PHYSICAL DOCUMENT ADAPTATION TO USER’S CONTEXT

AND USER’S PROFILE

Hassan Naderi and Béatrice Rumpler

INSA LYON, Bâtiment Blaise Pascal, 7, Av. Jean Capelle F69621 Villeurbanne Cedex, France

Keywords: Mobile device, Document adaptation, Physical Document Adaptation (PDA), Context, User’s delivery

preferences.

Abstract: Modern technology promises mobile users Internet connectivity anytime, anywhere, and using any device.

However, given the constrained capabilities of mobile devices, the limited bandwidth of wireless networks

and the varying personal preferences, effective information access requires the development of new

computational patterns. The variety of mobile devices available today makes device-specific authoring of

web content an expensive approach. The problem is further compounded by the heterogeneous nature of the

supporting devices and the users' behaviour. This research investigates the challenges posed by these

problems, and proposes a context-aware adaptation framework to bridge the gap between the existing

Internet content and today's heterogeneous computing environments.

1 INTRODUCTION

The rapid advances in wireless communication

technologies with their integration into mobile

devices such as mobile phones, PDA, and portable

computers provide a technical infrastructure that

enables a broad range of mobile services and

applications. These devices are designed to help

people access and to manage information. While

they provide greater access to online information,

they also suffer from significant limitations such as

small screen-sizes, slow connection, and limited

input capabilities. A typical cell phone screen is 200

times smaller than a standard PC monitor. This

imposes limits on one's ability to locate and display

the right information quickly at the right time, and

therefore the information overload problem itself

becomes more serious. Because these access devices

are becoming more and more popular, there is a

need to make the next generation of information

retrieval systems capable of automatically adapt the

found document to the user's situation (cf. § 4,

Figure. 1).

In section 2, we study the actual approaches to

adapt a document to a given situation. In section 3

we present our proposition to adapt a given

document to the user's situation. In section 4 we

define the concept of user's situation. In section 5 we

show how the user's situation influences the process

of Physical Document Adaptation (PDA). In order to

simplify and clarify the process of PDA we have

categorized the documents' components in two main

classes: structural components and logical

components (section 5). In section 6 we propose a

formula to estimate the optimal size of the adapted

document which is based on the user's situation.

2 RELATED WORKS

Since now, there have been great efforts to provide a

unified web which is accessible from many types of

devices. Device independent access to information

has been identified as a fundamental principle of the

W3C Device Independent Working Group (Gimson

2003). Several approaches for providing adaptive

web content delivery over heterogeneous and

dynamic environments have been investigated.

Often in these approaches, markup languages play a

major role to provide an adapted content for

different contexts. Some of these approaches try to

extend existing languages and content models with

new elements and attributes (Lewiz 2002, Mandyam

2002). The other approaches try to define new

languages which are completely device independent

(Grassel 2002). Their objective is to build a new

Naderi H. and Rumpler B. (2007).

PHYSICAL DOCUMENT ADAPTATION TO USER’S CONTEXT AND USER’S PROFILE.

In Proceedings of the Ninth International Conference on Enterprise Information Systems - HCI, pages 92-97

DOI: 10.5220/0002384300920097

 SciTePress

markup language that can be used to create the new

generation of documents which can be displayed on

various devices in different contexts. These

approaches use a new tag set to highlight the most

pertinent parts of a document according to different

contexts. These approaches have some limitations:

1- They require some efforts from the author and

they are time-consuming. 2- They produce different

versions of a document for different contexts.

Therefore finding an appropriate version of a multi-

version document to a given context is a difficult

task because there is not an ideal method to match

the user's context to the author's defined document's

context. 3- The author has to produce the different

versions of a document for each context. As there

are a lot of contexts, it's unlikely to produce a

specific version of a document for each context. 4- It

may be feasible for the next generation of

documents but as we have mentioned there is a

serious need to adapt the actual documents (Kurz

2004).

