On the e-Negotiation of Unmatched Logrolling Views
S.C. Cheung1, Patrick C.K. Hung2, Dickson K.W. Chiu3
Department of Computer Science, Hong Kong University of Science and Technology
2
Department of Management Sciences, University of Waterloo, Ontario, Canada
3
Department of Computer Science and Engineering, Chinese University of Hong Kong
email: [email protected], [email protected], [email protected]
1
Abstract
An e-Negotiation process comprises computerfacilitated tasks, each of which aims at resolving an
issue or a collection of co-related issues under
negotiation. It involves an iterative decision process in
which two or more parties make individual decisions
and interact with each other, aiming to arrive at a
contract. In each iterative decision process, a party
makes offers or counter offers that may reflect tradeoffs
of some issues. Tradeoff evaluation often requires
logrolling, which refers to the exchange of loss in some
issues for gain in others. The set of issues involved in a
logrolling process is referred to as a logrolling set. A
typical negotiation involves multiple logrolling
processes. The logrolling sets exercised by a negotiation
party in these processes constitute one logrolling view.
Problems arise when parties have unmatched logrolling
views. Alternative offers or counter offers that refer to
one logrolling view can be difficult to interpret under a
different logrolling view. In this paper, motivated by an
e-Negotiation example of a lease contract template, we
formulate a meta-model of unmatched logrolling views,
examine the involved e-Negotiation process, and
propose a mechanism for building an e-Negotiation
Support System (eNSS) to facilitate the process, without
the need of formulating utility functions. As a result, eNegotiation across parties with unmatched logrolling
views can be streamlined.
1
Introduction
A contract is a binding agreement between two or
more parties, defining the set of obligations and rewards
in a business process. Contracts are important for
attaining business process interoperability and enforcing
their proper enactment, because they reduce uncertainty
associated with the interactions among organizations. In
USA (Robinson 1997), the federal government spends
about USD$200 billion annually buying goods and
services from over 300,000 vendors. A typical
supermarket chain requires negotiating contracts of over
50,000 product items annually. It is obvious that
contract negotiation is a very important activity in
business processes.
Negotiation is a decision process in which two or
more parties make individual decisions and interact with
each other for mutual gain (Thompson 1998). Proposals
are sent to the other parties, and a new proposal may be
generated after receiving a counter proposal. The
process continues until an agreement or a deadlock is
reached, or even one or more parties quit. Each party
needs to determine reactions of the other parties and
obtain their responses. Each party also needs to estimate
the outcomes that the other parties would like to
achieve. Whereas each party has its own utility function
(Stigum and Wenstop 1983), they tend to be ignorant of
the others’ values and strategies, especially in a noncooperative environment. As a result, negotiations may
involve high transaction costs and do not always reach
the best solution.
Negotiation of contracts involves two or more
parties multilaterally bargaining for mutual gain in order
to achieve a mutual beneficial agreement, but each of
them has conflicting interests. The Internet has recently
become a global common platform where organizations
and individuals communicate among each other to carry
out various commercial activities and to provide valueadded services. As many business activities are
automated as electronic transactions, negotiation of
contracts between human becomes a bottleneck. A
major problem of this is its slowness, which is further
complicated by issues of culture, ego and pride
(Thompson 1998). With the growing acceptance of eCommerce, this problem can be much alleviated using
e-Negotiation (Hung and Mao 2002). An e-Negotiation
comprises computer-facilitated tasks, each of which
aims at resolving an issue or a collection of co-related
issues under negotiation. It involves an iterative
decision process in which two or more parties make
individual decisions and interact with each other aiming
to arrive at a contract.
Proceedings of the 36th Hawaii International Conference on System Sciences (HICSS’03)
0-7695-1874-5/03 $17.00 © 2002 IEEE
An e-Contract, which naturally follows a successful
e-Negotiation, is the computerized facilitation or
automation of a contract in a cross-organizational
business process. As current trends in e-commerce
accelerate the widespread use of e-Contracts in the
business world (Nabil and Yesha 1996), the ability of an
e-commerce system to quickly create mutually
beneficial e-Contracts becomes a critical success factor
for organizations. An automated e-Contract negotiation
system should conduct negotiation to create value by
interacting with different parties to create mutually
acceptable deals. This is particularly applicable to
standard business interactions that could take place over
the Internet, such as real-estate transactions, purchase
and sale of goods, etc. New contracts for these business
interactions are typically defined based on standard
contract templates. Specific business interactions not
covered by the clauses in standard contract templates
can be provided as contract variations or contract
escalations. A contract template is the reference
document based on which a new contract is negotiated.
A contract template consists of a number of contract
clauses, each addressing a specific concern in the
business interaction. Each contract clause contains a set
of template variables whose values are to be negotiated.
Figure 1 gives an example of a contract clause in a lease
contract template. The brackets help identify template
variables (i.e., Period, Start Date, Facilities, Mgt Fee and
Rent) in the clause.
1M if the lease starts within one month
Start Date = 2M if the lease starts between one month and two months
3M if the lease starts between two months and three months
Period
…This lease will start on <Date> and last for <Period>
year(s). The property will be rented <with / without>
Facilities
facilities. Tenant is required to pay landlord a monthly rent of
<Rent>. The rent <includes / excludes> the management fee.
…
Rent Mgt Fee
combination of values has to be evaluated by decisionmakers in each e-Negotiation iterative decision process.
The evaluation of a large combination of variables is
time consuming and could be difficult for decision
makers from a cognitive perspective. The partitioning of
issues into logrolling sets allows decisions to be
evaluated and made in a stepwise fashion so that
potential tradeoffs can be effectively recognized. In
addition, Foroughi and Jelassi (1990) have a similar
perspective. A typical negotiation involves multiple
logrolling processes. The logrolling sets exercised by a
negotiation party in these processes constitute one
logrolling view. However, problems arise when parties
have different sets of logrolling views. Alternative
offers or counter offers that refer to one logrolling view
can be difficult to interpret under a different logrolling
view. In this paper, motivated by an e-Negotiation
example based on a lease contract template, we
formulate a meta-model of unmatched logrolling views,
examine the involved e-Negotiation process, and
propose a mechanism for building an e-Negotiation
Support System (eNSS) to facilitate the process, without
the need of utility functions. As a result, e-Negotiation
across parties with different logrolling views can be
streamlined.
