Case study: Extending content metadata by appending user context

(1)

MSI Report 06099

Växjö University ISSN 1650-2647

Case study: Extending content metadata by

appending user context

Martin Svensson

Oskar Pettersson

School of Mathematics and Systems Engineering

Reports from MSI - Rapporter från MSI

(2)

Acknowledgments

Our first and most important acknowledgment is directed to Marcelo Milrad. Without your support, dedication and enthusiasm this wouldn't have been possible! We would also like to thank the remaining members of the CeLeKT-group for their willingness to help whenever there were questions to ask. Additionally, we would like to thank Jo Skamedal for the great tutoring given in the mysterious ways of writing theses.

A special acknowledgment goes to Per Flensburg and Gunnar Mosnik for the encouragement given to us during our time at Växjö University, both inside and outside of projects. Cheers and thanks for believing in us!

(3)

Abstract

Recent developments in modern computing and wireless networks allow mobile devices to be connected to the Internet regardless of their physical location. These mobile devices, such as smart phones and PDAs, have turned into powerful multimedia units allowing users to become producers of rich media content. This latest development contributes to the ever-growing amount of digital material existing on the World Wide Web, and at the same time creates a new information landscape that combines content coming from both, the wired and mobile Internet. Thus, it is important to understand the context or settings in which mobile devices are used, and what is the digital content produced by the different users. In order to gain more knowledge about this domain, we have investigated how to extend the standard metadata related to content with a metadata domain describing the context, or settings in which the content has been created.

In order to limit the scope of our work, we have focused our efforts in a specific case taking place in a project called AMULETS. The AMULETS-project contains all of the elements we need in order to resemble the contextual setting in a metadata model. Combined with the technical metadata associated to the digital content, we try to display the benefits of capturing the different attributes of the context that were present when the content was generated. Additionally, we have created a proof-of-concept Entity Relation (ER)-diagram which proposes how the metadata models can be implemented in a relational database. As the nature of the thesis is design-oriented, a model has been developed and it will be illustrated throughout this report. The aim of the thesis is to show how it is possible to design new metadata models that combine both relevant attributes of the context and content in order to develop new educational activities supported by location-based services.

(4)

Index of figures

Figure 1: Challenges and approaches in scenario-based design...7

Figure 2: A definition of context...13

Figure 3: The communications cycle...14

Figure 4: A Semacode bar code...17

Figure 5: Context added to the description of content...23

Figure 6: AMULETS-project case sketch...27

Figure 7: Nokia 6630...28

Figure 8: HP 6515...29

Figure 9: The CCS architecture...30

Figure 10: Metadata extraction and generation...30

Figure 11: Presenting the result of the game...33

Figure 12: Web-interface...34

Figure 13: Refining content search...35

Figure 14: Selecting content...36

Figure 15: Presenting content...37

Figure 16: Stage 1/5, real-life model...40

Figure 20: Stage 5/5, real-life model (finished)...43

Figure 21: Technical metadata model (finished)...44

Figure 22: Two separate domains...46

Figure 23: Two separate domains interrelated...46

Figure 24: Relations within a domain...47

Figure 25: Relations and their dimensions...48

Figure 26: Stage 1/4, modeling the database...57

(9)

1. Introduction

1.1 Background

“Information – anytime, anywhere” is a concept which describes the current trend in information sharing (Spyrou et al., 2004). Users are, in this respect, no longer limited to physical constraints due to wireless technologies applied in devices such as cellular phones, PDAs and laptops. While time tolerance between data availability and data accessibility decreases, the amount of information shared in this manner grows rapidly. Users may find themselves feeling lost in the information jungle, which is why time tolerance is not only a matter of technical improvements – it is also a matter of data structuring in order to contextually provide suitable content in applicableformat to wireless users.

Wireless devices come in many different shapes and sizes. For instance, different models of cellular phones have different technical limitations regarding the amount of memory, screen size and data format compatibility. Content generated andprovided by one multimedia device might be incompatible with other devices. Considerable amount of research has been conducted in separating content from presentation, where techniques such as XSLT1_{makes it possible to visualize}

content in different ways. The same content may, for instance, be presented as plain text or as an RSS-feed depending on the constraints of the device used. Existing technological tools, such as Apache Cocoon2_{, give developers the possibility to build dynamic applications, by automatically manipulating}

and generating content to match the target systems format constraints. Based upon these new ways of handling data, developers and IT architects find themselves having access to the building blocks for making data device compatible.

While certain content may be accessible to many recipients, via the technology previously mentioned, doesn’t automatically mean that users want content in a predetermined format. A device may be able to present data in several ways, such as plain text, video or text-to-speech. These customization aspects must be caught in order to bring as much value content to the user as possible. Users, on the other hand, won’t use services cumbersome to personalize. This is why metadata, the data about data, can be used asa valuable source for personalizing media by demanding minimaluser interaction. Examples of extractable metadata from the “information landscape” could be the identity of the content creator device or the time of day the content was created. This descriptive data attached to the media makes it possible for application logic to analyze the media (Shankaranarayanan & Even, 2006).

This metadata in combination with the explosion of the use of mobile devices opens a

(10)

whole new area of research and there are a lot of possibilities to draw advantage of. One of the areas where this is especially useful is within the field of Computer Support Collaborative Learning (CSCL), and one of the projects that has come furthest is the Savannah project.

The Savannah project (Benford, 2005), which is a project funded by several large corporations in the UK such as ESTA Futurelab, Hewlett-Packard Laboratories, the BBC and the Universities of Nottingham and Bristol has taken CSCL one step further. They have created a game for school children investigating the potential of wireless technology in a educational process called Savannah, in which children are to pretend that they are lions on the savanna. Instructions and content is sent back and forth as the children advance through the game by using the latest wireless technology. This project is currently being pilot-tested on a school in the UK and according to interviews with the students they have found the activity-based approach working very well.

