PERSONAL SHOPPING SUPPORT

FROM DIGITAL PRODUCT MEMORIES

Alexander Kröner, Patrick Gebhard, Boris Brandherm

German Research Center for Artificial Intelligence, Saarbrücken, Germany

Benjamin Weyl, Jörg Preißinger

BMW Forschung und Technik GmbH, Munich, Germany

Carsten Magerkurth, Selcuk Anilmis

SAP Research, CEC St. Gallen, St. Gallen, Switzerland

Keywords: Smart environments, Input and interaction technologies, User studies.

Abstract: Auto-ID as well as traditional identification technologies such as barcodes allow for linking physical

products with digital data. Thus products become “smart items”, which may contribute to the consumer's

retail experience in future retail environments. In this article, we discuss how digital assistants can utilize

so-called digital product memories for personalized support during tasks typically for the interaction

between consumer and product. A demonstration system allowed participants of an IT fair to explore

various approaches to personalized support on the basis of this technology in a storyline spanning several

spaces, some of them public, some private. Feedback gathered from 132 visitors indicates that this kind of

support is in general perceived well; however, it also emphasizes the diversity of people's interest in

interaction metaphors and means of privacy protection.

1 INTRODUCTION

The domain of retail is on the verge of a new era as

pervasive computing technologies become mature

and as ubiquitous as large scale mobile internet

services and consumer phones equipped with

barcode reading and Near Field Communication

(NFC) capabilities. Within well-defined smart

spaces it becomes feasible to access and utilize real-

world information from all kinds of different sources

for the potential benefit of various stakeholders such

as consumers, retailers, or manufacturers. By means

of a link to digital data, physical artifacts – including

products – may become “smart items” integrated

into this novel information structure. In this context,

Digital Product Memories (DPM) relate to the

notion of products that “keep a diary” of relevant

events gathered throughout their lifecycle in order to

provide valuable services: A product might warn

about critical incidents such as increased

temperature or pressure during transportation, or

provide information regarding its carbon footprint

that is calculated from its actual logistics and

production emissions (Kröner, 2010a).

Installed and initialized during production, this

functionality is of special interest for manufacturers

and logistics experts to capture and communicate

artifact-level information along the supply chain

(Stephan, 2010). Beyond supply chain management,

this approach may provide retailers with an

information source, which they can exploit to satisfy

the consumers’ growing interest in tracking the way

a product at hand went from production to

consumption (Horovitz, 2009).

Assuming that retailers might receive products

already equipped with such technology, we focus in

this article on the application of the DPM for

consumer support in a retail scenario. Particular

questions in such a setting address the customers’

preferences concerning the interaction with services

built on top of the DPM (and in consequence, which

data and which technical realization would be

Kröner A., Gebhard P., Brandherm B., Weyl B., Preißinger J., Magerkurth C. and Anilmis S. (2011).

PERSONAL SHOPPING SUPPORT FROM DIGITAL PRODUCT MEMORIES .

In Proceedings of the 1st International Conference on Pervasive and Embedded Computing and Communication Systems, pages 64-73

DOI: 10.5220/0003368200640073

 SciTePress

required for installing a DPM), and their interest in

using a DPM after the point of sale.

In order to research these questions and to reflect

the DPM’s unique feature of artifact-based, cross-

domain information transport, a system is required

which comprises not only different kinds of support,

but also spans different domains. Therefore, we

realized a demonstration system showcasing a

scenario comprising all steps of a typical shopping

trip. After a personal, interactive “shopping trip”

with that system, visitors of an IT fair were

interviewed concerning their opinion.

The remaining article continues with a brief

review of related work, followed by a summary of

the system’s technical setup. Afterwards, the

system’s various services and their interaction with

the user along a coherent scenario are discussed.

Then, feedback from visitors of an IT fair is

summarized. The article concludes with a wrap up of

the results and an outlook on future work.

2 RELATED WORK

The idea of utilizing the combination of physical

artifacts and linked digital data for consumer support

is subject of considerable research activities. Most

notably, Goggles (Google, 2010) relies on visual

features of a product at hand to retrieve linked data.