We believe that the user's situation can be used

to roughly determine the characteristics of the

adapted document. In this paper, we propose a

mechanism to automatically adapt a given document

based on the user's situation. For example, if the

user's situation indicates that the user has a mobile

device unable to display the pictures, then the

system must eliminate the document's pictures

3 DOCUMENT ADAPTATION

Indeed, our adaptation system is based on two types

of adaptations: Physical Document Adaptation

(PDA) and Semantic Document Adaptation (SDA).

By separating the PDA from the SDA, it will be

easier to build a document suited to the user's

situation. In the two following paragraphs we will

briefly explain each of these adaptation types and

then the PDA will be elaborated in this paper.

SDA answers to this question: which textual

parts of a document must be selected to be displayed

on the user's device to semantically satisfy the user's

request? The SDA approaches are similar to the

document summarization methods. In this step, the

wasted information will be eliminated from the

original document, based on the user's situation and

his query in order to enrich the document's content.

PDA is the process of selection, modification or

elimination of the physical components of the

document (cf. §5) based on the user's situation.

Recognizing the size of the adapted document (the

total size of the selected components to create the

adapted document) plays an important role in a

successful adaptation. In other words, a competent

adaptation system must be able to estimate the size

of the adapted document according to the given

situation.

The size of an adapted document is a dynamic

value related to the user's situation. More the user's

situation is restricted more the size of adapted

document must be reduced. An adaptation system

which is not able to estimate the size of an adapted

document may generate a too long or too short

document. For example if the system sends a lot of

information to a mobile user who is time restricted,

he can’t efficiently find what he wants. Inversely, he

will not be satisfied when the system sends him

insufficient information.

In what follow we explain our methodology to

create the textual part of adapted document. The

steps are:

1- Extracting and sorting the pertinent phrases

based on the user's requirements.

2- Estimating the size of the textual part of the

adapted document based on the user's

situation.

3- Calculating the number of phrases (N) which

must be displayed on the user's device.

4- Selecting the N most pertinent phrases form

the sorted list of phases.

5- Displaying N selected phrases to user by

respecting their order in the original

document.

Knowing the number of phrases to display on the

user's device is essential to respect the order of the

phrases in the original document. We propose a

formula in section 6 to estimate the optimal size of

the adapted document.

So far it has been shown that the user situation

has an important role in document adaptation. In

what follows we describe our definition of user's

situation, which is essential in the PDA process as

well as for estimating the optimal size of an adapted

document.

4 USER'S SITUATION

In most of the proposed systems, the user's

preferences and the context's characteristics (such as

user's device) are grouped in the same category. In

such systems, if a user uses two different devices,

the system will associate him two different profiles.

This problem is because the context includes the

user's characteristics.

PHYSICAL DOCUMENT ADAPTATION TO USER’S CONTEXT AND USER’S PROFILE

Time Availabl

Time

Location

Device

NetworkSoftware

Context Profile

atial-tem

oral EnvironmeAccess Platfor

Histor

Securit

User Profile

Preferences of

deliver

Domain of

interest

Identification

User Situation

Figure 1: User's situation components.

In order to solve this problem in our system, we

propose to separate the user profile and the context

profile. Thus in our system both of them characterize

the user's situation (Figure 1). The parameters in

Figure 1 are represented by three different colours to

illustrate their different roles in the semantic or

physical adaptation processes. The white elements

are related to the SDA, the black elements are

related to the PDA, and the components that can be

used for SDA and PDA (i.e. user's personal data) are

shown by the grey-white elements. Thus according

to Figure 1, SDA is roughly based on the user profile

and PDA is based on both the context profile and the

user profile (Figure 2).

As PDA is affected by both user and context

profiles, sometimes some conflicts may appear

between the user's delivery preferences and the

user's context. For example, a user may want to see a

picture (preference) but his screen (context) is not

able to display it. So a decision engine has been

designed in our system to solve such conflicts

between the user’s context and the user's delivery

preferences (Lum 2002, Naderi 2006a).

Figure 2: PDA based on context and user's preferences.