Another example illustrating new ways in which mobile technologies are used to support new ways of sharing and collaborating is being conducted at Berkley University in California. One of their efforts is an interesting prototype mobile application, which captures relevant image metadata when using camera-enabled mobile phones (Davis & Sarvas, 2004). When the client application sends an image with the attached metadata to the server, it is automatically categorized at the server repository via person- and location-guessing algorithms. This categorization creates derivable “knowledge” of the images that, for instance, makes it possible to select content sent to the user more accurately.

We hope to further develop and explore the use of this derivable “knowledge” of media within the field of collaborative learning environments.

1.1.1 Motivation

The reason for us working in this area is our personal interests in combination with the on-going efforts (the MUSIS project) in this direction at our institution. The MUSIS (Multicasting Service and Information in Sweden) is a research project run by the CeLeKT-group at Växjö University, investigating and developing innovative techniques for mobile multimedia handling and content generation to be delivered to smart phones using wireless multicasting techniques3_{. The scope of our}

work was discussed with the MUSIS research teamin order to bring possible value to their future projects. In fact, the problem domain arose from needs of an upcoming project within CeLeKT4

concerning subjects introduced in the previous section.

3 _{http://www.musis.se/}_{, 02/16/2006}

(11)

1.2 Purpose

Our thesis aims to investigate how to model a metadata schema describing content generated by mobile multimedia devices, as well as the context in which this content was created in order to provide users with context-aware support while conducting intellectual tasks.

When reading about at the Savannah-project and the Berkley initiative mentioned in the previous section we thought that it would be very interesting to see a combination of these ideas, where content created in wireless environments is organized according to the context existing upon creation. With this in mind, three core questions have beenidentified that will serve as guidelines in order to narrow the scope and set the focus for our thesis. These are as follows:

1. How do we model and structure a metadata schema for content generated by mobile multimedia devices which also captures the context in which this content was created?

2. How and to what extent can the metadata schema increase the understanding of the created content?

3. What benefits are acquired from the metadata schema when selecting appropriate content for a single or a group of users?

The purpose of the first question is to investigate if it is possible to develop a generic metadata model that may serve as a general guideline, so the potential solutions will provide a good foundation while trying to achieve the main purpose of the thesis. Answering question number two will require an investigation of the metadata schema in order to determine how metadata related to content can be derived from the created schema. By doing this, it will be possible to discuss the level of understanding of the content which these metadata contributes with. In this sense, “understanding” means the knowledge of content contained in its associated metadata that may, for instance, support applications when selecting relevant content for the user. By trying to find an answer to the third question, we will try to explore the possible benefits that can be gained when using the metadata schema we will develop. By “users” we refer to end-users navigating an application using an underlying implementation of the metadata schema. To be able to pull this off we need to add limitations to further narrow the scope of this thesis.

1.3 Limitations

(12)

report to a real-world case referred to as AMULETS. AMULETS is an on-going research project to be carried out by the CeLeKT group aiming to make use of advances within mobile and wireless technologies for outdoor learning related activities5_{. In this particular context, we will conduct our}

efforts for pursuing the efforts related to this thesis. Limitations such as the type of mobile multimedia device used and the way it sends content are predetermined, as well as real-life activities performed. This limits our work to the possible ways of extracting, structuring and making sense of the metadata and activities existing in the view of this scenario. As the nature of the work to be pursued in this thesis is of design character, we will create and develop a conceptual metadata model which will be technical implemented and evaluated in our future efforts related to our upcoming master thesis. These efforts will be conducted during the Fall 2006 and the Spring 2007.

1.4 Intended audience

The intended audience for this thesis are people with knowledge in the field of informatics and system science. Chapter 7 will demand some additional knowledge within the field of relational databases to be fully understood.

1.5 Technical approach

As part of the practical efforts related to our work, we will map the metadata model/models to an ER (Entity-Relationship) diagram. The reason for choosing this modeling technique is that it is the most popular high-level data model used in database design to support communication between users and designers (Connolly & Begg, 2002). The schema should be seen as a proof-of-concept implementation of the metadata model/models. As relational databases are well-known within the database community we decided that this would best fit its purpose as proof-of-concept.

1.6 Expected results

The outcome of our work is intended to be a first draft of the metadata schema to be used in the

AMULETS project. With this thesis we expect to be able to display the increased understanding of

content created when capturing the real-life context in which the content was created. Additionally, we intend to determine what benefits there are to gain from this increased understanding. Our ambition is that our work will allow the creation of an ER-diagram and an associated conceptual metadata model which support the description of any type of context – thus making it general enough to be used in other areas than in the AMULETS project.

(13)

1.7 Disposition

The initial part of the thesis will consist of a study based on an existing case, originally created by the CeLeKT-group. We will analyze every point in the case where data is gathered in order to explore different instances of metadata that can be collected and stored in a repository. This will be achieved through the use of the scenario-based methodology (Carrol, 2000), where we create scenarios out of small portions of the case description. This step will also include mapping the real-life concepts together and figuring out how these could be related to content created in the case. The models will be built step-by-step as we make conclusions by relating to the existing scenarios/case. The output of this step will be metadata model/models mapping the content metadata and the real-life concepts.

For the second part we investigate to what extent the metadata model/models can increase the understanding of created content. To investigate this we need to discuss the different aspects of relations and domains, which will give us enough knowledge to validate the models created. The validation will make sure that the aspects of relations and domains are applicable to the model/models created in the previous step. This work will answer question two, as well as contributing to the answer of question one in section 1.2. The third step aims to answer question three in section 1.2, which concerns the value of the model/models. As we are unable to measure the value of them in the case, we will use the scenario-based design methodology to create stories of the possible use of the collections of content created. The stories will then be analyzed and the use of content will be traced to the metadata model/models created in the first step.

The final step will consist of the creation of an ER-diagram supporting the metadata model/models created in part one. This will be done step-by-step by motivating the creation of tables, the relations among these and the new knowledge that can be inferred based on these relations. The output of this step will be a full ER-diagram which supports the implementation of the metadata model/models. With the creation of this diagram, along with the results of the previous steps described above, we intend to have answered the main purpose of thesis. We conclude this thesis by giving some reflections and directions of our future work.