However, this approach is not appropriate to

distinguish different instances of a product – a

feature essential for the DPM. Another characteristic

aspect of the presented work is user support across

services and contexts. Similarly, the MyGrocer

system (Roussos, 2002) provides information about

the product at hand in a scenario that combines retail

and stock-keeping in a smart home, although no

real-world context except for querying product IDs

is utilized. In this context, mobile shopping list

applications have been commonplace for some time

(e.g., RTM, 2011) and also mobile product

recommender systems are emerging in greater

numbers (e.g., von Reischach et al. 2010). The DPM

may be exploited to extend such services; its

uniqueness of our approach lies in the integration of

usage patterns of products previously bought.

Regarding interaction, Janzen and Maass discuss

how an interactive natural language communication

between users and products can be exploited for a

redesign of communication at the point of sale

(Janzen, 2008). Here, our vision is that the product

artifact becomes an integral part of a continuous

dialog between digital assistants and consumer that

may involve very different interaction types.

Beyond, “smart products” with embedded input

and output capabilities, actuators, sensors, and

product-specific data may proactively assist their

owners in performing their tasks (Miche, 2009) – an

extension of our assumption that digital extensions

to products may be beneficial for the consumer even

after the point of sale. However, while customers are

interested in such after sales services, they might

nevertheless prefer to destroy RFID chips at store

exits. This suggests to enhance trust through security

and privacy visibility, and to keep such privacy

enhancing technologies simple (Spiekermann,

2008). Here, feedback obtained in our demonstration

showed a positive impact of a security mechanism

with a high reputation.

3 TECHNICAL SETUP

In a previous demonstration (Kröner, 2010a), the

main goal was to illustrate how to build a DPM. The

demonstration described in this article focused on

the interaction with the DPM; particular goals

included:

• Communicating a continuous shopping

experience across public and private spaces

involving various services realized on top of

the DPM

• Illustrating the different roles physical items

may take in the interaction between service

and consumer

• Obtaining feedback concerning different

approaches to DPM-based personalization,

and to derive hints concerning the

deployment of such technology

The concept of the DPM is based on the idea that

some “smart label” attached to the product allows

for identifying a product on artifact level as well as

for continuously collecting data concerning this

artifact. This requires a data link between a product

and some digital storage. Depending on the

expectations to the DPMs behavior, this link can be

established via various technologies.

In the context of this demonstration, it was

sufficient to identify products and to get access to

linked data. In order to achieve these features across

the various stages of the demonstration – and to

enable a continuous shopping experience –, we

decided to employ barcodes and different RFID

technologies. In general, their appliance was guided

by well-known observations (Ngai, 2008). Products

in this demonstration are equipped with all of these

labels.

PERSONAL SHOPPING SUPPORT FROM DIGITAL PRODUCT MEMORIES

Figure 1: The user interacts with DPM-equipped products in a shopping scenario comprising

shopping-related actions at home (1), in the store (2-6), and in a car (7).

• Shopping Cart and Shelf. For the envisioned

kind of support it is important to know where

products are stored (e.g., to derive

manipulations such as place, take, move).

The use of many low-powered HF antennas

enables a fine-grained detection of passive

HF tags attached to the product.

• Mobile Phone. In this application it is crucial

to include a maximum number of end users,

so that we rely on barcodes and their

identification with standard consumer

phones.

• Checkout. Products have to be detected fast

and reliable. We use UHF (860 – 960 MHz)

on gate readers and related tags that are used

in logistic domains.

• User identification. In order to identify users

via a “Personal Token”, we rely on NFC

(13.56 MHz).

• Car. Since cars are made mostly out of metal,

we rely on active High Frequency (HF, 13.56

MHz) RFID tags and antennas.

A DPM may comprise data stored “on-product” (on

the smart label) and “off-product” (in external data

storages). The technical feasibility of the former

approach was shown in previous work (Stephan,

2010); however, for the purpose of this

demonstration, we considered this aspect less

relevant. Thus, all products were linked via a unique

identifier to an Object Memory Server (OMS). It

provides an open infrastructure for storing and

requesting data concerning some artifact (Schneider,

2007) – and thus implements the actual DPM.