As mentioned, adapting a document is based on

the user's situation. The more the user's situation is

restricted the more the original document must be

adapted. So characterizing the user's situation plays

an important role for adaptation systems. In what

follows we will explain the main parts of the user's

situation: the context profile and the user profile.

4.1 Context Profile

The context's components include the user's device,

the network, the user's available time, etc. (Figure

1). These components impose restrictions on the

process of PDA. For example a mobile device may

be unable to display a picture because of the

network's speed or because of the device's

capabilities. As it has been shown in Figure 1, the

only element from the context profile which is

related to SDA is Location. It means that this

element is a SDA related component and it has no

influence on PDA process.

In our proposition, we separate the contextual

information into three main categories:

Access platform: divided in three sub-categories:

- Device: represents the basic hardware features such

as screen size, memory capacity, etc.

- Network: such as the network’s price, bandwidth,

reliability, etc.

- Software: such as available browser, virtual

machine, etc.

Spatial-temporal: includes three parts:

- Available time: user's availability time.

- Time: represents the current time and date.

- Location: user’s absolute or relative address.

Environment: the conditions which can influence

the user’s behaviour (noise...).

In the section 5, we will explain how a document

can be physically adapted, based on the black

elements (Figure 1).

4.2 User Profile

The user profile is another component of the user's

situation, which plays an important role in the

process of the SDA. The main role of user profile is

to help the adaptation system to semantically adapt a

given document to the user's situation. Without a

user model, a system works exactly in the same way

with all users. But users are different: they have

different background, different knowledge about a

subject, different preferences, goals and interests.

Physical Document

Ada

tation

Context

User profile

User's delivery

Document

Adapted

document

ICEIS 2007 - International Conference on Enterprise Information Systems

The user's interactions will be registered in the user

profile in order to recognize his preferences and

domain of interest. As it has been shown in Figure 1

"user's delivery preferences" is the only black

element from the user profile which affects the

process of PDA. The user's preferences will be

recorded in this component to more properly display

a given document on his device. For example, a user

may indicate in his profile that he doesn't want to see

any pictures on his device, or he just wants to see the

title and a small summary of the document.

We have defined a typology of the user profile

based on five dimensions:

- Identification: includes the user's identity

information such as name, job, etc.

- Domain of interest: expresses the characteristics of

information that the user wishes to obtain. It affects

the process of SDA.

- Preferences of delivery: It covers the flexible

constraints for restitution of information to the user.

PDA depends on this part of the user profile.

- Security: includes all information about the

security and privacy such as login/password.

- History: manages the interactions between the user

and the system. It can be used to find pertinent

information to users (Naderi 2006b).

5 PHYSICAL DOCUMENT

ADAPTATION

In the previous section we have introduced two

profiles which define the user’s situation. Both of

them have an important role in the PDA process. In

this section we study the different possibilities for

the document adaptations to the user’s situation.

As we have mentioned before, PDA means

selecting, adapting and/or eliminating document's

components (or fragments) to satisfy user’s and

context’s restrictions. Defining the document

fragmentation is an important part of PDA. The

fragmentation is used to adapt the document to

different contexts. Once document fragmentation is

defined, the system can easily use the different

components according to the user's situation (context

+ user's preferences).

To simplify the PDA process, we have defined

two types of components: structural components and

logical components. According to these document’s

components, we have defined two types of PDA’s

respectively: structural document adaptation and

logical document adaptation. In what follows we

describe these two types of PDA.

5.1 Structural Document Adaptation

The structural components include the objects such

as: texts, figures, tables, diagrams, formulas, etc.

Different devices can display the different structural

components with different properties. For example a

device can display simple tables but it can’t display

complex pictures. By having the device’s

characteristics (in the context profile) such as the

screen size, screen resolution, number of colours etc;

the system can recognize the device's ability to

display a particular component. The aim of the

system, in the structural document adaptation, is to

select or modify the structural component. If the

adaptation is not possible the system will eliminate

the component before displaying the document.

The information concerning the ability of a

device to display a particular component can be

automatically extracted from the device’s

characteristics or it can be manually registered in the

device's profile by the user. By referring to the

device’s profile, the system determines whether a

component can be represented, or must be modified

or eliminated from a given document.

