Wireless Marketing of Ephemeral Personal

Goods: the Case of Auctioning Screen Estate for

Wireless Advertisements

Seng Wai Loke, Arkady Zaslavsky, Bijyendra Jain

School of Computer Science and Software Engineering

Monash University, VIC 3145, Australia

Abstract. We consider the notion of ephemeral personal goods that can be

traded wirelessly, capitalizing on the anytime anywhere capability of mobile

and wireless technologies which permits immediacy of action. More specifi-

cally, this paper considers a market-based mechanism for controlling the flow

of advertisements from businesses to the mobile devices of potential customers.

The mechanism relies on recognizing that the screen estate of a mobile device

user at appropriate contexts (e.g., time and place) is a commodity in itself, and

that businesses might pay to “purchase” or “rent” the screen estate of the

masses, and in doing so, purchase the right to send advertisements to users. We

describe an implementation using the JADE-LEAP multiagent toolkit, and dis-

cuss open issues.

1 Introduction

There is an unprecedented virtual proximity between businesses and customers, made

possible by the growth of wireless Internet technologies. Customers can be reached

anywhere, anytime, at least in principle. Such anywhere anytime capability permits

immediacy of action and so, opens doors for new forms of commerce. In this paper,

we consider the notion of ephemeral personal goods, which refers to items belonging

to an individual that can be traded for monetary gains but these items are ephemeral

in the sense that they might only have marketable value for a limited time and/or their

highest value is only at a specific period. Outside of this period, the item might not be

as valuable – the item should not be sold too early or too late. Due to the time factor,

immediacy of action in trading such goods is important. We consider an example of

an ephemeral personal good that is traded wirelessly via an auction in the area of

wireless advertising.

Indeed, mobile and wireless marketing,

including wireless advertising, will play

an important role for the growth of businesses, and so will location-based advertising

See http://www.mmaglobal.com/

Wai Loke S., Zaslavsky A. and Jain B. (2004).

Wireless Marketing of Ephemeral Personal Goods: the Case of Auctioning Screen Estate for Wireless Advertisements.

In Proceedings of the 3rd International Workshop on Wireless Information Systems, pages 127-133

DOI: 10.5220/0002667701270133

 SciTePress

[5,7]. However, an outstanding issue is the cost to customers, who might face unpre-

cendented spamming. Customers can opt-in to receive wireless advertisements of

interest, when they are ready or simply feel like it, but this is only one way customers

can control this channel. The incentives and potential benefits should continually

outweigh costs in order that customers might be encouraged to keep this channel

open.

This paper presents a market-based mechanism for controlling the flow of adver-

tisements from businesses to the mobile devices of customers. The mechanism relies

on recognizing that the screen estate of a mobile device user at appropriate contexts

(e.g., time and place) is a commodity in itself, and that businesses might pay to “pur-

chase” the screen estate of the masses. Various pay-to-surf or pay-to-view-ads com-

panies have surfaced in the past, and so this idea of purchasing, or renting for a pe-

riod of time, a user’s “desktop screen estate” (and hopefully, the user’s attention) is

perhaps not new. Here, we do not comment on the business virtues of this idea and

consider a variant of such an idea. We consider how a mobile device user could auc-

tion off its screen estate at certain times for certain periods, thus (for a small fee)

opening its doors to advertisements in a restricted way, with at least small guaranteed

incentives. By adjusting the price of his/her screen estate, the customer can regulate

the volume of advertisements he/she will receive. And the screen estate value of re-

sponsive users (users who respond often to such advertisements) might be higher than

non-responsive users. It is perhaps too early to comment on the business success or

failure of such a model, but we mainly consider the technical aspects in this paper.In

the rest of this paper, in Section 2 we detail what is it about screen estate that we have

viewed as a commodity. Section 3 describes the mechanism by which screen estate

can be auctioned off with the help of intelligent software agents. Section 4 concludes

the paper, and discusses open issues.

2 Screen Estate as a Commodity

Users of mobile devices sell screen-estate-items, which roughly quantifies the re-

sources a user is making available to view advertisements on his/her device. A

screen-estate-item is a tuple:

(<device location>,<date and time>,<duration>,<size>,<minimum price>)

where <device location> refers to the current location of the device, <date and

time> refers to the current date and time of day, <duration> refers to the duration for

which the user is making his/her screen space available to display advertisements,

<size> refers to the size of the screen estate for viewing advertisements (size can be

stated relative to the device, e.g. full screen or half-screen, or number of lines of text),

and <minimum price> refers to the minimum price the screen-estate-item will sell for.

Two other important pieces of information that accompanies a screen-estate-item are

the user’s interest profile, and device characteristics, the former is needed to help

128

vendors target their advertisements, and assess if they should join in the bidding,

and the latter, to help determine the actual characteristics (e.g. color, etc) of the adver-

tisement. These properties which define a screen-estate-item enables advertising that

is much more contextualized, and targeted, and enables the vendors to know more

about the situation of the user. For example, a screen-estate-item such as the follow-

ing tuple describes the situation of a user on Bourke Street Mall (a shopping Mall in

Melbourne) during the typical 1pm lunch hour, opting in to receive advertisements at

full-screen mode for the next 2 minutes.