In each stage of the demonstrator, the OMS is

involved to read, change, and update product

information (see Figure 2). Information concerning

each product was represented using a product

ontology comprising:

• General product knowledge, e.g., name,

price, and producer

• Product features, e.g., nutrition facts, or

technical properties

• Classification , e.g., ecological, traditional

• Product diary with events from production,

logistics, and retail

In order to illustrate how the DPM may support the

interaction with products from different categories,

we deploy in this setup different types of products,

including frozen or cooled products (e.g. pizza, ice,

and milk), typical everyday food products (e.g.

PECCS 2011 - International Conference on Pervasive and Embedded Computing and Communication Systems

Usage

Patterns

Product

Ontology

2/3

Creating a Shopping List

Exploring Products

Mobile Product

Information

Buying Products

Riding Home

Object

Memory

Server

SemProM Environment

User

Shopping

List

User Profile

Smart

Phone

Car

Key

Personal

Devices

Personal

Data

At Home At the Store

Product-centered Dialog

Usage

Patterns

Product

Ontology

2/3

Creating a Shopping List

Exploring Products

Mobile Product

Information

Buying Products

Riding Home

Object

Memory

Server

SemProM Environment

User

Shopping

List

User Profile

Smart

Phone

Car

Key

Personal

Devices

Personal

Data

At Home At the Store

Product-centered Dialog

Figure 2: A simplified architecture of the demonstration system. The Object Memory Server implements the DPM.

cereals, rice, and wine), and personal products that

belong to the user (e.g. PDA, car, and car key).

4 DEMONSTRATION

The system allows users without strong technical

background to explore a future retail scenario

spanning several linked stages ranging from home

over the store to the ride back home. Here, decisions

at one stage may affect services explored at another.

All stages share the idea that the product plays some

role in the interaction between consumer and

service.

4.1 Creating a Shopping List

In the beginning, the user has to create a shopping

list. This happens “at home”: In this private space, a

large-scale touch screen allows the user to

communicate with an embodied Virtual Character

that guides the creation of a shopping list. The

character acts in a kitchen environment (see Figure

1, Stage 1). It has a large repertoire of conversational

gestures and is able to generate all needed utterances

by using a template-based character control and text

generation system together with an actual TTS

system (Nuance). The character is aware of all

products at home and their status, which can be a)

available, b) need to buy, c) soon need to buy. In

addition, the character suggests recipes, whose

ingredients can be transferred to the shopping list.

Once the user is satisfied with the list, it is

transferred to a “Personal Token” – a personal item

the user owns and trusts: a car key that is equipped

with a NFC chip (Schöllermann, 2010). Such a key

can securely hold a shopping list, a credit card, and a

personality profile. The latter consists of favorite

products, allergy information, and individual

nutrition aspects. The Personal Token is supposed to

be taken to public spaces where it can be used to

reveal – at will – a small set of personal data in order

to obtain personalized support.

4.2 Product-centered Dialog

At the entrance of the store, the user takes a

shopping cart. In our demonstrator we carefully

distinguish between a public and a semi-private

shopping space. An instrumented shelf and a

refrigerated display case represent a public space.

These are aware of the position and amount of

locally stored goods. An instrumented shopping cart

represents a semi-private space. Once the user places

the key at the cart handle, the shopping cart system

retrieves the shopping list and the personality profile

from the Personal Token to enable a guided

shopping tour (see Figure 1, Stage 2).

Cart and shelf are equipped with a display

showing (different) Virtual Characters that

communicate via natural language and natural

conversational behavior with each other and the

user. In addition, the cart display shows the current

state of the shopping list. The characters react every

time a product is taken or placed.

The cart character guides the user through the

shopping list and recommends products matching

list and user profile. It resembles a personal advisor

that checks every product that is placed in the cart

with respect to individual needs and interests. The

underlying service exploits content of the respective

DPMs to reason about conflicts with the personality

PERSONAL SHOPPING SUPPORT FROM DIGITAL PRODUCT MEMORIES

profile. On a technical level, a real-time query to the

OMS for each involved product is made (see Figure

2). Emerging conflicts are addressed via natural

language by the cart character in a low voice –

respecting the privacy. Additional information is

presented on the cart’s display. In addition, the cart

character may ask the shelf character with a loud

(public) voice for help, e.g., if there is a product

alternative.