The remainder of this paper is organized as follows:
Section 2 describes a motivating example. Section 3
presents a conceptual model for e-Negotiation and eContracts. This is followed by a methodology in Section
4 to conduct tradeoff evaluation with unmatched
logrolling views. Next, Section 5 discusses related
works. Lastly, Section 6 discusses the conclusion and
future work.
2
Motivating Example
Landlord
Period
Figure 1: A Lease Template Example
In each iterative decision process of e-Negotiation, a
party makes offers or counter offers reflecting tradeoffs
of some issues. In the context of contract templates, an
issue corresponds to a template variable (Cheung et al.
2002). Tradeoff evaluation often requires logrolling,
which refers to the exchange of loss in some issues for
gain in others (Tajima and Fraser 1998). The set of
issues (or template variables) involved in logrolling
constitute a logrolling set. For instance, the issues
regarding the amount of deposit and the rent could
constitute a logrolling view if a party offers to exchange
loss in the issue of deposit for gain in the issue of rent.
Referring to the above example, e-Negotiation
involves the evaluation of several variables and of
several possible values per variable. As a result, a large
Mgt Fee
Basic Rent
Mgt Fee
Allowance
Facilities
Allowance
Rent
Rent = Basic Rent + Facilities Allowance + Mgt Fee Allowance
Logrolling Sets
Period
Basic Rent
Start Date
Tenant
Facilities
Facilities Allowance
Mgt Fee
Mgt Fee
Allowance
Rent
Figure 2: A Selection of Template Variables in a
Lease Contract
Figure 2 presents a scenario of negotiation over a
selection of template variables in a lease contract
Proceedings of the 36th Hawaii International Conference on System Sciences (HICSS’03)
0-7695-1874-5/03 $17.00 © 2002 IEEE
Facilities
Start Date
template. Although the motivating example involves
only two parties, the framework we discussed in this
paper is applicable to multiple parties. In this motivating
example, both parties have different tradeoff evaluation
model as reflected by their individual logrolling views.
The upper and lower halves represent the individual
logrolling view of a landlord and a tenant, respectively.
Let us first examine the landlord’s logrolling view. The
primary concern of the landlord is to rent the property
out as soon as possible since it is already vacant.
Therefore, the landlord is willing to exchange a
reduction in rent for an early start of the lease. The issue
Basic Rent and Start Date therefore form a logrolling set.
In each e-Negotiation iterative decision process, this
logrolling set is the first amongst the others to be
evaluated by the landlord. The additional allowance
required for the provision of facilities (such as
refrigerator, air conditioners and washing machines) or
inclusion of management fee can be evaluated with
reference to the Basic Rent. For instance, the landlord
requires an addition of $1,000 to the Basic Rent if
management fee is included. However, the tenant may
concern more on the lease period because of several
reasons. For example, his current employment contract
will be expired in a year. He wants to leave flexibility
for relocation after one year and is willing to exchange a
higher rent for a lease of shorter period. This is modeled
by the logrolling set containing period and Basic Rent.
This logrolling set is the first amongst the others to be
evaluated by the tenant in each e-Negotiation iterative
decision process. In real life, parties tend to insist on
using the evaluation model that they are most familiar,
based on their concerns and past experience. This
necessitates the support of unmatched logrolling views
in e-Negotiations. We say that parties have unmatched
logrolling views if they have different logrolling sets.
For example, the logrolling views of the landlord and
the tenant in the motivating example are unmatched.
The template variables (e.g., Period), which appear in
a lease contract template, are printed in non-italic. The
italic ones (e.g. Facilities Allowance) are referred to as
auxiliary variables, which are introduced to facilitate
tradeoff evaluation during the negotiation process.
These auxiliary variables are to be evaluated together
with the related template variables. As such, a logrolling
set contains both template and auxiliary variables,
which are to be logrolled. Detailed discussion on
auxiliary variables can be found in (Cheung et al. 2002).
Further, we do believe that an e-Contract negotiation
system should also depict an entire e-Contract template
so that negotiators can logroll among template variables,
focusing on tradeoffs among these instead of arguing
about single variables.
3
e-Negotiation of Contracts
Figure 3 presents the e-Negotiation conceptual
model and its relationships to the e-Contract conceptual
model in Unified Modeling Language (UML), which is
a standard language for object-oriented modeling (OMG
2001). A contract may contain a number of template
variables, such as the Rent and Period in a lease contract.
Each negotiation party expresses their reservation prices
of these variables. These prices are to be resolved in
negotiation. Auxiliary variables are not parts of a
contract but introduced to facilitate e-Negotiation and
tradeoff evaluations. There are various kinds of relations
among these variables. For instance, a template variable
precedes another variable if the former is to be
negotiated before the latter. We say that a set of
variables indivisibly relates to each other if they are in
the same logrolling set.
e-Negotiation
Conceptual
Model
Party
1..*
Task
1
1..*
1..*
Issue
Auxiliary
Variable
Template
Variable
1..*
Negotiation
Process
resolves
Contract
1
maps to *
*
Variable
1
makes
1..*
Offer
Accepted Offer
Logrolling
Set
maps to 1
*
Logrolling
View
1..*
Reservation
Price
*
relates to
1
holds
1..*
Accepted Reservation Price
Figure 3: A Conceptual Model for e-Negotiation and
e-Contract
A negotiation process is made up of tasks, each of
which aims at resolving an issue or a collection of corelated issues. For instance, issues in the lease example
are facilities provision, inclusion of management fee,
start dates of the lease and the deposit to be paid by the
tenant. Each of these issues maps to a set of template
and auxiliary variables. For instance, the issue of
facilities provision can be mapped to two variables
Facilities and Facilities Allowance. While Facilities is a
template variable representing the provision of facilities
in a lease, Facilities Allowance is an auxiliary variable to
indicate the additional amount required if facilities are
to be provided in the lease. Facilities Allowance is a part of
Rent. Its introduction helps the landlord to determine the
Rent. In our approach, a party does not need to be aware
of the auxiliary variables introduced by its counterparty.