To summarize, the work that will be described in the coming chapters has been organized and conducted according to three distinct phases:

• Literature review

• Creation of a conceptual model for the contextual metadata

• Modeling aspects related to this model and its validation

(14)

2. Method

In this section we will describe the methodology which will support us when pursuing the purpose of this thesis. Choosing a suitable way of organizing and planning the work at hand will increase the overall quality of the thesis as well as contributing to its validity. Before introducing methodologies chosen for this thesis we'll give a brief description of how we aim to conduct our work.

2.1 Methodological Approach

The specific domain related to the scope of this thesis was chosen after several meetings with Marcelo Milrad, who is one of the head researchers at the CeLeKT-group. We had been involved in previous projects within this group, which is why we knew that the group shared our interest in mobile communications and its possibilities. After the first meeting, we spent a couple of weeks reviewing various publications (including books and journal articles) within the chosen domain in order to identify a challenging purpose and focus for this thesis. During this time, we had regular contacts with Marcelo in order for us to stay on track. When conducting the literature review, we realized that the work at hand could be structured within three main phases (briefly mentioned in section 1.7). The first phase, literature review, is partly completed as this is being written. The following phases, creation of

a conceptual model for the contextual metadata and modeling aspects related to this model and its validation, was realized while we were gaining an understanding of the domain under investigation

In order to find out how to capture the context existing when content is created, we will have to relate this activity to some concrete actions. This is why phase two of our work involves modeling the embedded context when content is created in a specific case (mentioned in section 1.3). Once we have created a conceptual model describing this specific context, we can use the model to understand the possible use and benefits there are to gain when contextual metadata is appended to created content. This has been conducted as phase three of our work. Upon the completion of all these phases and the associated results, it can be claimed that the purpose of the thesis has been achieved. The results of our work to be presented in the coming sections will hopefully show (while applied to a specific case in this thesis), how and what benefits of appending context to content can be gained from modeling any contextual setting. Serving as a proof-of-concept, we will conclude our thesis by presenting an ER-diagram capable of storing any type of contextual metadata and their relations. Our ambition is that an overlying application, such as a “scenario generator”, can use an implementation of our ER-diagram for building custom contexts similar to the case context modeled in this thesis.

(15)

methodology and explain why it supports the work and efforts described in this thesis.

2.2 Scenariobased design

Scenario-based design is about envisioning and facilitating new ways of doing things and new things to do by analyzing how people accomplish tasks. The tasks to be analyzed are described in scenarios, which are stories about people and their activities. Scenarios have some characteristic elements which are always present (Carroll, 2000). For one, a scenario always include a setting which describes the “stage” in which the scenario takes place. Furthermore, scenarios includes agents or actors which are humans involved within the scenario that typically have goals and objectives. The goals and objectives are the changes that the actor wishes to achieve in the scenario setting.

Each scenario has a plot describing sequences of actions and events. These are the things that actors do and experiences as well as events that triggers changes in the settings of the scenario. The events may or may not contribute to the completion of an actors goal. Scenarios can be represented in several different ways, such as written text, storyboards or videos.

The figure below (figure 1) visualizes five challenges associated with system design as well as the corresponding response of based design. In order to display the value of scenario-based design we will now introduce five general challenges for system designers, their solutions through scenario-based design and why this methodology will be a good choice for our upcoming work.

(16)

2.2.1 Action versus Reflection

When developing a system with a group of professionals, the reflective side of a step in the design process is tricky to catch. All individuals take some pride in what they do, which may leave reflective thoughts out of a discussion in fear of being questioned competence-wise. The opinion of one individual may be trivial and irrelevant, as well as it might make the end-result a lot better if put to surface. Scenario-based design provides for reflection, as they describe an implementation of decisions already made (Carroll, 2000).

2.2.2 Design Problem Fluidity

Requirements always change, especially when using and/or developing cutting-edge technologies. The time-frame for a development project may be altered, developers may quit as well as new designers may join the development group. If not traced, the result of the development project may be a successful one in creating something which conforms to the initial requirements – but not to the requirements adjusted “along the way”. Scenario-based design is in its nature a flexible way of capturing requirements. If a scenario is changed, the list of requirements could be analyzed and altered according to the changes made in the scenario.

2.2.3 Design Moves Have Many Effects

Changes in the design process of any extent will most probably affect other elements involved in the design. As a scenario in itself is a story, the author of the story may decide the level of complexity and detail in which the story is written. This allows for the creation of several views of the same scenario that, when adjusted to fit the changes accordingly, may shed light on consequences that might had been left unattended for.

2.2.4 Scientific Knowledge Lags Design Application

In some cases the present scientific solutions lags design, either because they make the solution of the problem at hand more complicated than it needs to be or that there isn't a technology that solves the problem at hand. By using scenarios the dilemma can be abstracted and categorized, which provides for a clearer overview when applying the combined knowledge of the development group.

2.2.5 External Factors Constrain Design

(17)

Scenarios are design objects that describes the system by telling how the users tries to use parts of the system that provides for reflective thoughts of the front-end of the system.

2.3 Why scenariobased methodology?

When considering the choice of method we started off by determining our view of how to best fulfill the purpose of our thesis. After a lot of thinking, we concluded that a full-fledged inductive approach would have been too time-consuming and not contributing enough to the purpose of our thesis. Additionally, the fact that induction relies on general conclusions being made from empirical data, such as observations and interviews (Thurén, 1991), would have put a lot of load on the CeLeKT-group. They would undoubtedly have been our main source for collectible empirical material. Additionally, the project interested in the outcome of this thesis is still ongoing, which easily leads to change of requirements which may affect our work. As scenarios contains requirements within the stories told, these could be adjusted by simply updating the text.