Beside the context’s role, to select, modify or

eliminate the structural components; the user himself

can specify his preferences to display the different

structural component. The user, by specifying his

preferences in his profile, enables the system to

select the appropriate components to construct an

appropriate adapted document. Therefore the system

will be able to automatically select, modify or

eliminate the structural components from the

original document by referring to the device’s

profile (a part of context’s profile) and the user’s

delivery preferences (a part of user’s profile).

5.2 Logical Document Adaptation

In this step, the structural document adaptation

subsystem sends a partially and structurally adapted

document to the logical document adaptation

subsystem. The logical components are the textual

parts of a document such as title, authors, abstract,

keywords, sections, subsections, conclusion,

references, document's summary, etc. The context

profile has no impact on the selection of the logical

components. Only the user can specify his

preferences about the representation of the logical

components. For example a user may only be

interested in reading the title and some keywords of

a document. Another user may be interested to have

an automatic summary of the document.

PHYSICAL DOCUMENT ADAPTATION TO USER’S CONTEXT AND USER’S PROFILE

Consequently the system can define the

appropriate document modifications by referring to

the user profile (structural and logical adaptation)

and the context profile (structural adaptation).

Therefore roughly all of the possible document's

modifications (structural, logical and semantic) can

be found in the user profile and in the context

profile. Thus by having enough information about

the situation, a document can be automatically

adapted to the user situation. The size and the

content of the generated document are the only

important elements which can't be directly

determined from the two mentioned profiles. In what

follows, we propose a formula to estimate the

optimal size of an adapted document, based on the

user situation.

6 SUMMARY’S SIZE

ESTIMATION

There have been some attempts to calculate the

optimal size of the document’s summary. Sweeny

(Sweeney 2003) has mentioned some static

percentage to generate an adapted document. For

example he suggests that a summary of 3% is the

most optimal value for the mobile phones (without

considering the size of the document!). But as

previously mentioned, the size of an adapted

document is not static and it must be dynamically

calculated, based on the user's situation. Although

the user can’t explicitly specify the size of the

needed information for the different situations,

however it can be calculated from the user's

preferences and the other contextual parameters. In

this section we try to list all of the parameters which

can influence the adapted document's size.

Subsequently, we propose a formula to show the

relationship between these parameters and the size

of the adapted document.

- Screen size N(ch): This is one of the most

important parameters for calculating the adapted

document’s size. The more the screen size is small,

the more the size of the delivered information must

decrease. The unit we choose to measure the screen

size is the number of characters which can be

displayed on the device’s screen. So it depends on

some of the user’s preferences such as the font's size

registered in the user profile.

- User’s availability time T

(s): The more the user is

time restricted the more the size of the adapted

document must decrease. The value of T

can be

explicitly supply by user, extracted from his agenda,

or it can implicitly be estimated based on the user’s

daily behaviours.

- Battery’s availability time T

(s): Contrary to the

desktop devices, the energy of a mobile device is

limited, so the system must consider the remaining

time of battery’s charge.

- Network network_effect: As mobile networks are

slower and more unreliable than wired networks, the

efficient time of connecting a mobile device to

network may be less than a desktop PC one.

- Speed of reading V(ch/s): the size of adapted

document depends on the speed of reading of user.

This speed can be calculated explicitly by testing the

user, or implicitly by observing the user’s behaviour.

- Input facility α: This factor represents the required

time to get a new page. When the screen size is

small the user needs more time to interact with

system to display the new page and the system also

needs some time to display the required page.

The above situational parameters limit the

efficient users time. By considering these parameters

the appropriate size of the adapted document (S) can

be calculated as follow:

(1)

In this formula t is the elapsed time because of

contextual restrictions such as the small screen size.