Proceedings of the 36th Hawaii International Conference on System Sciences (HICSS’03)
0-7695-1874-5/03 $17.00 © 2002 IEEE
involves
2..*
e-Contract
Conceptual
Model
An issue may be logrolled with other issues. For
instance, a landlord may logroll the issues of facilities
provision and lease period if a long lease is the
condition to facilities provision. These logrolling issues
collectively form a logrolling set. Unmatched logrolling
views occur when each party has different combinations
of logrolling sets, as we will discuss in Section 4.
3.1
e-Negotiation Process Model
Figure 4 depicts our e-Negotiation process in UML
activity diagram. Each party has to participate in each
constituting activity of the process. The e-Negotiation
process either leads to a successful creation of an eContract; otherwise leads to nothing. When an eNegotiation process starts, it triggers two parallel
activities. One activity is to identify issues and define
the criteria to be adopted in the e-Negotiation process.
For instance, the landlord may set criteria on the
minimum rent such as $20,000 (in Hong Kong Dollars)
and on the minimum lease period such as one year. The
other activity is to select an appropriate e-Contract
template in the domain of the e-business under
negotiation.
defineissues
andcriteria
introduce
auxiliaryvariables &
specifyrelations
identify
logrollingissues
[inconsistent]
select e-Contract
template
[reachconsensus on
all reservationprices]
creationof
e-Contract
[all issues have
mapissues to beenmapped]
variables
[consistent]
makeoffers &
counter offers
[quit]
validate
consistency
organize
logrollingviews
Figure 4: e-Negotiation Process Model in UML
Activity Diagram
Then, the sets of logrolling issues are identified and
refined with the introduction of necessary auxiliary
variables. Relations between auxiliary variables and
template variables are specified. For example, the
template variable Rent is a sum of three auxiliary
variables Basic Rent, Facilities Allowance and Mgt Fee
Allowance. Issues are mapped to collections of auxiliary
and template variables. Variables to which an issue or a
collection of logrolling issues is mapped constitute a
logrolling set. Each party makes offers and counter
offers in turn to attain a set of acceptable reservation
prices for these variables. In real-life, parties need to
arrive at an agreed negotiation plan before contract
negotiation (Hung and Mao 2002). Furthermore, the
plan, which may not often be captured precisely in oneshot, must be consistent with the inter-relationships
among the variables (Cheung et al. 2002).
4
Resolution of Unmatched Logrolling
Views
As mentioned in Section 2, tradeoff evaluation often
requires logrolling, which is most effective if this can be
conducted based on the logrolling view held by each
individual party. Each logrolling view represents the
tradeoff evaluation model of a negotiation party. To
facilitate evaluation, a party may introduce auxiliary
variables to a logrolling set. For example, three
auxiliary variables (Basic Rent, Facilities Allowance and
Mgt Fee Allowance) are introduced to facilitate tradeoff
evaluation covering basic terms, inclusion of facilities
and inclusion of management fee, respectively. Each of
these auxiliary variables constitutes a part of Rent.
However, unlike template variables, such as Rent,
auxiliary variables do not appear in contract templates
and their introduction is a party’s own decision. As
such, there could be different interpretations on
auxiliary variables across negotiation parties. For
instance, the landlord ties the Basic Rent to the date
when the lease may start, while the tenant ties it to the
period of the lease. In fact, as each party may also have
a different set of auxiliary variables, exchanges of these
auxiliary variable values may not be meaningful.
Unlike auxiliary variables, template variables are
derived directly from a contract template. Negotiation
parties therefore share a common set of template
variables and the interpretation on these variables.
While tradeoffs are evaluated based on a logrolling set,
offers can be exchanged in an aggregated view, which
contains only template variables. As such, e-Negotiation
involves a methodology to aggregate the values
specified in the logrolling sets held by one party and to
dissect these aggregated values into the logrolling sets
held by another party. The mechanism is facilitated by
an e-Negotiation Support System (eNSS). Each party
runs a copy of eNSS, which exchanges aggregated
values with the eNSSs of counter-party as shown in
Figure 5. Let us first revisit the lease contract example
with further details and then elaborate the methodology.
LogrollingSets
e-Negotiation
Support System(eNSS)
1M
1M
Excluded
Excluded
1yr
Included 20K
2 yrs
Excluded 21K
2 yrs
Included 22K
Excluded
Excluded
1yr
Included 21K
2M
Included
2 yrs
Excluded 22K
2M
Included
3M
Excluded
3M
Excluded
1yr
2 yrs
1yr
StartDate Basic Rent
1M
19K
2M
20K
3M
21K
Excluded 19K
1yr
Included
Included
2M
2M
Excluded 20K
Included 23K
Excluded 21K
1yr
Included 22K
3M
Included
2 yrs
Excluded 23K
3M
Included
2 yrs
Included 24K
Facilities Period FacilitiesAllowance
Excluded 1 yr
0
Included 2 yrs
2K
e-Negotiation
Support System(eNSS)
Mgt Fee MgtFeeAllowance
Excluded
0
Included
1K
Negotiation
Party
Figure 5: Exchanging Offers in Different Views
4.1
Logrolling View of Landlord
Figure 6 shows the logrolling view and reservation
prices held by the landlord. Since the property is vacant,
the landlord is eager to let it out as soon as possible and
Proceedings of the 36th Hawaii International Conference on System Sciences (HICSS’03)
0-7695-1874-5/03 $17.00 © 2002 IEEE
1M
1M
Start Date Facilities FacilitiesAllowance
1M Included
2K
2M Included
2K
2M Excluded
0
3M Excluded
0
Mgt Fee Mgt FeeAllowance
Excluded
0
Included
1K
LogrollingSets
Start Date Facilities LeasePeriod Mgt Fee Rent
Period Basic Rent
1yr
18K
2yrs
17K
Negotiation
Party
AggregatedView
offers a monthly Basic Rent as low as 19K if the lease
can start within one month. A higher Basic Rent of 20K
and 21K is required should the lease starts within two
and three months, respectively. On top of the basic
terms, provision of facilities and inclusion of
management fees are possible if the tenant pays an
additional allowance of 2K and 1K, respectively.