(Bourke Street Mall, 4 March 2003;1 pm, 2 minutes, full screen, $0.50)

This clearly describes an opportunity for nearby (within 150m of the user’s location,

say) food vendors to target this user. Other attributes might be worth adding which

adds value to the screen-estate-item by providing more information to vendors of

what the user might spend money on, including the mood of the user (e.g., bored, sad,

nervous), the intentions of the user (e.g., hungry and looking for food), or the user’s

wish-list.

3 Selling Screen Estate via Location-Based Auctions

In our prototype implementation, the user selects a set of values to define a screen-

estate-item to be auctioned. Our prototype is implemented using a multiagent system

based on the JADE-LEAP toolkit

(http://leap.crm-paris.com/) running on a

Palm OS emulator.

3.1 JADE-LEAP

Intelligent multiagent systems refers to systems consisting of a set of interacting

agents, where each agent is a software entity (a process or a thread) that exhibits

autonomous, proactive, communicative (with other agents and/or with users using

high-level interaction protocols), and adaptive behaviour. Such agents should also be

capable of limited amounts of reasoning (e.g., to carry out a pre-programmed bidding

strategy). JADE-LEAP [3] is a toolkit for building multiagent systems where the

agents reside on mobile devices and stationary devices. Agents can run in the Java2

MicroEdition environment on a small device, and interact with agents on stationary

devices over a wireless network.

3.2 A Prototype System

The reason for using agents is to automate the process of auctioning screen-estate-

items, both for vendors, which potentially deals with hundreds or thousands of sellers

However, vendors might want to send the advertisements regardless of the profile, even if the

product might not match the user’s interest profile. This is because, the profile might not be

an exhaustive description of the user’s interest, or the user might like what is being adver-

tised.

129

at a time, and for sellers who are on-the-move and would not want this process to be a

burden to them. Moreover, we leverage on existing FIPA work on standardizing

interaction protocols and toolkits that implement such protocols, thereby shortening

implementation time. Figure 1 shows the user interface for defining a screen-estate-

item. Figure 2 sketches the architecture of the system.

Fig. 1. A series of pull-down menus to help the user define a screen-estate-item.

Fig. 2. The multiagent system for a location-based auction

130

The basic architecture should have the following agent types:

• User Agent: its runs on the mobile device. It interact with the user and server-

side agent. It accept queries from the user, sends it to Broker Agent and gives the

result back to the user. The protocol between the User Agent and the Broker

Agent is based on the FIPA Brokering Interaction Protocol Specification

(http://www.fipa.org/specs/fipa00033). Figure 1 are screen dumps of the user

agent.

• Broker Agent: the main role of the broker agent is to coordinate the auction proc-

ess, from assembling interested vendors, regulating the auction (as the auction-

eer), to finalizing the transaction if a sale takes place.

• Vendor Agents: it represent the actual retailers like Myers etc. It decides what

price should be offered in a particular round of the bidding process. Such agents

might make decisions based on pre-programmed strategies.

The Broker Agent interacts with Vendor Agents using the FIPA English Auction

Interaction Protocol Specification (http://www.fipa.org/specs/fipa00031). User agents

might interact with Broker agents over short-range wireless networks such as Blue-

tooth, Infrared, or Wireless LANs, in which case location information comes for free.

Once the user has defined a screen-estate-item for sale, he/she sends it to the Bro-

ker Agent which then initiates an auction with vendors to sell the item. Once the

screen-estate-item has been sold, the winning vendor is permitted to send advertise-

ments to the user in the way prescribed in the definition of the screen-estate-item. The

User Agent and Broker Agent can ensure that the restrictions specified in the screen-

estate-item definition are adhered to. For example, the vendor can only send adver-

tisements for a fixed period and not longer – the Broker Agent or the User Agent

might stop the flow of advertisements after the specified period.

The utility of the auction mechanism is that the user can set a price of his/her

screen-estate-items, and this price acts as a filter for the amount of advertisements

that gets sent to the device. The higher the price, the fewer the advertisements. Hence,

our approach is more general than simply having the user opt-in or not opt-in to re-

ceive advertisements. The user has the option of opting-in (by setting the price to

zero) or opting-out (by setting the price to infinity), or choosing an option in between

these two extremes, i.e. receiving some advertisements.

Multiple instances of the agent types can be used to cope with large numbers of

customers concurrently.