The shelf character resembles a salesperson. It

provides assistance by giving in shelf navigation

hints for faster product localization. In addition, the

character communicates general product-related

information (e.g., price, producer…) in a natural

conversational style.

The user becomes part of the dialog between the

characters (see Figure 1, Stage 3). Knowledge

retrieved from the DPM of the involved products

helps to create the illusion that Virtual Characters

reacting intelligent to the consumer’s interaction

with the product.

4.3 Exploring Products

An information kiosk enables the customer to

interact with contents stored in the DPM. The so-

called DPM Browser uses a kiosk metaphor; if the

user places a product on an RFID antenna, then a

“product diary” is presented – a temporal sequence

of events from production over transportation up to

product display at the retailer (see Figure 1, Stage 4).

While most of these events were fixed for the

purpose of this particular demo, the user may

observe changes in temperature of products taken

from the refrigerated display case, thus experiencing

the idea of quality control via DPM.

The browser visualizes the diary as a timeline

with event-associated icons. Clicking on such an

icon reveals more detailed information of the

corresponding event, e.g., the temperature value

measured, and the time when this event was

recorded. Depending on the event’s content, it is

combined with external information sources (e.g.,

GPS data is presented in an online map).

This interaction relies on public data provided by

arbitrary entities along the supply chain. However,

we assume that there is considerable potential for

user support from products with DPMs carrying

personal data. This idea is demonstrated by a weekly

medicament blister with DPM. Such a blister is

individually produced for each patient by a fully

automatic packaging plant; the contained medicines

are arranged in solid oral forms and in the correct

dosage, pre-sorted for the seven days of the week

and for taking four times a day (see, e.g., 7x4

Pharma, 2011). Samples of such a product were

Figure 3: A screenshot of the mobile product information

service.

equipped with a DPM containing similar data as the

regular products as well as information about the

intake of the contained medicine.

However, since such a blister’s DPM may allow

for undesired conclusions concerning their owner,

access is controlled by means of a roles and rights

management concept with identification by

electronic identity card (Brandherm, 2010). In the

demonstration, such access to private data in a

public space requires the customer to prove his or

her identity via prototypes of the upcoming German

electronic identity card. Once access is granted, the

user can browse the diary of the blister and retrieve

hints containing the intake of the medicines. In

addition, a drug interaction alert service is enabled.

If additional products are placed on the browser’s

surface (e.g., food, supplement or drug previously

selected in other parts of the store), then this service

checks for interactions based on contents retrieved

from the blister’s and these products’ DPMs. If an

interaction is detected, then external information

sources are used in order to support patient and

physician with different views on an interaction.

4.4 Mobile Product Information

Access to the user profile from the shopping cart

introduced before was semi-public in nature, as the

rather large display on the cart is unsuitable for

completely private interaction.

In order to provide explicit means for accessing

and even altering user profile dimensions in a public

space using a private interaction device, we

introduce an Android based mobile phone

application called “SOPHIA” (Wisdom) that stands

for “SemProM Product Hybrid Information App”

(see Figure 1, Stage 5 and Figure 2) and effectively

links information stored on the product at hand and

PECCS 2011 - International Conference on Pervasive and Embedded Computing and Communication Systems

Figure 4: The customer’s car detects products (left-hand side) and exploits their DPMs for personalized assistance (right-

hand side, example: recommended time to drive home).

relates it to a user profile stored on the mobile

phone.

The user profile can be automatically created by

analyzing the characteristics of products of a similar

category that a user already owns. The central idea

of this mechanism addresses the fact that the DPM

technology allows for storing product-related events

that occur in the lifetime of that particular artifact.

Consequently, a product that is often used should

provide many events in its DPM and many events

should then equal a high relevance to its owner,

simply because he uses the product frequently

(which is a strong simplification, of course). In this

respect, due to the utilization of DPM technologies,

the product recommendation can be based on actual

usage histories and not merely on the purchase

history, even without a semantic analysis of the

events contained in the DPM. In the demonstrator

setup, we utilized a small number of products from

the previously mentioned categories.