However, the landlord requires a lease period of two
years, should provision of facilities are included. As
such, there is no reservation price for the inclusion of
facilities and a lease period of one year. Note that the
landlord does not need to specify reservation prices for
all possible combinations of issues. The three tables
reflect the logrolling view used by the landlord to
evaluate tradeoffs. Each logrolling set is represented in a
table. It should be noted that logrolling sets held by a
party do not have fields in common.
Start Date
1M
2M
3M
Facilities
Excluded
Included
Period
1 yr
2 yrs
Mgt Fee
Excluded
Included
Basic Rent
19K
20K
21K
Facilities Allowance
0
2K
Mgt Fee Allowance
0
1K
variables into template variables based on a set of
relations R.
Aggregated View = fR(L1 × L2 × … × LN)
Start Date
1M
1M
1M
1M
2M
2M
2M
2M
3M
3M
3M
3M
Facilities
Excluded
Excluded
Included
Included
Excluded
Excluded
Included
Included
Excluded
Excluded
Included
Included
4.2
Aggregation of Logrolling Sets
When the landlord is satisfied with the reservation
prices specified at a favorite logrolling view, an
aggregated view can be constructed based on the
Cartesian product of his/her logrolling sets, where the
template variable Rent is a lump sum of the values in the
related auxiliary variables. In general, given a set of N
logrolling sets L1, …, LN, an aggregated view can be
constructed by applying a function fR upon the Cartesian
product of Li, where i = 1, ..., N and fR converts auxiliary
4.3
Logrolling View of Tenant
Period
1 yr
2 yrs
Start Date
1M
2M
2M
3M
Basic Rent
18K
17K
Facilities
Included
Included
Excluded
Excluded
Mgt Fee
Excluded
Included
Facilities Allowance
2K
2K
0
0
Mgt Fee Allowance
0
1K
Figure 8: Three Logrolling Sets and the Associated
Reservation Prices of Tenant
Before examining the mechanism of offer matching,
let us consider the reservation prices at the tenant side.
Suppose the potential tenant prefers a shorter lease to
allow for relocation after a year. He is willing to pay a
higher monthly Basic Rent of 18K for a one-year lease
contract, but only 17K for a two-year period. The tenant
is able to start the lease within a month, should the
landlord provide the facilities. The tenant does not mind
paying facilities allowance up to 2K. The provision of
facilities can be optional if the lease starts within two
Proceedings of the 36th Hawaii International Conference on System Sciences (HICSS’03)
0-7695-1874-5/03 $17.00 © 2002 IEEE
Rent
19K
20K
21K
22K
20K
21K
22K
23K
21K
22K
23K
24K
Hereafter, let us represent the table entries generated
by the negotiation tool eNSS in italic fonts. Figure 7
gives the aggregated view computed from the landlord’s
logrolling sets given in Figure 6 and the relation Rl
above. This aggregated view, representing a landlord’s
offer, is sent to the tenant for evaluation
In addition to specifying reservation prices at the
logrolling sets, a party also needs to define a set of
relations R, mapping auxiliary variables in the logrolling
sets to the associate template variables. In this lease
example, the template variable Rent is related to the
auxiliary variables Basic Rent, Facilities Allowance and Mgt
Fee Allowance using the following formula:
Rent = Basic Rent + Facilities Allowance + Mgt
Fee Allowance
Mgt Fee
Excluded
Included
Excluded
Included
Excluded
Included
Excluded
Included
Excluded
Included
Excluded
Included
Figure 7: The Aggregated View Derived from the
Reservation Prices in Figure 6
Figure 6: Three Logrolling Sets and the Associated
Reservation Prices of Landlord
Rl:
Period
1 yr
1 yr
2 yrs
2 yrs
1 yr
1 yr
2 yrs
2 yrs
1 yr
1 yr
2 yrs
2 yrs
months. Should the lease start within three months; the
tenant would like to buy the facilities himself. A
management fee allowance of 1K can be added to the
Basic Rent if management fee is included. Like the case
of the landlord, the tenant needs to specify a set of
relations mapping his auxiliary variables to the template
variables.
Rent = Basic Rent + Facilities Allowance + Mgt
Fee Allowance
In general, each party may define its own set of
relations although Rl and Rt are the same in our lease
example. An aggregated view as shown in Figure 9 can
be computed based on the reservation prices given by
the tenant in Figure 8.
Rt:
1. Identify the template variables that do not have a
finite set of discrete values, such as the Rent in the
lease example.
2. Rename these template variables in aggregated
views based on the role of each negotiation party.
For instance, the variable Rent in the aggregated
view of the landlord and that of tenant is renamed to
Rentlandlord and Renttenant, respectively.
3. Obtain an offer matching table by performing a
relational join of the renamed aggregated views, and
inserting additional columns showing the differences
between the corresponding pairs of reservation
prices specified for the renamed variables.