When we had decided that induction wasn't a viable option for us we investigated the deductive approach. Deduction relies on a logical conclusion being made from a set of related hypotheses initially considered as true, which are tested empirically throughout the ongoing research (Graziano et al., 2002). This methodology was, from our point of view, not a viable one as we considered the purpose of this thesis unmeasurable in terms of either true or false.

After determining that neither induction or deduction would do for our upcoming work, we realized that the purpose of this thesis would benefit from a methodology focusing on aspects such as design and creativity. We agreed that the scenario-based methodology would be a very good selection for the sake of our purpose. It lets us gather requirements in a reflective manner and triggers discussions in the ongoing work which we think will affect the end-result in a positive manner. We will also be able to set the appropriate level of complexity for our work by writing the scenarios more or less technical/detailed. This will create a natural boundary for us to work within. Additionally, as the purpose of this thesis aims to display the benefits of the result for end-users, we can create scenarios where the value of the repository is shown. These scenarios will most likely trigger further ideas contributing to the end-result.

Engineering research efforts conducted within the field of information systems research can rely on many different research techniques. One of them is called construction6_{, involving the}

conception, design and creation of an artifact or a technique, which is a perfect fit for the purpose of this thesis. When using the construction technique, the design is based on theoretical foundations and is usually tested in performance to measure whether the construction was successful or not. As we have selected the scenario-based design methodology to support our upcoming work, we will not be

(18)

able to determine the actual value of the design in figures and numbers. Although this would have been interesting to review, it wouldn't contribute to the main purpose of the thesis of creating a metadata schema. Instead, the benefits of the schema will be displayed and discussed thoroughly, while the actual measurement and implementation of the result of this thesis is planned to be pursued as a future effort in our master thesis.

2.4 Reliability

Reliability means that the surveys are correctly executed, and regardless of who is conducting the survey the end result has to be the same (Thurén, 2003). Because of the methodology chosen we will have to see the thesis reliability a little bit differently. The case which this thesis will be built upon was supplied to us by the CeLeKT-group, and we will use this case as a boundary for writing scenarios. The design of our metadata models and structures will be described and motivated throughout the thesis, and can therefore be recreated. Furthermore, we will not claim that our solution is the best possible, but rather state the benefits which can be gained from using it. Reliability regarding conclusions made throughout the thesis will be grounded on existing research material. In this way we aim to solidify the reliability of the outcome as far as possible.

2.4.1 Case: AMULETS

A case had to be selected for this thesis to build scenarios upon, and after several meetings with Marcelo Milrad the decision fell on the AMULETS case. As the purpose of our thesis was outlined in close collaboration with Marcelo, the case created by the CeLeKT-group was a perfect fit for our work. One reason for this choice is the possibility to arrange meetings without much hassle as we live close to the research center. Clarification of uncertainties in the case report is only a stroll away, plus, discussing face to face eliminates potential misinterpretations of e-mails and letters. Another reason is the enthusiasm of our work shown by the CeLeKT-group. The outcome of this thesis will provide them with valuable information of how to structure the metadata repository. All of the scenarios created by us will be presented to and discussed with the CeLeKT-group to assure that they stay within the intended scope of the case.

2.5 Validity

(19)

(20)

3. Theoretical foundations

The theoretical foundations of the thesis aims to cover different areas that are relevant for the scope of our work. We will begin by discussing issues related to metadata, followed by a brief introduction to ubiquitous computing. In the third and fourth section we will cover relevant positioning technologies and web technologies respectively. When these topics are explained, we will move on to explaining the state of the art within ubiquitous computing utilizing positioning technologies and web technologies which finally leads us to the section “Putting it all together”. This section shows why the different topics are relevant for our upcoming work, as well as explaining how our approach differs from the state of the art of ubiquitous computing.

3.1 Context

There have been a lot of effort put into defining the word context, amongst the better ones are by Brown (1996) which defines the word “elements of the user’s environment which the computer knows

about“. That is a perfectly good definition of the word, but it's not quite enough for the scope of this

thesis and as there already has been published a very good definition of the word within the CeLeKT group (Kurti et al., 2006) we are going to use the definition “information and content in use to

support a specific activity (being individual or collaborative) in a particular physical environment”

and this will be the bases for our whole design approach for context.

This definition makes our definition of context into a three-layer structure consisting of the following attributes: location / environment attributes, activity / task attributes and personal / interpersonal attributes (figure 2). All these attributes are interdependent, meaning that they all depend on each other and all of them are contributing to the end result of the thing it is describing.

(21)

3.2 Metadata

Metadata literally means “data about data”, this definition implies that metadata can be anything that describes something else, and that makes this a very wide subject that can be implemented in a lot of various places. Traditionally, cultural heritage and information professionals such as museum and library workers has been doing this without digital help for a long time, so metadata is not something recently invented. There exist a lot of standards in this area and they all have different advantages and disadvantages but the lot of them ensures that information professionals have a wide array of standards to choose from so that they can do their work efficiently. All of these perspectives and standards lead to a very broad conception of metadata and what it is, so to understand it better it is very helpful to break it down into distinct categories – administrative, descriptive, preservation, use and technical metadata (Gilliland-Swetland, 1998).

Administrative metadata is used for such tasks as location information and version control. Descriptive metadata is used to describe or identify resources for such tasks as specialized indexes or annotations by users. Preservation is used for metadata related to the preservation management of information resources, which include such tasks as documentation of the physical condition of resources. Technical metadata is related to how a system functions or metadata behaves

(22)

and include things such as hardware and software documentation and formats, compression rations, scaling and such. The last type is the use metadata and it handles metadata related to the level and type of use of information resources, this includes, for example, use and user tracking and exhibition records. A combination of these categories of metadata generates an understanding of something on different levels that can be used to let technical devices adapt its behavior accordingly. This idea is often referred to as ubiquitous computing.