Interaction with the system is straightforward.

The user accesses the DPM of a product simply by

reading its ID via the built-in barcode reader of the

phone and then accessing the relevant product

properties via the aforementioned OMS. By relating

the subjective importance of each relevant profile

dimension to the properties of the concrete product,

a “suitability index” is calculated and displayed that

is a measure for how well the product matches the

profile of the user. Due to the private nature of the

interaction device, the weighing of individual profile

dimensions can be explicitly altered by dragging

respective slider controls as shown in Figure 3

within the product category of mobile phones, for

which the price, community ratings, carbon

footprint, compatibility, and raw functionality are

the relevant dimensions shown on the screen.

By evaluating for a given product category the

DPMs of all products belonging to a user, it

becomes possible to implicitly create user profiles

based on deduction, following Cattell’s notion of

past behavior predicting future behavior (Cattell,

1969), i.e. it is plausible to assume that a new

product is more likely to appeal to the user if it

matches the properties of the user’s existing items.

From this perspective it is possible, but not

always necessary, to control and adapt explicitly

profile dimensions that have meaningful default

values based on the past interaction with DPMs.

4.5 Buying Products

At the next stage, the user has to pay – just by

moving a shopping basket with the previously

selected products a through a gate and authorizing

the payment process via the Personal Token (see

Figure 1, Stage 6). A regular screen displays all

products recognized; if the user places actively the

car key at a specific location nearby, the DPM of

each detected product is updated with the

information that it is now owned by a specific

customer (respectively its id). In addition, time and

location (store) are stored to build a seamless

product history. Furthermore, a Matrix Code is

displayed that serves (in addition to a printout) as

receipt. By taking a picture of this code, the receipt

can be transferred electronically to the mobile device

for further exploitation – e.g., for housekeeping

books to track costs.

4.6 Riding Home

New external communication interfaces, fixed and

wireless, increasingly become integral part of

automotive on-board networks. These wireless

interfaces enable new value-added info- and

entertainment services, such as internet services and

seamless mobile device integration. Besides

technologies such as GSM, Bluetooth and USB,

RFID technology enables new proactive scenarios

PERSONAL SHOPPING SUPPORT FROM DIGITAL PRODUCT MEMORIES

within the vehicle (Steffen, 2010). In combination

with the processing of personal information as well

as context information, this technology can be used

to detect and infer user intentions inside the vehicle

and thus provide means to enhance usability of

functionality and new personalized value-added

services.

During the demonstration, this idea is illustrated

by a car, which reacts on products deployed in the

passenger cell (see Figure 1, Stage 7). Products

which are equipped with (active) RFID technology

can be automatically detected via respective

receivers inside the vehicle. Thus, the vehicle can

process information from its own sensors, the

respective DPMs, and further information describing

the user’s context available via the vehicle’s

connectivity services. Eventually, a user intention

can be inferred. Services exploit this information to

assist the customer in the area of:

• Configuration and personalization of

software and services according to the

provided context

• Object-centric execution and triggering of

services inside the vehicle as well as

integration of infrastructure based services

• Seamless recommendations and assistance

with continuous evaluation of the DPM and

context information

Thus, groceries bought within the store interact with

the vehicle; it evaluates their DPMs in conjunction

with additional context information, such as outside

and inside temperature, planned route and estimated

time of arrival. Then, the car computer offers

services matching the particular kind and state of the

product at hand (see Figure 4, left-hand side).

For instance, the vehicle can advice the driver at

which time she/he needs to drive back home in order

to put the products into the fridge right in time (see

Figure 4, right-hand side). Further advice can be

given with respect to products which have not yet

been bought but are still on the shopping list: for

instance, the vehicle may suggest the next shopping

mall, where the product can be purchased, as a new

destination. Furthermore, the vehicle compares

products inside the car with the list of paid products

stored on the key in order to warn the driver if she

forgot something at the checkout point.

If the customer has bought a new mobile device,

the vehicle can detect based on the DPM, which

software is needed to install in the car in order to

pair the device accordingly. This software can be

downloaded to the on-board system once the item is

detected inside the car, so that the device may be

used right away.