Start Date
Facilities
Period
Mgt Fee
Rentlandlord
Renttenant
Rentdiff=(Rentlandlord
1M
Excluded
1 yr
Excluded
19K
N/A
1M
Excluded
1 yr
Included
20K
N/A
N/A
1M
Included
1 yr
Excluded
N/A
20K
N/A
1M
Included
1 yr
Included
N/A
21K
N/A
1M
Included
2 yrs
Excluded
21K
19K
2K
1M
Included
2 yrs
Included
22K
20K
2K
2M
Excluded
1 yr
Excluded
20K
18K
2K
2M
Excluded
1 yr
Included
21K
19K
2K
2M
Excluded
2 yrs
Excluded
N/A
17K
N/A
2M
Excluded
2 yrs
Included
N/A
18K
N/A
2M
Included
1 yr
Excluded
N/A
20K
N/A
2M
Included
1 yr
Included
N/A
21K
N/A
2M
Included
2 yrs
Excluded
24K
19K
5K
2M
Included
2 yrs
Included
25K
20K
5K
3M
Excluded
1 yr
Excluded
21K
18K
3K
3M
Excluded
1 yr
Included
22K
19K
3K
3M
Excluded
2 yrs
Excluded
N/A
17K
N/A
3M
Excluded
2 yrs
Included
N/A
18K
N/A
3M
Included
2 yrs
Excluded
23K
N/A
N/A
3M
Included
2 yrs
Included
24K
N/A
N/A
- Renttenant)
4.4
Offer Matching
Since the aggregated view of the landlord’s offer
contains only reservation prices of the template
variables, these prices may not be directly mapped to the
logrolling view of the tenant, which consists of the
tenant’s own auxiliary variables We explain the
construction of an offer matching table and its
utilization for tradeoff evaluation based on the tenant’s
logrolling view as shown in Figure 9.
Start Date
1M
1M
1M
1M
2M
2M
2M
2M
2M
2M
2M
2M
3M
3M
3M
3M
Facilities
Included
Included
Included
Included
Excluded
Excluded
Excluded
Excluded
Included
Included
Included
Included
Excluded
Excluded
Excluded
Excluded
Period
1 yr
1 yr
2 yrs
2 yrs
1 yr
1 yr
2 yrs
2 yrs
1 yr
1 yr
2 yrs
2 yrs
1 yr
1 yr
2 yrs
2 yrs
Mgt Fee
Excluded
Included
Excluded
Included
Excluded
Included
Excluded
Included
Excluded
Included
Excluded
Included
Excluded
Included
Excluded
Included
Rent
20K
21K
19K
20K
18K
19K
17K
18K
20K
21K
19K
20K
18K
19K
17K
18K
Figure 9: The Aggregated View Derived from the
Reservation Prices in Figure 8
The offer-matching table is an internal data structure
of the e-Negotiation system. It is transparent to the
negotiation parties, who concern mainly with the
evaluation and specification of reservation prices at their
own logrolling views. At the tenant side, an offermatching table can be constructed based on the offers by
the landlord and the tenant. The offer-matching table is
derived in the following steps.
Figure 10: An Offer Matching Table
Figure 10 gives an offer-matching table computed
from the aggregated view of the landlord and that of the
tenant. An additional column Rentdiff is inserted in order
to show the differences between Rentlandlord and Renttenant.
Matching offers are found if Rentdiff contains zero or
negative values. If so, the e-Negotiation process has
attained a reachable state; otherwise a difference table is
computed. Note that the offer-matching table may
consist of not applicable (N/A) entries, as some
combinations are not available in either the landlord’s
offer or the tenant’s offer. For example, the first two
entries in Renttenant do not have applicable values because
the tenant requires the inclusion of facilities if the lease
is to be started within a month.
Proceedings of the 36th Hawaii International Conference on System Sciences (HICSS’03)
0-7695-1874-5/03 $17.00 © 2002 IEEE
N/A
4.5
Tradeoff Evaluation Based on
Logrolling Sets
Period
1 yr
2 yrs
Basic Rent
18K
17K
Min Difference
2K
2K
Figure 11: A Logrolling Set for Tradeoff Evaluation
Period
1 yr
Basic Rent
19K
Min Difference
1K
Figure 12: Revision of an Entry
The tenant now has three options: (a) continue to
visit another alternative in the same view for tradeoff
evaluation, (b) visit another logrolling set for tradeoff
evaluation, or (c) discontinue and send the revised
counter offer to the landlord. Options (a) and (c) are
similar to the steps discussed above. Let us consider
option (b) in more details. Suppose the subsequently
picks another logrolling set, such as the [Start Date,
Facilities, Facilities Allowance] for tradeoff evaluation after
revising the entry in Figure 13. The figure gives the
existing reservation prices held by the tenant under the
logrolling set [Start Date, Facilities, Facilities Allowance].
The field Min Difference indicates the minimum amount
to revise in order to meet the landlord’s asking prices.
Here, Min Difference is computed under the assumption
that Period = 1 year and Basic Rent = 19K. The last two
Start Date
1M
2M
2M
3M
1M
3M
Facilities Allowance
2K
2K
0
0
N/A
N/A
Min Difference
N/A
N/A
1K
2K
N/A
N/A
Figure 13: Further Tradeoff Evaluation based on
Period = 1 year and Basic Rent = 19K
A Tradeoff Evaluation Scenario (I)
[Period = 1yr, Basic Rent =18K] (cf. Figure 14)
[Start Date = 1M, Facilities = Included, Facilities Allowance = 2K] (cf. Figure 15)
[Mgt Fee = Excluded, Mgt Fee Allowance = 0K]
[Mgt Fee = Included, Mgt Fee Allowance = 1K]
[Start Date = 2M, Facilities = Included, Facilities Allowance = 2K]
...
[Start Date = 2M, Facilities = Excluded, Facilities Allowance = 0K]
...
...
[Period = 2yr, Basic Rent =17K]
...
Figure 14: A Tradeoff Evaluation Scenario taken by
the Tenant
Figure 14 depicts a possible scenario taken by the
tenant for tradeoff evaluation. In this scenario, he
considers the entry [Period = 1 yr, Basic Rent = 18K] in
the view [Period, Basic Rent], then the entry [Start Date =
1M, Facilities = Included, Facilities Allowance = 2K] in the
view [StartDate, Facilities, Facilities Allowance], then the
entry [Mgt Fee = Excluded, Mgt Fee Allowance = 0K] in
the view [Mgt Fee, Mgt Fee Allowance], and so on. Figure
15 depicts another possible tradeoff evaluation scenario
for the tenant, except that it is not as systematic as the
previous one.
