ANNOTATIONS AND HYPERTRAILS WITH SPREADCRUMBS

An Easy Way to Annotate, Refind and Share

Ricardo Kawase, Eelco Herder and Wolfgang Nejdl

L3S Research Center, Leibniz Universität Hannover, Appelstr. 4, 30167 Hannover, Germany

Keywords: Annotation, Trail, Social Media, Social network, Online Collaboration, User Interface, SpreadCrumbs.

Abstract: Annotations have been shown to be an important activity during reading, especially during “active reading”.

Annotations support understanding, interpretation, sensemaking and scannability. As valuable as in paper-

based contexts, digital online annotations provide several benefits for annotators and collaborators. To study

the impact of these benefits we present in this paper SpreadCrumbs, a straightforward Web annotation tool.

SpreadCrumbs offers simple annotation’s interactions and metaphors that support most of the users’

annotations needs in the digital environment by enhancing the web experience with “in-context” annotations

and providing a unique form of social navigation support with hypertrails. The results of our studies with

the tool show the importance of annotations, the empirical outperformance of “in-context” annotations over

other methods, and the outcome benefits of supporting social navigation.

1 INTRODUCTION

The World Wide Web is arguably the biggest source

of information nowadays. Whereas the exchange of

ideas on the Web was predominantly one-way, the

Web 2.0 now offers a new means of interactions and

has shifted more power and influence to users.

However, there are still a number of features missing

that are essential for supporting information

classification, retrieval, processing and

understanding.

Most of these issues have been reported already

during the early inception of the Web, mainly from

the hypertext community (Halasz 1991) (Vitali

1999). In particular, frequently mentioned are: the

lack of typed or annotated links; the absence of

hypertrails; limited browser history mechanisms;

and the lack of support for annotations.

In order to bring these missing features into the

Web, a common workaround is to create

applications that enhance the Web usability, such as

search engines, tagging systems, annotation systems,

social networks and others. The competitive

character within the Web 2.0 has arguably led to a

more powerful reincarnation of the rich features that

once were part of the classic hypertext systems

(Millard 2006); albeit as a collection of diverse,

disconnected applications, interoperating on top of a

common Web platform. Surprisingly, despite the

prevalence of interactive applications and social

networking, thus far Web annotation systems

haven’t seen a significant take-up (Karger 2003).

Given the absence of any dominant mature

annotation system, it appears that there is still no

generally accepted, concrete method for

straightforward online annotation. This is surprising,

given the abundance of literature showing the

importance of annotations for comprehension and

their benefits for reading and writing proposes

(O’Hara 1997). Similar to the paper-based

environment, digital annotations are expected to be

useful for supporting comprehension and

interpretation (Marshall 1998). Moreover, comments

and references are known to stimulate associative

thinking, which can be even better reproduced

digitally, by what we call “hypertrails”. For this

reason, our research goal is to understand users’

annotation behaviors and identify the benefits and

drawbacks of online annotations and trails.

Based on insights gained from earlier work and

an analysis of the reasons that hampered wide-

spread adoption of earlier annotation systems, we

created SpreadCrumbs (Kawase 2009a).

SpreadCrumbs is an online annotation tool that

allows the users to place annotations within Web

resources, either for themselves or for other users. In

this paper we introduce the application, its main

functionalities and present a system evaluation.

Kawase R., Herder E. and Nejdl W.

ANNOTATIONS AND HYPERTRAILS WITH SPREADCRUMBS - An Easy Way to Annotate, Reﬁnd and Share.

DOI: 10.5220/0002778100050012

In Proceedings of the 6th International Conference on Web Information Systems and Technology (WEBIST 2010), page

ISBN: 978-989-674-025-2

The rest of this paper is structured as follows: First,

in section 2, we discuss related works on the fields

of annotations and annotations systems followed by

the description of Spreadcrumbs. Later, in section 4

we present a concise summary of a set of

experiments and studies using our tool and the

respective results. We finally draw our conclusions

in section 5.

2 RELATED WORK

2.1 Paper Annotations

We adopt the definition of annotations as set forth

by MacMullen (MacMullen 2005) and Marshall

(Marshall 1997) – as any additional content that is

directly attached to a resource and that adds some

implicit or explicit information in many different

forms. Annotations may serve different purposes,

such as: signalling a foreshadow, aiding memory

and interpretation or triggering reflection.