3.3 Ubiquitous computing

Ubiquitous computing, originally brought up by Mark Weiser, aims to move the concept of computing from being a distinct object into an integrated part of the environment7_{. In gaining understanding of}

the environment and the surroundings of a user, systems can be programmed to contextually adapt their behavior accordingly. The ultimate goal of ubiquitous computing is to reach a level of interaction with the environment (figure 3) where the computing disappears into the fabric environment, so that the user only sees the task8_.

A lot has happened in ubiquitous technology since Mark made his statement in 1988. Most people carry a cellular phone wherever they go, which has computing-power exceeding mainframes of the late eighties by far. But in terms of ubiquitous computing the cellular phones are still in the crib. As of today, most of these devices don’t have tools built-in for determining its exact location. Nevertheless, ubiquitous computing regarding cellular phones has great potential. Imagine the following scenario; a person visits the local library when his cellular phone receives a call. Instead of sending out a loud

7 _{http://en.wikipedia.org/wiki/Ubiquitous_computing}_{, 03/10/2006} 8 _{http://semacode.org/}_{, 03/10/2006}

(23)

signal in order to notify the person of the call, the cellular phone could be “aware” of its present location and act more discretely. Perhaps by using the vibrator instead of the sound signal in order to avoid an embarrassing situation for the owner. While the technology aren’t there yet, there are some interesting initiatives worth studying closer.

3.4 Positioning technologies

This section will introduce several technologies which can be used to inform or to determine the position of people. Some are by means of satellites and others are by means of some kind of digital notes with embedded coordinates. Knowing the location of a user is an important ingredient in ubiquitous computing, and as the AMULETS-project will let the multimedia devices act based on their position-awareness, we will discuss some of the existing technologies within this field.

3.4.1 GPS

The Global Positioning System (GPS) was founded as an American military project in the 1960’s under the US air force. The project was originally developed for use in tactical aircrafts that needed to be able to find out their position in three dimensions, anywhere in the world at all times to be able to locate where they, for example where suppose to drop bombs.

GPS uses 24 satellites that orbits around the earth every 12 hours at a height of 20-200km. Four satellites are located at each of six planes inclined at 55 degrees to the plane of the earth’s equator. Each of the satellites covers half of the earth with a signal about it’s own location, so the GPS device is with the information from the satellites able to calculate it’s own position on earth. It does this in a very accurate way with a fault marginal as little as 2 meters (Getting, 1993).

You can use this technology in so many implementations that listing them would be an essay alone, but this technology is best used in mobile devices such as phones, PDAs and computers within vehicles. It is for example used by firefighters to map forest fires via mobile devices (Patterson, 2006) and is also widely used in navigational systems on various vehicles like cars and airplanes (Tiano et al., 2001). GPS is an important part of the AMULETS-project (briefly mentioned in section 1.3) as it will be used to monitor the position of mobile multimedia devices during the learning activities. As GPS doesn't work indoors, we will now discuss other positioning technologies relevant for our work not sharing this constraint.

3.4.2 RFID

(24)

years, introduced as early as in the 1940s, the technology wasn't applied into industries until the late 1980s. The reason of this was the lack of a standard and the large implementation cost related to RFID (Holmqvist, 2006). Times have changed and today the technology is used commercially.

RFID are built upon electronic tags which content can be read by using tag readers, such as hand scanners or antennas. The tags comes in two main flavors – either active or passive. The active tags are able to send outgoing signals without first being contacted. As this variant depends on battery power its length of life is limited to 5 years. On the other hand, the passive tags doesn't need a battery to function. Due to this it cannot transmit a signal on its own, but instead collects enough power from the incoming signal to be able to send a response9_{. Although a viable option, the}

AMULETS-project uses GPS in combination with another kind of positioning technology – Semacode.

3.4.3 Semacode

Semacode is a system for ubiquitous computing combining camera phones, bar codes and URLs in order to bring the real world into the virtual world10_{. The bar code is a 2D-image (figure 4) that}

generates a URL-sting when scanned, which then can be accessed in a web browser. As Semacode only needs a Java environment, with a good-enough camera phone, the software can be run on many different cell-phone vendor models. Additional parameters can be passed in the URL-string as GET-variables, which the target system can extract and analyze. If location-based data, such as coordinates, were embedded in the Semacode URL, it would, for instance, be possible to determine the current location of the device scanning the bar code. As Semacode technology requires the direct interactivity of a user, it can be used to determine choices made at the location of the bar code. One example of Semacode usage within the AMULETS-project is question answering, where a set of related semacodes will be used as selectable options for a question.

(25)

3.5 Web technologies

To let devices of various kinds behave closer to the ubiquitous ideal, the devices must be able to determine its own limitations in creating and presenting content. This section will introduce web technologies relevant for the devices used in the AMULETS-project, as well as some technologies used in this thesis to display the value of content created in the AMULETS-project. The presentation application of the scenarios (sections 4.7 - 4.10) is one example of the usage of technologies that is part of the concept Web 2.0 (O'Reilly, 2005), including powerful technologies such as Ajax, but Web 2.0 is just as much a way of thinking as a collection of technologies. The web based presentation application is heavily user customized and relies upon dynamically generated content, which all are characteristics for an application to be a part of the Web 2.0. The first sections will introduce the reader to three technical solutions which the devices used in the AMULETS-project are limited to when creating content, while the two last sections will speak of two valuable techniques that can be used when presenting created content.

3.5.1 3GP

3GP is a multimedia container format which is defined and maintained by the third Generation Partnership Project11_{and is intended to be used in 3G phones. It is a simplified version of MPEG-4}

Part 14 and usually goes with the extension *.3gp or *.3g2.

3GP stores video streams as MPEG-4, H.263 or H.264 and audio streams as AMR-NB or AAC-LC. It does also transfer the most important bits first so the format is a streaming friendly

11 _{http://en.wikipedia.org/wiki/3GP}_{, 05/10/2006}

(26)

format. This format is commonly accepted as the most popular format to use when using videos in mobile devices.