Suppose after the tradeoff evaluation exercise, the
tenant makes two changes in the [Period, Basic Rent]
view: (a) revising the Basic Rent from 18K to 19K for a
lease period of 1 year and (b) revising the Basic Rent
from 17K to 18K for a lease period of two years. Note
that, the tenant may still choose not to examine the two
new combinations available in the landlord’s offer as
indicated in the last two entries in Figure 13. The
revised reservation prices in terms of logrolling sets are
shown in Figure 16.
Proceedings of the 36th Hawaii International Conference on System Sciences (HICSS’03)
0-7695-1874-5/03 $17.00 © 2002 IEEE
Facilities
Included
Included
Excluded
Excluded
Excluded
Included
Time
A party starts tradeoff evaluation when the
underlying eNSS determines no acceptable offers. A
party selects a logrolling set that he would like to first
evaluate the tradeoffs. For instance, the tenant first
selects the view [Period, Basic Rent] due to the strong
preference of a shorter lease. In our proposed
methodology, a party may start tradeoff evaluation at
any logrolling set, which may even vary across different
rounds of negotiation. Once a view is selected, the party
is presented with the minimum difference between his
reservation price and that of his counter-party. For
instance, Figure 11 shows the additional column, Min
Difference, computed by the underlying negotiation
support system when the tenant has selected the view
[Period, Basic Rent]. The minimum difference is
calculated from the minimum Rentdiff amongst the
records from the offer matching table, where Period = 1
or Period = 2.
If the party decides not to make any revision in that
logrolling set, it may select either another logrolling set
for tradeoff evaluation or an entry to be revised in the
logrolling set. Suppose the tenant now revises his
reservation price to, say, 19K. The underlying system
updates the corresponding entries in the offer matching
table and a new minimum difference as shown in Figure
12.
entries represent the combinations that are found from
the counter party’s offer (i.e., landlord) but have not yet
been considered by the tenant.
4.6
A Tradeoff Evaluation Scenario (II)
[Period = 1yr, Basic Rent =18K] (cf. Figure 14)
Time
[Start Date = 1M, Facilities = Included, Facilities Allowance = 2K] (cf. Figure 15)
[Mgt Fee = Excluded, Mgt Fee Allowance = 0K]
[Mgt Fee = Included, Mgt Fee Allowance = 1K]
[Start Date = 2M, Facilities = Included, Facilities Allowance = 2K]
Offer Acceptance
Similarly, the landlord performs tradeoff evaluations
based on the tenant’s submitted aggregated view (Figure
17) and makes any necessary revisions to reservation
prices. Suppose the landlord is willing to lower the
monthly Basic Rent by 1K. Figure 18 gives the
landlord’s revised offer.
[Period = 2yr, Basic Rent =17K]
Start Date
1M
2M
3M
...
...
[Start Date = 2M, Facilities = Excluded, Facilities Allowance = 0K]
...
...
Facilities
Excluded
Included
Figure 15: Another Tradeoff Evaluation Scenario
taken by the Tenant
Period
1yr
2 yrs
Start Date
1M
2M
2M
3M
Rent
19K
18K
Facilities
Included
Included
Excluded
Excluded
Mgt Fee
Excluded
Included
Mgt Fee
Excluded
Included
Facilities Allowance
2K
2K
0
0
Mgt Fee Allowance
0
1K
The revised logrolling sets are then aggregated with a
lump sum value in Rent and submitted to the landlord as
a counter offer. The aggregated view is shown in Figure
17.
Facilities
Included
Included
Included
Included
Excluded
Excluded
Excluded
Excluded
Included
Included
Included
Included
Excluded
Excluded
Excluded
Excluded
Period
1 yr
1 yr
2 yrs
2 yrs
1 yr
1 yr
2 yrs
2 yrs
1 yr
1 yr
2 yrs
2 yrs
1 yr
1 yr
2 yrs
2 yrs
Mgt Fee
Excluded
Included
Excluded
Included
Excluded
Included
Excluded
Included
Excluded
Included
Excluded
Included
Excluded
Included
Excluded
Included
Rent
21K
22K
20K
21K
19K
20K
18K
19K
21K
22K
20K
21K
19K
20K
18K
19K
Figure 17: The Aggregated View of Tenant’s
Counter Offer
Mgt Fee Allowance
0
1K
As a result, there are four sets of accepted values,
which are summarized in Figure 19. The e-Negotiation
process has attained a reachable state, leaving both
parties mutually settled on one set of accepted values.
On the other hand, negotiation failure occurs if all
parties refuse to revise their reservation values and there
is no solution across these reservation values.
Start Date
1M
1M
2M
2M
Facilities
Included
Included
Excluded
Excluded
Period
2 yrs
2 yrs
1 yr
1 yr
Mgt Fee
Excluded
Included
Excluded
Included
Rent
20K
21K
19K
20K
Figure 19: Candidate Offers Acceptable to Landlord
and Tenant
Figure 20 presents the resultant e-Contract based on
the first accepted set of reservation prices.
Start Date = 1M
Period
…This lease will start on June 20, 2002 and last for
two year(s). The property will be rented with
facilities. Tenant is required to pay landlord a
Facilities
monthly rent of $20,000. The rent excludes the
management fee. …
Rent
Mgt Fee
Figure 20: An Example Lease Contract with
Accepted Values Underlined
5
Related Work
Modeling of e-contracts can be dated back to the
Contract Net Protocol (Smith 1980). However, they
only concentrated on low-level transaction aspects.