Additionally, annotations may occur in many

different forms; for example: by highlighting,

encircling or underlining text, we emphasize the

importance of a certain part of the document; a

strikethrough indicates that something is wrong,

misplaced or not relevant; arrows signal relations

between two or more elements.

Interacting with a document is known to

stimulate critical thinking and reflection, a process

that can be called ‘active reading’ (Adler 1972),

which is in contrast to passive consumption of text.

In particular, text in the margin of a document may

support a better understanding of the topic during

later reading.

In (Millard 2006), the authors draw a comparison

between the early Hypertext pioneers visions and the

present-day Web applications, commonly known as

Web 2.0. The results of their analysis show that most

of these systems support both private and public

annotations and provide support for collaboration.

Even though these features are identical with the

first ideas of the Hypertext, the annotations are

limited, because they reside exclusively bound to

individual Web 2.0 services providers and they are

not “in-context” – More specifically, they are not

visualized together and associated with the

annotated content (the topic of interest), which the

benefits will be exposed later.

2.2 Digital Annotation Systems

The Fluid Annotations projects (Zellweger 2002)

introduce an online annotation system that supports

in-context annotation in an extension of the open

hypermedia Arakne Environment (Bouvin 1999).

Their studies focused on evaluations and the

presentation of the annotations in terms of visual

cues, interactions, accommodation and animated

transactions. Their main approach to in-context

notes uses between-lines annotations. Their

evaluations give valuable insights into the usability

and manipulation of annotations. Nevertheless, we

believe disrupting the original layout of the

annotated content may be more confusing and

disruptive than beneficial.

Another annotation system is MADCOW

(Bottoni 2004) (Bottoni 2006) a digital annotation

system organized in a client-server architecture,

where the client is a plug-in for a standard web

browser allowing users to annotate Web resources.

Although MADCOW supports different

representations for annotations, previous work

comparing paper-based and digital annotations

(Kawase 2009b) suggests that paper-based

annotations should not be mimicked by similar

representations but by providing the means to

achieve the same goals. In addition, the placeholders

of the annotations are inserted between the HTML

content which can be disruptive, distractive and may

lead to the problem of orphan annotations. Finally,

usage complexity will impact the dissemination of

any new technology, and in particular, will always

be an obstacle for the non engaged users. The

annotation interface in their work has not been

evaluated.

A more full-fledged annotation tool is Diigo

Using the Diigo toolbar, users can highlight text or

attach 'inline sticky notes' to Web pages. Despite the

wealth of features, Diigo cannot boast a big user

population. According to online user comments, this

is due to both usability issues and the fact that all

annotations are public by default. We understand

that sharing annotations is one of the main possible

advantages of digital annotations systems; however

in light of Diigo, we believe that a ‘shared’

annotation must not be mistaken for a ‘public’ one.

The benefits of reliable collaborators are not fully

applicable in the ‘public’ scenario; we elaborate,

further on this point in sub-section 2.3.

http://www.diigo.com/

WEBIST 2010 - 6th International Conference on Web Information Systems and Technologies

In summary, there are numerous and similar

annotations systems - most of them are discontinued

works which have neither developed further nor

been presented in further studies.

2.3 Social Navigation

Social navigation support (SNS) describes

techniques for guiding users through specific chosen

resources (Brusilovsky 2001). In AnnotatEd (Farzan

2006) the authors introduce two types of SNS:

traffic-based and annotation-based. Our model more

is related to the annotation-based style, in that every

annotated page becomes a step in a trail.

Annotation-based social navigation support has

been shown to be more proficient and reliable than

traditional footprint-based social navigation support

(Farzan 2005). When the annotated resource reflects

the interest of the annotator, it appends more value

to the SNS. Annotation based SNS assists users in

gathering information by making it easier to re-

access the information and by showing the collective

wisdom of the collaborators.

Allowing users to “attach” their personal insights

to a resource increases the reliability of annotation-

based navigation support. Previous study of

annotation-based SNS shows that users are

particularly interested in being informed about

resources annotated by others. Annotated resources

are significantly more likely to be visited by users,

specifically after being annotated (Farzan 2005).