In addition, a product may trigger services

supporting communication concerning the product,

e.g., a Twitter application. The Twitter application

registers to a specific tweet in the context of the

product, e.g., the mascot of the football world

championship registers to the respective tweets on

this topic. The product may also trigger the

download of an application (e.g., a travel guide) or

media content and thus adapt the car's infotainment

system to the user’s interests based on what she

takes into the vehicle.

Finally, in order to support safety and

transparency during its own maintenance, the car

exploits information from the DPM of spare parts to

inform about their compatibility and correct usage.

5 FEEDBACK

The demonstration was presented at the CeBIT 2010

technology fair. Accompanied by an expert for the

respective demo stage, a visitor could explore the

various system components. The resulting “shopping

tour” followed the fixed sequence of stages depicted

in Figure 1. Each stage provided some degree of

freedom (e.g., concerning the products in the

shopping list). Visitors who made a complete tour

spent up to 40 minutes at the exhibit.

While the setup did not aim at an evaluation of

the DPM, this event was nevertheless an interesting

opportunity to identify trends and gain hints

concerning future extensions of the system.

Therefore, visitors who explored all demo stages

were asked to answer a couple of questions from

three areas related to the interaction with the DPM:

user interface, usefulness of service, and privacy.

The questionnaire addressed:

• Demographic data and purpose of the visit

• Knowledge about RFID and similar

technologies

• Preferences regarding interaction device and

modality

• Utility of car-related services and factors

affecting a buying decision

• Conditions motivating a user to keep a

product's DPM intact after purchase

• Trust in the protection of personal data, and

rating of privacy at the different demo stages

• Effects of the application context

For most of these questions, potential answers were

arranged on a four point Likert scale. Filling out the

questionnaire took between further 10-20 minutes; a

PECCS 2011 - International Conference on Pervasive and Embedded Computing and Communication Systems

Whichfeatureswouldmotivate

youtokeepaproduct'sDPM

intactafterpurchase?

10%

20%

30%

40%

50%

answer

Strongly

disagree

Disagree Agree Strongly

agree

Qualitycontrol

Productusage

Productfeatures

Complaints

Maintenance

Users

Isprivacyprotectionacrucialfeature?

10%

15%

20%

25%

30%

35%

40%

45%

answer

Strongly

disagree

Disagree Agree Strongly

agree

Cart

MobileInfo

Checkout

Car

Blister

Users

Figure 5: Visitors’ feedback – would they keep the DPM intact (left-hand side), where is privacy protection (especially)

crucial (right-hand side)?

project member guided the visitors through the

questionnaire and provided additional explanations.

5.1 Participants

132 visitors answered the questionnaire, 71% male

and 24% female. 65% of them were up to 30 years

old, further 23% were between 31 and 50 years old.

The reason of the CeBIT 2010 visit was in most of

the cases a private one (66%). Thus, the answers

might be biased towards the perspective of male

customers. The visitors' experience with RFID was

surprisingly low (for an IT fair) - 50% of the

answers expressed little or very little experience

with this technology, opposed by 41% with (strong)

experience.

5.2 User Interface

The demonstrator intentionally mixed a wide range

of different interaction types. These involved

specific mechanisms such as a smart shelf as well as

general purpose access mechanisms such as the

user's mobile phone. 84.85% of the answerers

expressed a preference towards the latter approach.

The interaction relied on implicit mechanisms

(e.g., product detection during checkout) and on

explicit ones (e.g., exploration at the kiosk).

Employed modalities included point & click (touch,

pen), tangible interfaces, and speech (output only).

Here, the visitors could express multiple

preferences. The majority of the answers (73%)

expressed a preference on graphical user interfaces,

followed by tangible ones (16%) and speech-based

ones (11%).

Discussion. For the interface to the DPM, we

conclude that the utilization of novel user interface

types might not be crucial for the success of the

DPM. However, we should devote special attention

to access mechanisms and devices, which are

familiar to users – such as the personal mobile

phone.

