificate of membership.
c2 stands for the commitment of the payment for
the software.
c3 stands for the payment for the software.
We have the following trust scope labels:
1. T SL1 =< {c1}, {p1} >
2. T SL2 =< {c1, c2}, {p1, p2} >
3. T SL3 =< {c1, c2, c3}, {p1, p2} >
TYPE AND SCOPE OF TRUST RELATIONSHIPS IN COLLABORATIVE INTERACTIONS IN DISTRIBUTED
ENVIRONMENTS
335
Based on the rules to compare two trust scope labels,
we have
T SL1 cannot be compared with T SL2 (or T SL3).
There is no obvious relationship between T SL1
and T SL2 (or T SL3).
T SL2 > T SL3. It means that the trust scope of
T SL2 is less strict than that of T SL3.
In the analysis and modelling of trust relationships,
the trust scope label may be quite complicated and
the above comparison rules provide helpful tools in
making judgements.
6 CONCLUDING REMARKS
Based on the formal definition of trust relationship
with a strict mathematical structure proposed in our
previous work, in this paper, we have focused on the
modelling of trust relationships in collaborative in-
teractions in distributed environments. We have dis-
cussed different forms of trust under the our proposed
taxonomy framework. We believe that authentication
constitutes layer one of trust and it plays a foundation
role for other trust types on layer two. The hierar-
chy of layer two trust relationships proposed is based
on the nature of the four tuples of a trust relation-
ship. This hierarchy provides a bird’s eye view of the
purposes of trust relationships in the real world. The
properties of trust direction and trust symmetry have
been discussed and a set of definitions has been pro-
vided. In real implementations, these properties can
be customized and configured based on the specific
requirements. The trust scope label has been defined
under our taxonomy framework and it could be used
in the analysis of the scope and diversity of trust rela-
tionships.
The proposed properties of trust relationships and
taxonomy framework are currently being used in the
development of the overall methodology of life cycle
of trust relationships in distributed information sys-
tems. We believe that the classification of trust are
helpful for better understanding of the trust in distrib-
uted systems. We believe that the definitions about
trust direction, trust symmetry and trust scope label
provide suitable terms for the related properties and
they can be used as tools for enabling the design and
analysis of trust in collaboration of entities in real sys-
tems.
REFERENCES
Abrams, M. (1995). Trusted system concepts. In V., J. M.,
editor, Computers and Security, pages 45–56.
Abrams, M. and Joyce, M. (1995). Trusted computing up-
date. Computers and Security, 14(1):57–68.
Anderson, S. and et al (2004). Web services trust
language (ws-trust)(version 1.1). http://www-
106.ibm.com/developerworks/library/ws-trust/.
Blaze, M., Feigenbaum, J., and Keromytis, A. (1999).
KeyNote: Trust management for public-key in-
frastructures (position paper). Lecture Notes in Com-
puter Science, 1550:59–63.
Blaze, M., Feigenbaum, J., and Lacy, J. (1996). Decen-
tralized trust management. In Proceedings of IEEE
Symposium on Security and Privacy, pages 164–173.
Chu, Y. H., Feigenbaum, J., LaMacchia, B., Resnick, P., and
Strauss, M. (1997). REFEREE: Trust management for
Web applications. Computer Networks and ISDN Sys-
tems, 29(8–13):953–964.
Della-Libera, G. and et al (2002). Web services trust
language (ws-trust) (version 1.0). http://www-
106.ibm.com/developerworks/library/ws-trust/.
Ding, Y. and Petersen, H. (1995). A new approach for dele-
gation using hierarchical delegation tokens. Technical
report, University of Technology Chemnitz-Zwickau
Department of Computer Science.
Grandison, T. and Sloman, M. (Fourth Quarter, 2000). A
survey of trust in internet application. IEEE Commu-
nications Surveys.
Huhns, M. N. and Buell, D. A. (2002). Trusted autonomy.
Internet Computing, IEEE, 6(3):92–95.
Jøsang, A. (1996). The right type of trust for distributed
systems. In Proceeding of the 1996 New Security Par-
adigms Workshop. ACM.
Landauer, J., Redmond, T., and Benzel, T. (1989). Formal
policies for trusted processes. In Proceedings of the
Computer Security Foundations Workshop II, 1989,
pages 31–40.
Marsh, S. (1994). Formalising trust as a computational
concept. Phd thesis, University of Sterling.
TCSEC (1985). Trusted computer system evaluation crite-
ria. Technical report, U.S.A National Computer Secu-
rity Council. DOD standard 5200.28-STD.
Wang, Y. and Vassileva, J. (2003). Trust and reputation
model in peer-to-peer networks. In Proceedings of
Third International Conference on Peer-to-Peer Com-
puting.
Winsborough, W. H., Seamons, K. E., and et al (2000). Au-
tomated trust negotiation. In Proceedings of DARPA
Information Survivability Conference and Exposition.
Winslett, M., Yu, T., and et al (2002). Negotiating trust in
the web. IEEE Internet Computing, 6(6):30–37.
Xiong, L. and Liu, L. (2003). A reputation-based trust
model for peer-to-peer e commerce communities. In
IEEE International Conference on E-Commerce.
Zhao, W., Varadharajan, V., and Bryan, G. (2004). Mod-
elling trust relationships in distributed environments.
In Lecture Notes in Computer Science, volume 3184,
pages 40–49. Springer-Verlag.
ICEIS 2005 - INFORMATION SYSTEMS ANALYSIS AND SPECIFICATION
336