3 SPREADCRUMBS

SpreadCrumbs is an in-context Web annotation

system, which has been implemented as an

extension of the Mozilla Firefox Web browser. The

underlying assumption of SpreadCrumbs is that

users can annotate Web resources with keywords or

sentences and create hypertrails through a set of

annotations. These annotations can not only be used

for one’s own reference, but can also be shared

within a social network. The design of

SpreadCrumbs has deliberately been kept

minimalistic. Following the approaches seen in

related work, we chose the basic visual metaphor for

the annotations: Post-it notes.

The Post-it representation has an optimized

approach to simulate the most common paper based

annotations forms namely underlining, highlighting

and notation in margins. The idea is not to mimic

different representations but to provide a way to

achieve the same goals: signalling for future

attention, comprehension and summarization. In

addition post-it notes are extremely efficient as “in-

context” landmarks which are the main purpose of

the research.

Furthermore, by bringing the annotation

behaviour to the digital online environment we also

add valuable features that are not applicable in the

paper-based scenarios. The most prominent are the

re-finding and the social sharing possibilities. The

content of an annotation is easily searchable within

the tool and shareable with other users.

3.1 The Browser Add-on

The SpreadCrumbs Browser add-on is a Javascript

implementation based on AJAX principles. We used

the AJAX and Javascript library from Yahoo, The

Yahoo! User Interface Library (YUI). The library

provides functionalities for drag & drop and other

manipulations used in SpreadCrumbs. A simple

client server architecture stores all the data on the

server providing the user the possibility to access her

data anytime from any computer where the client

application is installed.

Once the client add-on is installed to the browser

the user can access the sidebar. Through the sidebar

the users have access to straightforward ordinary

actions like creating account, profile management,

access to a contact managing webpage and a tabbed

annotation-browser-window. From the right-click

context menu an option is available to annotate the

page, the same as from a small annotation button

near the address bar.

3.2 Networking

As a non-mandatory step, new users may add their

social network contacts to become collaborators in

SpreadCrumbs. From the sidebar the users have

access to the ‘contact manager’ webpage, from

which they can import their contacts from their

Facebook Network using Facebook Connect

technology. Once the contacts are imported they

become part of the user’s SpreadCrumbs network

and the user is able to share annotations with her

contacts. If at some point these contacts join

SpreadCrumbs and grant permission to Facebook

Connect; their accounts will be synchronised and all

the annotations previously shared by some other user

will be retrieved.

ANNOTATIONS AND HYPERTRAILS WITH SPREADCRUMBS - An Easy Way to Annotate, Refind and Share

3.3 Annotating

Annotations (which we will refer to as ‘crumbs’) are

added via the right-click context menu by the option

“Add Crumb”, which results in the opening of a

pop-up window that contains three fields: the

receivers of the annotations, a topic and the content.

By default, annotations are private. An auto-

completion drop-box helps the user in adding

receivers from her contact list.

Once the annotation is created, a post-it note

appears in the screen, originally on the clicked spot

but easily relocated by drag and drop (Figure 1).

When any of the involved users in the annotation

accesses the annotated website, post-it note will be

displayed. Additionally, if the user keeps her

connection to Facebook through SpreadCrumbs, the

receivers of the annotation will get a notification on

Facebook and a notifying e-mail about the new

annotation.

Figure 1: Conference page annotated with SpreadCrumbs.

3.4 Reacting

Each annotation is an entity in a thread (a crumb in a

trail) and diverse actions can be taken over it. When

visualizing an annotation, any of the involved users

has the ability to interact with it: moving it around,

closing it, following trails and replying.

3.4.1 Connect and Disconnect

Each user has her individual status in the context of

one annotation. The status “Connected” is the

normal status to visualize the annotations;

“Disconnected” means that she will not visualize the

annotation anymore once she comes back to the

website; and “Stand by” means that she will not

visualize the annotation again until some

modification has occurred in the annotation thread.

3.4.2 Replying

The reply link on an annotation brings up the same

window pop-up as adding an annotation offering to

the user just the content field to be filled. Once

confirmed, the reply is attached to the first post-it

note and the same notifications actions are triggered.

Any user involved in the annotation is able to add a

reply to the running thread, which is visible to all

participants. This action simulates a micro in-context

forum on each annotated web page.

3.4.3 Following Trails (SNS)

What makes SpreadCrumbs unique is that the

annotated pages are not simply a loose collection,

but the resources become interconnected. Each

annotation is associated with links that can be

followed from the crumb: the user trail and the topic

trail. Near the name of each user who annotated the

page and near the topic text there are two small

linked arrows indicating the path to the previous and

to next annotation in the hypertrail. Following the

previous/next link next to the name of a user will

redirect the current user to the next/previous

annotated page where both users share another

annotation.