5.3 Usefulness of Service

Regarding the application of the DPM in the context

of a car, 85% considered reminders concerning

missing products as a relevant or very relevant

feature; 73% answered similarly regarding

information on spare parts, and at least 68%

considered information from transported products as

useful or very useful.

Feedback regarding information affecting the

actual buying decision revealed that 74% of the

answerers would use product ratings provided by

other customers. 64% take the similarity to products

they already own into account.

Discussion. For the design of DPM-based

services, we conclude that the customer’s car might

serve as one hub for DPM-based services related to

shopping. Furthermore, people’s interest in

comparing and combining product-related data from

a DPM at hand with other information sources (e.g.,

user ratings, other DPMs) indicates a need for

standards and services, which explicitly support the

alignment of DPM data as well as the

communication about a product between customers.

Finally, this feedback indicates an interest in

product-related services across the actual shopping

process. The latter observation is of special interest

for the application of the DPM, whose open nature is

meant to support such services explicitly.

5.4 Privacy and Trust

For applications of the DPM beyond the shopping

process, e.g., transportation via the own car, it is

crucial that the DPM hardware of a product stays

PERSONAL SHOPPING SUPPORT FROM DIGITAL PRODUCT MEMORIES

intact after payment. 77% of the answers indicated a

will to keep the DPM in place for quality control,

information concerning product application,

information concerning product features, and/or

issuing complaints concerning the product (multiple

choices possible, see Figure 5, left-hand side).

However, there are considerable differences in the

feedback to the various services.

The DPM and related technologies can be

exploited to collect and transport data about a user.

While the visitors’ trust in the protection of their

data was limited (32% positive, 63% negative

answers), the means of privacy protection employed

for the medicine blister were perceived positively

(40% positive, 50% negative answers). The need for

data protection was perceived differently for the

individual elements of the scenario. It was rated as

crucial for checkout, car, and blister (see Figure 5,

right-hand side).

Interestingly, 71% of the visitors expressed that

they wouldn’t differentiate between a professional or

a private use of a DPM.

Discussion. Regarding privacy concerns and trust

in the DPM, we conclude that the deployment of the

DPM should be judged with respect to the kind of

the product. A special need for data protection exists

if a DPM is applied in a public space to retrieve or

communicate sensible personal data, in this case:

about (driving) behavior medication, and payment.

Furthermore, we speculate that there might be a

relationship between indirect user interaction with

the DPM and trust - the use-cases presented at

checkout and car both included pro-active access to

the DPM of personal items, which was not a direct

response to an explicit user action. The presented

mechanisms for privacy protection were perceived

positively; however, they might not sufficient to

resolve trust issues in general.

6 CONCLUSIONS AND

OUTLOOK

Future retail may exploit novel technologies in order

to enrich the customer’s retail experience. Such a

technology is the digital product memory (DPM) – a

record of digital data linked with a physical product.

We reported about a demonstration system, which

illustrates how this information source can be

utilized for the realization of personalized, product-

centric services, which support the customer during

shopping and beyond.

By means of a survey, we collected feedback

from users of this system (visitors of a public

presentation at an IT fair). From that feedback we

conclude that user interaction with the DPM should

integrate smoothly into technology people are

familiar with. Feedback concerning services

indicates a strong interest in applications of the

DPM beyond the point of sale. However, despite

powerful means of privacy protection, trust in this

technology nevertheless stays limited. Therefore, we

recommend limiting the storage of personal data in

the DPM to particular combinations of products and

services, such as medicine and quality control.

These conclusions have limitations, which might

require additional experiments. Thus, the technically

affine CeBIT visitors do not necessarily reflect the

preferences of arbitrary shoppers. Furthermore, in

order to facilitate the demonstration at a fair, the

presented setting did not allow the visitor to “switch

off” characteristic DPM features (e.g., logging) and

thus to compare user support with products equipped

with a DPM, any other smart item technology, or no

such technology at all.

Future work will have to address these questions

in order to enable retailers to judge more precisely

where DPM-based applications are most promising.

In addition, mechanisms for facilitating the

deployment of DPM-technology are required. Here,

we are planning to devote special attention to a

standardization of memory structure and content

(Kröner, 2010b), and to management tools which

support setup and configuration of DPM-based

services.