3.5.2 JPEG

JPEG is a compression format developed and maintained by the Joint Photographic Experts Group used to compress photographic images in computers, digital cameras and other devices12_{. The format}

provides us with a better compression-format for photos than GIF-images does and the cameras in the devices used in the project uses this format, thus we use it in the project.

3.5.3 MP3

Invented by the German technology group Fraunhofer and Thomson, the music compression format MP3 quickly became the buzz of the web when it entered the stage and basically changed the whole music market overnight (Hacker, 2000).

The format MP3 is a quite complex algorithm to compress raw audio files, like the ones that exists on a CD. These files are quite large however, over 10 megabytes for a minute of music. MP3 however can reduce this size to around 1 megabyte for a minute of music which enabled music lovers all over the world to transfer their CD-collections onto their hard drives and still have an acceptable quality of the music.

The format does not analyze redundancies like the zip format does, it analyzes patterns in the music stream and compares it to what the human hearing can comprehend. The algorithm then discards all of the information that is unnecessary in the file to reduce the size.

As this is the most established standard for music today the choice naturally fell on the MP3 format.

3.5.4 Ajax

Ajax is not made up of one technology, but rather HTML, Javascript, DHTML and DOM combined. The word itself – Ajax – is shorthand for Asynchronous Javascript and XML, with “asynchronous” being the most interesting part, as its purpose is to bridge the gap between the functionality and interactivity of a desktop application and the always-updated web application by allowing asynchronous communication between client and server in web environments13_{. Normally, a web}

application requires the present page in the web browser to reload in order to communicate with the web server, but thanks to Ajax the client can initiate such communication instantly without reloading.

12 _{http://en.wikipedia.org/wiki/JPEG}_{, 05/12/2006}

(27)

3.5.5 Google Maps

Google maps is a map service provided by the company Google, partly as a application that you download, partly as a application directly in the browser and partly as an API that developers can use to make their own variations and enhancements of Googles application. The API allows developers to send in coordinates to Google maps and present the locations dynamically on the map14_.

Google maps also offers driving instructions and ways for local businesses to get themselves on the map. It also features satellite photos and a mixed view for maps that enables you to have the navigability of a standard map with the photorealism of a satellite photo. It is also possible to search for businesses in the area that you are looking at and so on15_.

3.6 State of the art in ubiquitous computing

The following sections will present a small overview describing the state of the art so far within the area of ubiquitous computing incorporating positioning techniques and web technologies.

3.6.1 Location based learning

Location based learning is an implementation of the ubiquitous computing in a teacher/student environment and is a new way for students to get to learn new things in a new and exciting way. A good example of this is the savanna project developed by NESTA Futurelab, Hewlett-Packard Laboratories, The BBC and the Universities of Nottingham and Bristol, which let’s the students actually live and interact with a savanna as lions. This is done on the field via PDAs with wireless connections and a “command central” where the teachers can follow the pupil’s movements and actions on the field.

The teachers can set the students' devices to different contexts such as food hunting, water drinking and so on so that students interact with the same field differently depending on what they are supposed to do. This contextual information is a great pillar for ubiquitous computing and the uses for this are many.

Interviews with students from the savanna project have shown that sound helps to improve the experience greatly. There are various other examples of projects like this already implemented and that proves that ubiquitous computing is very much a reality today (Benford, 2005).

The main goal of the study was to investigate the value of the game, described in the case framework, created from a pedagogical view – thus the application and data repository was tailored for the purpose of the game. Furthermore, content created was suited to fit the scope of the

(28)

game. The application running in the PDAs was programmed to behave in a certain manner upon interaction, which let the back-end be structurally static.

3.6.2 Contextual navigation

Ubiquitous computing allows the possibility of capturing contextual information characterized by the activity performed in combination with the current object carrying out the action (Hoppe et al., 2006). In order to describe, integrate and retrieve information, researchers at the University of Duisburg-Essen and Spanish Open University introduced a learning object repository (LOR). The LOR is made up of an ontology consisting of learning objects, actions, goals and complementary. As these four concepts contain the learning objects’ contextual information and references between resources, it is possible to create navigation based on group context and/or semantic navigations. For instance: two learning objects could be semantically related if authored by the same person although they differ in subject – depending on the context of the user browsing the system. While context-aware, there are no indications of support for geographical location-based decision making within their definition of context.

3.6.3 Media Creation System for Mobile Images

Significant research within the mobile communication area has been conducted at Berkley University. One of their recent efforts includes a content metadata creation process for images taken with a mobile phone, where the intention is to automate the entire process from creating image content and contextual metadata to have it physically stored and categorized in a repository (Sarvas et al., 2004). As proof-of-concept a prototype was constructed, which created semantic metadata “on the fly” as images were captured. Location-based metadata was included in the prototype as contextual metadata, but the authors states that capturing these data is a problem as most cellular phones do not support GPS. The data repository structure for handling the captured metadata is structured similar to ontologies, with a facet structure. In this way the different images can be semantically related to each other via the category it belongs to. In the Berkley paper the different facets were Person, Location, Object and Activity. For example: the Location facet for a building in Växjö could be Location > Geographical > Sweden > Småland > Kronoberg > Växjö. By creating relations between the different facet structures, semantic relations as well as image-specific metadata can be derived from the repository.

(29)

metadata repository seems to be made to fit the needs of their work without a deeper explanation of the value of the repository besides the description of the images. Furthermore, the Berkley initiative focuses solely on images, leaving other possible content, such as audio and text, out of their scope.

3.6.4 uLearn

Another initiative within the frame of ubiquitous computing supporting learning was conducted at Lancaster university. The project called “uLearn” aimed to investigate whether ubiquitous computing could contribute in the education of young children (Mitchell & Race, 2005). Their study was conducted by photographing bar codes next to pictures representing different animals with cellular phones. When the bar code was photographed, a request for more information about the picture belonging to the bar code was sent to a server. The server processed the request and returned a HTTP response to the cellular phone, which presented more information about the picture photographed in the default cellular phone web-browser.