3.3 Broader Scope of the Work

The broader scope of this work is the integration of a number of special purpose loca-

tion-based services grouped together under the concept of a location-based e-

marketplace that is superimposed on a physical marketplace [4]. Such a location-

based e-marketplace will virtually integrate the vendors of the physical marketplace

and provide new ways for users to interact with these vendors while the user is in

sufficiently close proximity to these vendors. Figure 3 shows the opening menu of the

user agent with three options: (1) modifying the user profile, (2) entering a user query

to initiate a location-based reverse auction as described in [4], and (3) selling a

131

screen-estate-item as described in this paper. Selecting the third option leads to the

screens as shown in Figure 1. Other shopping helpers (e.g., [2,6,8]) might be inte-

grated as another option within the user agent. The enabling networking technology

for such location-based e-marketplaces might be the proliferating hot-spots [1].

Fig. 3. The user agent’s opening menu.

4 Conclusions and Open Issues

We have proposed the concept of auction-based filtering of wireless advertisements

by considering parameterized screen estate as a commodity. We see that it is possible

to define screen-estate-items and to sell them to vendors anytime anywhere as long as

there is supporting infrastructure in the environment (e.g., appropriate servers). How-

ever, users might not want to spend time configuring screen-estate-items while on-

the-move. For greater convenience, users should be able to pre-configure a variety of

screen-estate-items and sell them in just one-click. Filtering of advertisements will

also benefit from information about user interests stored in user profiles. But such

profiles should be stored locally (on the mobile device) for privacy reasons. This

implies filtering on the mobile device which might be resource intensive - we envi-

sion future mobile devices to have greater resources for such purposes.

Our current on-going implementation is only simulation work and shows that the

method proposed is feasible in principle, but there are a number of outstanding issues

to tackle for the idea to be feasible in practice:

• The system should allow the user to adjust the price even after advertise-

ments start to arrive. For example, if the user finds that too many advertise-

ments are coming in, the user can raise the price (or conversely, if too few

advertisements are coming in, the user can lower the price). Alternatively,

the user agent might do that automatically given some pre-programmed

132

rules. This feature might require continual auction processes whereby the

user’s new prices can be employed without delay.

• There is a need for a reputation mechanism for users. For example, it is al-

ways possible that a user signs up for advertisements, i.e. effectively opts-in,

and then not view any of the advertisements at all, or that when advertise-

ments arrive, they are immediately deleted. One way to overcome this prob-

lem might be to use interactive advertisements or to keep a history of the

user’s responsiveness to advertisements viewed. A user who often responds

to advertisements viewed might have a higher screen-estate value. If a user

never responds to any advertisements, he/she might lose credibility and few

vendors might bid for his/her screen estate.

• The user has little control over the advertisements sent to the user’s device.

Offensive or inappropriate material might find their way into the user’s de-

vice, unless the user agent performs some local filtering - in this regard, tex-

tual materials might be easier to analyze than multimedia.

Our approach might also be integrated with existing location-based advertising

middleware (e.g., [5]).

We think that there is a rise in personal assets that the individual can sell on-

demand and screen estate for wireless advertisements is only one possibility. There

are other digital assets such as information or services that can be wirelessly marketed

in an ad hoc manner. For example, one could sell information, recommendations,

referrals, or be a shopping advisor for certain period of time.

References

1. Ahlund, C., Zaslavsky, A., and Matskin, M. Supporting Mobile Business Applications in

Hot Spot Areas with Pervasive Infrastructure. In Proceedings of the 1

International Confer-

ence on Mobile Business, Greece, July 2002.

2. Fano, A. SHOPPER’S EYE: Using Location-Based Filtering for a Shopping Agent in the

Physical World. In Proceedings of the International Conference on Autonomous Agents, pp.

416 – 421, ACM Press, 1998.

3. Bergenti, F. and Poggi, A. LEAP: A FIPA Platform for Handheld and Mobile Devices. In

Proceedings of the ATAL Conference, Special track on agents on handheld, mobile or em-

bedded devices, Seattle, August 2001.

4. Loke, S.W. An Exploration of Agent Assistance for Physical Marketplaces: Proximity-Based

Reverse Auctions. In Proceedings of the 2003 International Conference on Intelligent

Agents, Web Technologies and Internet Commerce (IAWTIC 2003),Vienna, Austria, (ed.)

M. Mohammadian, February, 2003, 124 - 135.

5. Kolmel, B. and Alexakis, S. Location Based Advertising. In Proceedings of the 1

Interna-

tional Conference on Mobile Business, Greece, July 2002.

6. Menczer, F., Street, W.N. and Vishwakarma, N. IntelliShopper: a Proactive, Personal, Pri-

vate Shopping Assistant. In Proceedings of the 1

International Joint Conference on

Autonomous Agents and Multiagent Systems, Bologna, Italy, ACM Press, 2002.

7. Ververidis, C. and Polyzos, G.C. Mobile Marketing Using a Location Based Service. In

Proceedings of the 1

International Conference on Mobile Business, Greece, July 2002.

8. Yang, J. . MORPHEUS: A More Scalable Comparison-Shopping Agent. In Proceedings of

the 5th International Conference on Autonomous Agents, Montreal, Quebec, Canada, ACM

Press, 2001.

133