ACKNOWLEDGMENTS

This research was funded in part by the German

Federal Ministry of Education and Research under

grant number 01 IA 08002 (project SemProM), grant

number 01 IS 08025B (project INTAKT), and grant

number 01 IC 10S01H (Software-Cluster –

cooperative project EMERGENT). We are grateful

for the Virtual Character technology and the

extensive support provided by the Charamel GmbH.

The responsibility for this publication lies with the

authors.

REFERENCES

7x4 Pharma, 2011. 7x4 Box, http://www.7x4box.de/,

accessed on January 11, 2011.

Brandherm, B. et al., 2010. Roles and Rights Management

Concept With Identification by Electronic Identity

Card. In Proc. of PerCom-2010, IEEE Computer

Society, 768-771.

PECCS 2011 - International Conference on Pervasive and Embedded Computing and Communication Systems

Cattell, R. B., 1969. The profile similarity coefficient, rp,

in vocational guidance and diagnostic classification.

The British Journal of Educational Psychology 39,

131-142.

Google Mobile Labs, 2010. Goggles for Android.

http://www.google.com/mobile/goggles/, accessed on

September 3, 2010.

Horovitz, B., 2009. Tracking deliveries of all kinds is on

everyone’s radar. USA Today, http://www.usatoday

.com/money/media/2009-07-26-pizza-products-

tracking_N.htm, accessed on October 27, 2009.

Janzen, S., and Maass, W., 2008. CoRA – Interactive

Communication with Smart Products. In Proc. of the

2nd European Workshop on Smart Products: Building

Blocks of Ambient Intelligence (AmI-Blocks'08), 9-15.

Kröner, A. et al., 2010a. Demonstrating the Application of

Digital Product Memories in a Carbon Footprint

Scenario. In Proc. of IE 2010, IEEE Computer Society

(2010), 164-169.

Kröner, A. et al., 2010b. W3C Object Memory Modeling

Incubator Group. World Wide Web Consortium,

http://www.w3.org/2005/Incubator/omm/, accessed on

January 10, 2011.

Miche, M. et al., 2009. Core Services for Smart Products.

In Proc. of the 3rd European Workshop on Smart

Products: Building Blocks of Ambient Intelligence

(AmI-Blocks'09).

Ngai, E.W.T. et al., 2008. RFID research: An academic

literature review (1995–2005) and future research

directions. International Journal of Production

Economics 112, 2, 510-520.

von Reischach, F., Guinard, D., Michahelles, F., Fleisch.,

E., 2010. A mobile product recommendation system

interacting with tagged products. In Proceedings of the

2009 IEEE International Conference on Pervasive

Computing and Communications (PERCOM '09).

IEEE Computer Society, Washington, DC, USA, 1-6

RTM, 2011. Remember The Milk: Online to do list and

task management, http://www.rememberthemilk.com/,

accessed on January 10, 2011.

Roussos, G. et al., 2002. A case study in pervasive retail.

In Proc. of the 2nd international Workshop on Mobile

Commerce (WMC '02), ACM Press, 90-94.

Schneider, M., 2007. Towards a General Object Memory.

In UbiComp 2007 Workshop Proceedings, 307-312.

Schöllermann, T. et al., 2010. Schlüsselerlebnis: NFC im

BMW Key als Schlüssel für ein neues

Mobilitätserlebnis. Elektronik Automotive 1+2, 40-43.

Spiekermann, S., 2007. Privacy Enhancing Technologies

for RFID in Retail – An Empirical Investigation. In

Proc. of UbiComp 2007, Springer, LNCS 4717, 56-72.

Steffen, R. et al., 2010. Near Field Communication (NFC)

in an Automotive Environment – Use Cases,

Architecture and Realization. In Proc. of the 2nd

international IEEE Workshop on Near Field

Communication, Monaco, 15-20.

Stephan, P. et al., 2010. Product-Mediated

Communication through Digital Object Memories in

Heterogeneous Value Chains. In

Proc. of PerCom-

2010, IEEE Computer Society, 199-207.

PERSONAL SHOPPING SUPPORT FROM DIGITAL PRODUCT MEMORIES