Proceedings of the 36th Hawaii International Conference on System Sciences (HICSS’03)
0-7695-1874-5/03 $17.00 © 2002 IEEE
Facilities Allowance
0
2K
Figure 18: Revised Offer of the Landlord
Figure 16: Revised Reservation Prices Given by the
Tenant
Start Date
1M
1M
1M
1M
2M
2M
2M
2M
2M
2M
2M
2M
3M
3M
3M
3M
Period
1 yr
2 yrs
Rent
18K
19K
20K
Gisler et al. (2000) presents a framework for legal econtracts, but not a mechanism for modeling econtracts. Grosof (1999) introduces a declarative
approach to business rules in e-commerce contracts by
combining Courteous Logic Program and XML.
Marjanovic and Milosevic (2001) model a contract with
demotic logic, based on obligation, permission and
prohibition. Recently, Karlaplem and et al. (2001) have
proposed a meta-model of e-contracts with entityrelationship diagrams for generating workflows to
support e-contract enactment. But they have not
considered the notion of logrolling views or
commitments in e-contracts. On the other hand, we have
done some work on e-Contract enactment (Chiu et al.
2002) and e-contract enforcement (Chiu et al. 2003). We
have (Cheung et al. 2002) also made a study on the
details of variable relations within the context of
contract templates and proposed a mechanism for
negotiation plan generation.
In addition, Griffel et al. (1997) presents an
application of contract negotiation by mobile agents. A
contract is represented as an object that can be accessed
by the negotiation partners. Each negotiation partner has
the opportunity to change or insert clauses. They only
provide a conceptual view of their approach and do not
describe any formal logical model to support their
approach. Further, Tan and Thoen (2002) proposed a
conceptual view to represent the contents of business
contracts with Formal Language for Business
Communication (FLBC) for contract negotiation based
on the event semantics. However, they do not provide
any conceptual and logical model to model the
negotiation of contracts. Yu and Mylopoulus (1996)
consider dependencies of business goals but not down to
practical details of variable dependencies of contract
negotiation. In summary, none of these researches
considers the logrolling process.
In political sciences, logrolling means the trading of
votes by legislators (Wilson 1969). For example, a
legislator offers to another to trade his vote on a certain
bill in return for the other's votes on a second and a third
bill. After it, logrolling (Froman and Cohen 1970) is
defined as one-actor trades a concession on one or more
issues for reciprocal concessions on other issues with
his bargaining opponent. In general, negotiators have
cognitive biases (Foroughi and Jelassi 1990), i.e.,
negotiators tend to consider issues one at one time that
is also called compromise (Froman and Cohen 1970),
because it is cognitively difficult for negotiators to
integrate multiple issues into a single package. Beside
the area of political sciences, not much research work in
logrolling has been done in other research areas such as
Operation Research (OR) and Information Systems (IS).
This paper is applying this concept into an e-commerce
application.
6
In this paper, motivated by a lease contract template
example, we have proposed a pragmatic approach to eNegotiation based on e-Contract templates. Tradeoff
evaluation in e-Negotiation often requires logrolling,
which refers to the exchange of loss in some issues for
gain in others. The set of issues involved in a logrolling
process is referred to as a logrolling set. Problems arise
when parties have different combinations of logrolling
sets. Alternative offers or counter offers that refer to one
combination can be difficult to interpret under a
different combination. Furthermore, we have formulated
a meta-model of unmatched logrolling views, examined
the involved e-Negotiation process, and proposed a
mechanism to facilitate the process.
Furthermore, facilitated by E-ADOME WFMS
(Chiu et al. 1999, 2001, 2002), a prototype eNegotiation system can be built rapidly, supporting the
cross-organizational process via contemporary Internet
technologies (Cheung et al. 2003). With such a solid and
convenient platform, we are prototyping an eNegotiation system based on the techniques developed
in this paper. In particular, we are using the lease
contract template as our primary test application. There
are two major reasons that traditional Operations
Research approaches such as utility functions and
normative models cannot be applied into this research
problem. First of all, utility functions are very sensitive
to negotiators. Therefore, it is extremely difficult and
almost impractical in reality to obtain the utility
functions from negotiators during negotiation (Soo
2000). Second, normative models can provide
compelling analyses of rational behavior in competitive
situations. However, normative models (Thompson
1998) do not adequately describe the behavior of most
negotiators in the negotiation activity. It is because
negotiators do not always behave according to
principles of normative models. Therefore, this paper
introduces a feasible approach to build such an eContract negotiation system.
This work can be expanded in several directions. We
are currently investigating one-to-many (more than two
parties at one time) negotiation of contract. We are also
interested in the role of negotiation mediators
(especially real-estate brokers and agents) and
applications in customer relationship management (Chiu
et al. 2003), and how our current solution can help them,
because this is a large business in Hong Kong. For
applications, we are also interested in sales contracts of
real estates, supply chain and investment. In addition,
we are investigating the ranking of different types of
issues and criteria (Brams and Kilgour 2001) for
logrolling issues.
Proceedings of the 36th Hawaii International Conference on System Sciences (HICSS’03)
0-7695-1874-5/03 $17.00 © 2002 IEEE
Conclusion and Future Work
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
Bertino, E., E. Ferrari and V. Atluri (1999). “The
Specification and Enforcement of Authorization Constraints
in Workflow Management Systems,” ACM Transactions on
Information and System Security, vol. 2, no. 1, pp. 209-240.
Brams, S.J. and D.M. Kilgour (2001). “Fallback
Bargaining,” Group Decision and Negotiation, Kluwer
Academic Publishers, vol. 10, pp. 287-316.
Cheung, S.C., P.C.K. Hung and D.K.W. Chiu (2002). “A
Meta-model for e-Contract Template Variable Dependencies
Facilitating e-Negotiation,” Proceedings of the 21st
International Conference on Conceptual Modeling (ER
2002), Tampere, Finland, October 2002.
Cheung, S.C., D.K.W. Chiu and S. Till (2003). “A DataDriven Methodology to Extending Workflows to E-services
over the Internet,” Proceedings of the 36th Hawaii
International Conference on System Sciences (HICSS-36).