Following the topic trail will lead the user to web

pages on which the user has annotations with the

same topic description. A simple illustrative

example: one user privately annotates five different

pages with the topic “Conference” adding specific

content for each annotation. Once it is done, each

conference page annotated has a link connecting to

each other. A temporal defined (and connected)

collection of web resources was created and at any

time the user is able to remove, edit or add new stop

points in this trail. The final output is a simulation of

the Memex idea where the resources are now

annotated and associated in accordance with the

user’s preferable organization.

Providing sharing capabilities of these trails,

SpreadCrumbs grants Social Navigation Support in a

very concrete and defined manner. Differently from

others SNS systems, the resources are not only a

collection of links but they have a well-defined

temporal order, each resource becomes

interconnected and they hold in-context insights

from the annotation authors.

3.5 Browsing Annotations

The SpreadCrumbs’ sidebar contains a browser pane

with three different tabs that shows the three facets

WEBIST 2010 - 6th International Conference on Web Information Systems and Technologies

of the organizational dimensions of a trail: topics,

pages, people. Additionally, a small pane in the

bottom shows detailed information on the selected

trail.

The tab topics shows the trails grouped by topic

description. The user visualizes distinct items that

represent the different trail-topics she created. From

this pane, the user is able to access the annotated

page, edit the topic description and change her status

in the topic. By clicking or selecting one of the

topic-trails the bottom pane loads and displays all

the crumbs belonging to this trail assembled by

page. In this pane the user has the same possibilities

to directly access the annotated page, to edit the

crumb and to reply it.

The second tab, page, shows the trails grouped

by the resource annotated. The visualization has the

title extracted from the Webpage and the trail last

modified date as well. The user has the possibility to

edit the name of the page, if she wants to. It is

important to notice that although trails mainly

contain the same page title in this facet they will not

be grouped together since the grouping is based on

the URL location of the annotation. By clicking or

selecting one of the page-trails the bottom pane

loads and displays all the crumbs belonging to this

trail assembled by the different topics existing on the

selected page with same management capabilities.

Finally, the people tab shows items that represent

the trails from the user’s contacts. The item

visualization shows the name of the contact and her

last activity on the trail. It also indicates whether the

contact is already connected to SpreadCrumbs’

network or not (due to the fact that is possible to

share annotations to imported contacts that are not

subscribed to SpreadCrumbs). By clicking or

selecting one of the people-trails, the bottom pane

works in the way as the topics tab previously

described.

4 EVALUATION AND STUDIES

To evaluate the usability and performance of

SpreadCrumbs, we ran a series of laboratory

experiments and processed the usage logs. The aim

of our experiment was threefold: 1) more fully

understand the desired annotation features needed on

the web, 2) examine the possible benefits of

annotations over bookmarks, and 3) evaluate social

navigation support in an arbitrary scenario. In this

section we will describe the experiments and

significant results.

4.1 Understanding Annotations

In order to better understand the real use of

annotations and Web annotations we conducted a

field-study examining the paper-based annotations

of 22 PhDs students and pos-Docs in their own work

environment (Kawase 2009b). For each participant,

we looked at the last 3 research papers or articles

that they have printed and read. In total we have

collected 66 articles, covering a total of 591 pages of

text. We found 1778 annotations and an average of

3.08 annotations per page. The Table 1 below shows

the average of each type of annotation based on

Marshall’s proposed classification (Marshall 1997)

by forms and functions.

Table 1: Collected annotations classified by type.

Annotation types

Highlighting/Mark sections headings 153 8.6%

Highlighting/Mark text 1297 73%

Problem solving 2 0.1%

General notes (Notes in the margins) 326 18.3%

Although most of the annotations consist of

highlighting activities, we identified in our previous

study that it does not imply that mimicking this

feature is the most appropriate approach to be

followed. We identified that paper-based highlights

are used for signalling and attributing different

levels of importance and to help memorization

during the reading activity. However, digital

highlight is usually a non-persistent activity to help

focusing on the text and re-finding – users highlight

the text with the mouse cursor while reading.

Excessive amounts of digital highlighting turns out

to be more distractive than helpful. The conclusion

of this work led us to the consideration that

annotation systems should emphasize re-finding,

visual overviews, grouping, sharing and

collaborating rather than to try and mimic the ‘old-

fashioned’ paper-based annotation.