3.6.5 Context aware systems

The complex use of the new generation of mobile devices such as mobile phones, combined with their relatively limited ability to actually present content on their small screens and limited technology support, forces the content creators to tailor their content for different devices. This indicates that understanding in which context the user is in would be highly beneficial as the user would want one piece of information presented in one way at his home computer and another one on his mobile phone.

In recent research efforts described by Beale and Lonsdale (2004), they discussed a model that describes context as a dynamic process with historic dependencies; this is to support context constructed trough the learner's interactions with the items available and the surrounding environment at the time. They also suggested a software architecture to evaluate the effectiveness of their ideas, using technologies such as XML.

Their metadata structure is a strictly hierarchical one with a root element that is the context object and then branch of into different sub-parts of the root element (Beale & Lonsdale, 2004).

3.6.6 The context project

The context project is a collaborative effort with the Department of Computer Science, the HIIT Basic Research Unit, both from the university of Helsinki and is funded by the Academy of Finland.

(30)

allowing you to “tag” pictures with different data manually via an interface directly in the phone. This project has also resulted in an application called Merkitys which basically is the ContextMedia program customized for use with flickr.com which is a picture database that supports these tags16_.

The things they want to achieve with linking this to flickr is to link mobile phones with an already popular place to upload photos in a contextual manner. You are also able to send GPS data and similar data with your pictures if you have a GPS device available and plugged into your phone in some way.

This differs it from the efforts in California as it is released under the GPL and is already a publicly available product which anyone can keep on develop as it is open source.

3.7 Putting it all together

After this brief overview within the area of ubiquitous computing, along with positioning and web technologies that can be used support its cause, the time has come to explain why this is important for our work. As previously stated, the idea of this thesis aroused after discussing the AMULETS-project with the CeLeKT-group. As this project focuses on mobile and ubiquitous technologies for outdoor activities, it was necessary to explain the topic of ubiquitous computing. Context and metadata has very central roles when automating devices – thus it was natural to begin these sections. The next section, which handled position technologies, was very relevant for this thesis as the AMULETS-project uses both GPS and Semacode for determining the whereabouts of persons and devices. The section of web technologies was just as important, as these technologies are heavily used when creating and presenting content that is created in the case within the AMULETS-project. Furthermore, we needed some insight in the various types of content that can be created within the scope of the case. Section 3.6, displaying the current state of the art within ubiquitous computing combining positioning technologies and web technologies, was necessary in order to explain where our contribution is placed within the area of ubiquitous computing and to clarify in what way we aim to differ from work already conducted.

3.7.1 Our contribution

When looking at the state of the art within ubiquitous computing (section 3.6), we can tell that much has already been conducted within the area of interest which we can use as inspiration for our upcoming work. We think that our approach differs from those efforts as there seems to be that a lot of focus on either displaying benefits of technology used or displaying the benefits of various metadata structures. Our approach aims to display the benefits we can gain by extending collectible metadata

(31)

generated by cutting-edge technology with metadata describing the context or setting in which the content was created. An overview of our thoughts can be viewed in figure 5 below.

Neither the Savannah-approach or the Media Creation System example seems to have set their focus in this manner when their work was performed, and the uLearn project didn't focus on the creation of content. Additionally, the Media Creation System and The Context Project limited their experiments to photographs, while we intend to let the metadata structure be able to support any type of metadata. In relation to the particular focus our work, our efforts differs from those conducted by Beale & Lonsdale (2004) in connection to associated model and data structure. While their model is strictly hierarchical, our schema aims to allow many different types of relationships between content. Further on, we will try to make our foundation as generic as possible to support all kinds of pre-created metadata models. The back-end should be able to support the description of any context in accordance to what is to be described, by relating the context in which a user or group of users exists to content created by these. By adding a contextual dimension in our approach we expect to gain:

• the possibility of implementing automatic refinement of the selection of content presented to the

end user by comparing the end user context with the contextual metadata describing content.

• the possibility of comparing and selecting content sharing the same context.

• the possibility of comparing and selecting content based on similarities in their contexts.

• the possibility to increase the probability of content relevance for the end user when automating

content selection by taking proper advantage of both technical metadata and contextual metadata.

(32)

(33)

4. AMULETS - Case and scenarios

This chapter will introduce the case and the scenarios in which we will apply the scenario-based methodology. The case foundations section of this chapter contains a description of the boundaries, or setting, in which the scenarios are created. In the following section we talk a little about the technical aspects in the scenarios, which are extremely important for us as the limitations of the devices used determines what metadata that can be collected. After the foundations has been explained we carry on with the scenarios.

As previously stated, AMULETS-project aims to make use of mobile and ubiquitous technologies in learning activities. This learning activity involves an outdoor game to be conducted by children at a local school, where the goal of the game is to educate children about trees. The original material we received from the CeLeKT-group describing the case can be found in Appendix 3 (in Swedish). The rules and proceedings of the game will be described in the following section.

4.1 Case foundations

The structure of the game is divided in three different activities, the first part will include a lecture for the students that are going to take part in the experiment and will educate them in the subject and explain how to use the technology used in the experiment, if needed. The second part is conducting the actual game which will involve the students to take part in a game in the surroundings of the school. The third part will be a reflective part where everything gets analyzed and what might have gone wrong will be corrected.

4.1.1 Previsit

The idea behind the whole game is that students learn things in a better way if they have both theory and practical teaching in a subject. The plan here is that the education goes in three steps where the students first listen, then do something practical and finally examine what they have done. To their aid, the students have two agents called ”Veta” and ”Finn”, which are flash animations appearing on-screen on mobile multimedia devices handed out to the groups participating in the game. There is also a teacher available to the groups for further assistance. The groups will consist of 4 students each, and as there are 32 children in the class the study will involve 8 groups. As previously mentioned, each group will be equipped with mobile multimedia devices serving different purposes in the game.

(34)

the real world and experience the study. To their help they have the two agents as previously mentioned.