It's equal to the number of displayed page (S/N-1)

multiplied by the required time to change one page

(α

). According to the above formula, the more the

screen size (N) decreases, the more the number of

displayed pages (S/N) increases and therefore the

more the elapsed time increases. T is the efficient

user’s availability time that can be calculated form

the following formula:

(2)

The above formula permits to calculate the size

of a textual summary which simply can be used to

calculate the number of phrases which must be

displayed on the user's device. When the user is just

interested in the textual contents (or the original

document doesn't include the other physical

components such as picture) the above formula can

effectively work. But sometimes the users may be

interested in the other physical components. In this

case, the user requires some time M to read the

1_0

_)T,(TminT

≤≤

×=

effectnetworkwhere

effectnetwork

VαN

α)(TVN

)-T(V

])1

(-T[Vt)-(TV TV S

−××

×+

⇒×−×=×=

′

×=

ICEIS 2007 - International Conference on Enterprise Information Systems

physical components (i.e. pictures, tables, graphs

etc). So the formula (1) will be replaced by:

(3)

The calculation of the required time to read the

non textual physical components is out of the scope

of this paper and requires more efforts. The user can

explicitly specify the average value of M in his

profile.

Therefore by the above formulas our system will

be able to estimate the optimal size of the

information which must be displayed on the user’s

device.

7 CONCLUSION

We have proposed a methodology to create

documents adapted to the user's situation. The user's

situation in this paper includes two main profiles:

context profile and user profile. Physically

Document Adaptation (PDA) and Semantically

Document Adaptation have been defined and then

we have explained how the situational parameters

can be used to physically adapt a given document to

the user's requirement. We have divided the PDA

process into two sub-processes: structural document

adaptation and a logical document adaptation. We

have explained that knowing the size of an adapted

document plays an important role in an adaptation

system. So we have proposed a new formula which

is able to estimate the size of the adapted document,

based on the user’s situation.

ACKNOWLEDGEMENTS

This research is partially supported by the French

Ministry of Research and New Technologies under

the ACI program devoted to Data Masses (ACI-

MD), project #MD-33.

REFERENCES

Gimson, R. et al., 2003, Device Independence Principles,

W3C Working Group, http://www.w3.org/TR/di-princ

Grassel, G., Lauff, M., and Spriestersbach, A., 2002,

Definition and Prototyping of Renderer-Independent

ML. Nokia, SAP and IBM Germany. In W3C

Workshop on Device Independent Authoring

Techniques, SAP University, St. Leon-Rot, Germany.

Kurz, B. Popescu, I. and Gallacher, S., 2004, FACADE - a

Framework for Context-aware Content Adaptation and

Delivery, In Second Annual Conference on

Communication Networks and Services Research,

Canada.

Lewis, R., 2002, A Device Independent Method for Web

Site Authoring. In W3C Workshop on Device

Independent Authoring Techniques, SAP University,

St. Leon-Rot, Germany.

Lum, W.Y. & Lau, C.M., 2002, A Context-Aware

Decision Engine for Content Adaptation, In Pervasive

Computing, IEEE Press, Vol. 1, No. 3.

Mandyam, S., Vedati, K., Kuo, C. and Wang, W., 2002,

User Interface Adaptations: Indispensable for Single

Authoring. In W3C Workshop on Device Independent

Authoring Techniques, SAP University, St. Leon-Rot,

Germany.

Naderi H., Rumpler B., Pinon J.M., 2006a, A Two

Layered personalized Information Retrieval System, In

MAPS'06 : International Workshop on Multimodal

and Pervasive Services ICPS’06. IEEE Press, France.

Naderi H., Rumpler B., Pinon J.M., 2006b, An Efficient

Collaborative Information Retrieval System by

Incorporating the User Profile, In AMR 2006, to

appear in Springer Lecture Notes in Computer

Science.

Sweeney, S., Crestani, F., 2003, Supporting Searching on

Small Screen Devices Using Summarization. In

Proceedings of Mobile HCI 2003 International

Workshop. Lecture Notes in Computer Science,

volume 2954.

αVN

α)-M-(TNV

t)-M-(TV TV S

×+

=×=

′

×=

PHYSICAL DOCUMENT ADAPTATION TO USER’S CONTEXT AND USER’S PROFILE