4.2 Annotations vs. Bookmarks

For the comparison of Annotation and Bookmarks

we had a pool of participants consisting of 24 males

and 10 females, with an average age of 28. Our

participants were randomly and equally split into

two groups: the first group created annotations using

the Delicious social bookmarking service, the

second group made use of SpreadCrumbs.

ANNOTATIONS AND HYPERTRAILS WITH SPREADCRUMBS - An Easy Way to Annotate, Refind and Share

After a short introduction to the basic features of the

tool (either SpreadCrumbs or Delicious), each

individual session was conducted. We asked the

participant to find answers for ten random questions.

All questions were specific information-finding

tasks that could be solved by a brief internet search

with any popular search engine. We ensured that the

questions were sufficiently obscure, to minimize the

chance of participants knowing the answers

themselves.

Five months after the initial round of the studies,

the participants were invited to participate again.

This time, their task was to relocate the answers that

they had previously found during the first task. The

long time interval ensured that the participants

remembered neither the answers they had provided

nor the resources they had used to find the answers.

In total, 30 out of the initial 34 participants were

involved in this phase of our study (21 males and 9

females, average age 28 years).

The participants were divided into three

equivalent groups of 10 people, each one

corresponding to a specific refinding methodology

and corresponding tool. As a base line, the first

group used a search engine in their efforts to carry

out their tasks (in other words, they had to search

again for the same information). The second group

used bookmarks to refind the information. This

group consisted of those subjects that used Delicious

in the previous session and had the URLs of the

visited resources at their disposal. The third group

consisted of the SpreadCrumbs users. The members

of this group had the in-context annotations at their

disposal.

We ensured that all participants accomplished all

of their tasks under the same conditions and that

their performance is compared on an equal basis.

After the appropriate Web resource was found, thus

completing the ‘searching stage’, the participant had

to locate the answer in the page and highlight it

using the mouse – the browsing stage. There were

no instructions or restrictions on how to proceed at

this stage: the participants were allowed to perform

this task the way they would in a non- controlled

environment. Upon completion of all tasks, the

subjects were asked to fill out two questionnaires,

one regarding the information refinding experience

and another one investigating their opinion on the

tool they used. The necessary data for estimating and

evaluating the average and overall browsing time per

individual were collected using screen capture and

data-logging software that recorded all participants’

actions.

From this refinding task we collected a total of 297

successful activities, evenly distributed across the

conditions. With an average mean of 21 seconds, the

annotation group was significantly faster than the

bookmarking group (38 seconds; t(98)=3.88, p<0.01,

r=.36) and the search engine group (46 seconds;

t(98)=4.07, p<0.01, r=.38) plotted in figures 2 and 3

. The differences between the two latter groups

turned out to be non-significant.

Figure 2: Distribution of refinding tasks by time.

Figure 3: Average refinding time per participant in

ascending order.

We have seen that current digital annotation systems

mainly address the goals of future refinding and

sharing – which makes them very similar to social

bookmarking systems.

A full description of the entire experiment is

beyond the scope of this paper. As an ongoing

activity, we will detail the usability analysis and

present design issues. For example, we consider how

annotations diminish wasted time on refinding tasks

by providing landmarks and improving scannability.

We also intend to present work with reduce the

usage of browser's find functionality (CTRL+F).

Our experiments to date are promising and insightful

and we have identified significant benefits and

crucial need for annotations. Apart from the

WEBIST 2010 - 6th International Conference on Web Information Systems and Technologies

cognitive support for understanding and

interpretation while reading, these annotations

enhance scannability upon later reading,

outperforming bookmarks for refinding tasks.

4.3 Shared Trails and Annotations

To evaluate the usability and benefits of annotations

we asked the same 34 participants from the previous

study to play a role in a scenario on collaborative

decision making. The participants were asked to

plan a trip to London, by reviewing the options, as

collected by their ‘partner’ (the experimenter). Via

either SpreadCrumbs or Del.icio.us, the participant

received a number of annotations/bookmarks on

suitable hotels, restaurants, museums and musicals

in London. The participants evaluated the given

options – by visiting the bookmarked sites and/or by

reading the annotations – and finally decided for one

option in each category. After having finished both

tasks, the participants were asked to fill out a short

usability questionnaire and to evaluate the tools.