4.1.2 The agents

”Veta”

Veta (the Swedish word for knowledge) is a digital agent that provides the students with both theoretical and practical knowledge. It’s animated and is presented to the students when semacodes are photographed. This agent will in this case be represented by a Nokia 6630.

“Finn”

Finn (the Swedish word for find) is the other digital agent. Finn is going to educate the children in how to find their way in an environment with the help of a map and some other technical aids, such as a Pocket PC and a mobile phone. This agent will in this case be represented in a HP pocket PC with GPS support and semacodes.

4.1.3 The game

The students will try the experiment out in close proximity to the school and will face a few problems that needs to be solved and the outcome of this little competition between the groups is both dependent on their answers and the time in which they complete the course. The quick sketch below (figure 6) visually represents the main idea of AMULETS17_{. The numbers on the sketch marks where}

different tasks in the game are located in the area. The arrows drawn to the box labeled CCS (CCS is a content storage and distribution system further explained in section 4.2.3) means that there is an exchange of content when content is created by or sent to the participants of the game. The arrow pointing at the map representation refers to a map representation in the PDAs used by the participating groups. The arrow between the map representation box and the CCS states a possible use of the metadata gathered in the CCS, in this case a representation with Google Maps technology.

(35)

The map representations in the PDAs will be used by the groups to navigate throughout the game. Additionally, in order to determine the winner, the game has a point system in which all groups starts with 50 points and for each accomplished mission they get 10 extra points.

There will be semacodes placed out in the fields which the students use for reporting, mark completed missions and get help and tips from the system. But if the students use the semacodes to get help it will give a point penalty. The types of semacodes will be distinguished by different color codes.

In all missions there is a part where the students needs to produce a photo/video/audio file or answer a question to prove that they have visited the place and get their points for the mission. If they get everything right the group will finish the game with 100 points. However, if they take too much time they might still loose as their total time takes into account.

4.2 Technological foundations

In this section we will give a short presentation of the technical devices that are used in the scenarios and what techniques they support. We will also describe the CCS-system illustrated in figure 6.

4.2.1 Nokia 6630

The Nokia 6630 (figure 7) is what is commonly known as a 3G phone, meaning that it operates in the WCDMA network if possible and enables the phone to reach previously unmatched transfer speeds in the area of 384kbps when receiving data and 128kbps when sending data. The phone does also

(36)

support edge but as edge isn’t supported by any network in Sweden it is not really usable for us18_.

The 6630 is running the 60-series operating system which is a Symbian-based operating system developed by Nokia. The use of this operating system ensures that the phone is fairly easy to develop applications for, which is good for the AMULETS-project.

It also support a 1.3 mega pixel camera which is sufficient for what it is going to be used for, there is also a 176x208px color screen on the phone which is more or less required to be able to preview pictures taken by the phone. The phone also supports exchangeable memory cards which makes it possible to store large files such as movie clips directly on the phone. It is also quite comfortable in size and it weights 127 gram, which is a bit more than a regular phone but still pretty light for a smart phone.

It does also support browsing the web via a web browser and sending e-mails. This is combined with an ability to send SMS, MMS and view most common formats in movies and images which combined makes this phone an excellent device choice for the case.

4.2.2 HP 6515

The HP 6515 (figure 8) is a Personal Digital Assistant, PDA for short, which per definition is a small computer which either has a touch screen or a small keyboard connected to it and is intended to be used as a personal organizer, it has however evolved into much more areas of use19_.

18 _{http://www.nokia.com/6630/}_{, 05/22/2006}

19 _{http://www.hp.com/}_{, 05/22/2006}

Figure 7: Nokia 6630

(37)

The device is running the Windows Mobile 2003 – Phone edition operating system, has a slot for a Integrated Secure Digital (SDIO) to expand the memory and features a 3” TFT screen. It has a 1.3 mega pixel camera with a LED flash, supports Bluetooth, IrDa, GSM, GPRS, EDGE and supports a lot of the common codec’s for video, audio and images.

It does also support a GPS solution, which is the most important feature of all in this device for us, as the project relies heavily on data describing where people are when they take part in certain activities. All this makes this device sufficient enough for the case.

4.2.3 CCS

While the AMULETS-project will be used as the guideline for pursuing the purpose of our work, the outcome is intended to be the first draft to a component in something larger. This “something” is called the CCS. With the help of the upcoming scenarios, we will display the advantages extending the CCS-system with our contribution.

The CCS (Collect Convert Send) system was originally developed during the MUSIS-project (mentioned in section 1.1.1) with the intension of serving as a a middleware between Internet information resources and mobile devices (figure 9). However, CCS is not at all limited to the MUSIS-project and can be used in and enhanced for future projects as well. The collect, convert and send processes relies on a data structure (1) and a scheduler (2) to be able to send appropriate content tailored for a user or a group of users.

(38)

The “Repository with metadata” is where the CCS system stores all content that is collected by the system, along with its descriptive metadata. Figure 10 illustrates an example of a user creating content and sending it to the CCS system with a mobile device (1). The attached metadata is extracted (3) and structured in a repository (4) along with the content created (5). The content combined with its descriptive metadata can then be used in order to make automated selection and conversion of content sent to users or groups of users more accurate and sophisticated without demanding user interaction (Kurti et al., 2006).

As previously stated, the result of this thesis aims to contribute to the CCS, and by looking at the purpose of this thesis (section 1.2) we can determine that our upcoming work involves refining the “Repository with metadata” in the data structure in figure 9. Applied to the case within the AMULETS-project, our upcoming work will be to investigate how to capture the context existing when the children create content in the game, and to figure out how this can be structured in order to enhance the selection and conversion of content in the CCS for users or groups of users (figure 9). In section 3.7.1, we have already stated some assumptions of possible enhancements the contextual metadata are expected to contribute with. Before we begin modeling the schema, we will dedicate the following section to the scenarios belonging to the case.

Figure 9: The CCS architecture

Case study: Extending content metadata by appending user context