Chiu, D.K.W., W.C.W. Chan, G.K.W. Lam, S.C. Cheung
and F.T. Luk (2003). “An Event Driven Approach to
Customer Relationship Management in e-Brokerage
Industry,” Proceedings of the 36th Hawaii International
Conference on System Sciences (HICSS-36)
Chiu, D.K.W., S.C. Cheung and S. Till (2003). “A Three
Layer Architecture for E-Contract Enforcement in an EService Environment,” Proceedings of the 36th Hawaii
International Conference on System Sciences (HICSS-36)
Chiu, D.K.W., K. Karlapalem, Q. Li and E. Kafeza (2002).
“Workflow View Based E-Contracts in a CrossOrganizational E-service Environment,” Distributed and
Parallel Database, vol 12, no. 2-3, pp. 193-216.
Chiu, D.K.W., Q. Li and K. Karlapalem (1999). “A Meta
Modeling Approach for Workflow Management System
Supporting Exception Handling,” Information Systems,
Pergamon Press, Elservier Science, vol. 24, no. 2, pp. 159184.
Chiu, D.K.W., Q. Li and K. Karlapalem (2001). “Web
Interface-Driven Cooperative Exception Handling in
ADOME Workflow Management System,” Information
Systems, Pergamon Press, Elservier Science.
Erikson, H.E. and M. Penker (2000). “Business Modeling
with UML: Business Patterns at Work,” John Wiley, New
York.
Foroughi, A. and M.T. Jelassi (1990). “NSS Solutions to
Major Negotiation Stumbling Blocks,” Proceedings of the
23th Hawaii International Conference on System Sciences,
vol. 4, pp. 2 -11.
Froman, L.A., Jr. and M.D. Cohen (1970). “Compromise and
logroll: comparing the efficiency of two bargaining
processes,” Behavioral Science, vol. 15, pp. 180-183.
Gisler, M., K. Stanoevska-Slabeva and M. Greunz (2000).
“Legal Aspects of Electronic Contracts,” In CAiSE*00
Workshop of Infrastructures for Dynamic Business-toBusiness Service Outsourcing (IDSO'00), Stockholm.
Griffel, F., M.T. Tu, M. Munke, M. Merz, W. Lamersdorf
and M.M. da Silva (1997). “Electronic contract negotiation
as an application niche for mobile agents,” Proceedings of
First International Enterprise Distributed Object Computing
Workshop, pp. 354 -365.
Grosof, B.N. (1999). “A declarative approach to business
rules in Contracts: Courteous Logic Programs in XML,”
Proceedings of the 1st ACM Conference on Electronic
Commerce (EC99), Denver, Colorado, USA, Nov. 3-5.
Herring, C. and Z. Milosevic (2001). “Implementing B2B
contracts using BizTalk,” Proceedings of the 34th Hawaii
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
Proceedings of the 36th Hawaii International Conference on System Sciences (HICSS’03)
0-7695-1874-5/03 $17.00 © 2002 IEEE
International Conference on System Sciences, pp. 40784087.
Hung, P.C.K. and J.Y. Mao (2002). “Modeling e-Negotiation
Activities with Petri Nets,” Proceedings of the 35th Hawaii
International Conference on System Sciences (CD-ROM).
Hwang, C.L. and K. Yoon (1981). “Multiple Attribute
Decision Making: Methods and Applications,” SpringerVerlag Berlin Heidelberg, New York.
Karlaplem, K., A.R. Dani and P.R. Krishna (2001). “A
Frame Work for Modeling Electronic Contracts,”
International Conference on Conceptual Modeling (ER2001).
Lewicki, R.J. and J.A. Litterer (1985). “Negotiation,” Irwin.
Marjanovic, O. and Z. Milosevic (2001). “Towards formal
modeling of e-Contracts,” Proceedings of 5th IEEE
International Enterprise Distributed Object Computing
Conference, pp. 59-68.
Meister, D.B. (1993). “A Prescriptive Analysis Negotiator
Support System,” Ph.D. Thesis, Department of Management
Sciences, University of Waterloo.
Nabil, R.A. and Y. Yesha (1996). “Electronic Commerce:
Current Research Issues and Applications,” Springer-Verlag
Berlin Heidelberg.
OMG: Object Management Group (2001). “Foreword UML
specification 1.4,” September.
Robinson, W.N. (1997). “Electronic brokering for assisted
contracting of software applets,” Proceedings of the 13th
Hawaii International Conference on System Sciences, vol. 4,
pp. 449-458.
Smith, R. G. (1980). “The contract net protocol: High Level
Communication and Control in a Distributed Problem
Solver,” IEEE Transactions on Computers 29(12), December
1980, pp. 1104-1113.
Soo, V.W. (2000). “Agent negotiation in trusted third party
mediated uncertain games,” Proceedings of Fourth
International Conference on Autonomous Agents, pp. 265266.
Stigum, B.P. and F. Wenstop (1983).“Foundations of Utility
and Risk Theory with Applications,” D. Reidel Publishing
Company.
Tajima, M. and N.M. Fraser (1998). “Modelling Multi-Issue
Negotiation,” Trends in Multi-criteria Decision Making,
Lecture Notes in Economics and Mathematical Systems, pp.
465-474.
Tan, Y.H. and W. Thoen (2002). “Using Event Semantics for
Modeling Contracts,” Proceedings of the 35th Hawaii
International Conference on System Sciences (CD-ROM).
Thompson, L. (1998). “The Mind and Heart of the
Negotiator,” Prentice-Hall Inc.
Wilson, R. (1969). “An Axiomatic Model of Logrolling,”
The American Economic Review, vol. 59, no. 3, pp. 331341.
Yu, E. and J. Mylopoulus (1996). “Using Goals, Rules, and
Methods To Support Reasoning In Business Process
Reengineering,” International Journal of Intelligent Systems
in Accounting, Finance and Management, special issue on
Artificial Intelligence in Business Process Reengineering,
John Wiley & Sons, vol. 5, no. 1, pp. 1-13.