In this study, 50% of the users who received the

suggestions from their ‘partner’ via Delicious did

not read or even did not notice the additional

comments on each bookmark, which were displayed

just below the page title and the URL. One

participant explicitly told us that she noticed them

only in the middle of the task. Another participant

said that she noticed the comments, but did not read

all of them because she thought they were irrelevant.

By contrast, all the participants who received the

suggestions via SpreadCrumbs did notice and read

the comments, which were displayed as post-it

notes. They all accessed the bookmarked pages and

read the shared comments in the context. During the

interview after the task, some of them confirmed that

their choices were influenced by those comments.

The results show that if annotations are meant to

provide additional information and to influence the

receiver’s opinion or choices, they should be

presented as such, in the context. A text snippet

below the title, as provided by many social

bookmarking sites, is clearly not sufficient to catch

the receiver’s attention and may be overlooked

during a collaborative knowledge building process.

4.4 User Feedback

After completing the set of tasks, each participant

was asked to fill out a questionnaire, with the aim of

distilling opinions on the tool used as well as the

experiments in general. The answers were given by

selecting the appropriate value on a 7 point Likert

scale.

The user experience survey consisted of 13

questions, taken from established surveys on user

satisfaction, frustration and disorientation. The

Crombach a of 0,762 indicated a good reliability and

the results were grouped nicely into the three factors.

Without going too much into detail, the error bar

charts show that participants from the Bookmarking

and the Annotating group reported less frustration

than participants from the Search group. Further, the

participants from the Annotation group reported a

marginally significant lower value of difficulties in

finding the right information (see Figure 4)

Figure 4: Error Bars for survey questions on frustration

(left) and difficulty in finding information (right).

Whereas most other questions did not result in

significant differences in answers, the overall trend

indicated positive effect of bookmarking - and of

annotation in particular - on the subjective user

experience.

It is also worth mentioning that five participants

of the annotation group marked the same page, a

page that had been changed during the time interval

between the first and the second session of the study.

As a result, the annotations they had posted were

misplaced in all the five cases, which caused a slight

delay in the refinding task. Two of them suggested a

more intuitive way of attaching annotations that

involved arrows. Even though this way could well

solve the issue of misplaced annotations, it will still

be of no help for orphaned ones, which is in the

cases where the annotated information has been

completely removed. This issue is actually

considered as one of the most complicated and

challenging problems of the in-context annotation

approach (Cockburn 2001) (Wang 2005).

5 CONCLUSIONS

In this paper, we presented the SpreadCrumbs Web

annotation tool and demonstrate how it is able to

overcome the limitations of previously existing

annotation. In SpreadCrumbs, users can place Post-

it-like notes at any location of a Web page. From our

user studies and a literature survey we identified that

users’ needs for making annotations in the Web

ANNOTATIONS AND HYPERTRAILS WITH SPREADCRUMBS - An Easy Way to Annotate, Refind and Share

environment do not differ significantly from their

needs in the paper environment (Fu 2005). In

addition, Spreadcrumbs supports different user tasks,

not only private annotations, but also personal

reminders, refinding enhancer, and social

bookmarking/annotation with a unique form of

supporting social navigation and collaboration.

We also presented empirical results that show the

important role of annotations in the digital

environment, the outperformance of in-context

annotations over bookmarks in terms of supporting

information refinding, the analysis and the impact of

in-context annotations on social and collaborative

scenarios and finally the usability and users’ opinion

feedback.

Although we have seen the importance and

benefits of annotations, no annotation system is

widely adopted. This implies that there are still

several issues to be studied and solved. The main

challenge for annotation systems is on the user

interface level. It is necessary to balance the classic

tension between full-fledged features and ease of

use. Particular attention should be paid to the

question to what extent annotation systems should

provide and emphasize social bookmarking features.

Addressing issues such as this is intended in our

future work.

REFERENCES

Adler, M.J. and van Doren, C. (1972) How to Read a

Book. Simon and Schuster, New York, NY.

Bottoni, P., Civica, R., Levialdi, S., Orso, L., Panizzi, E.,

and Trinchese, R. (2004) MADCOW: a multimedia

digital annotation system. In Proceedings of the

Working Conference on Advanced Visual interfaces

(Gallipoli, Italy, May 25 - 28, 2004). AVI '04.

Bottoni, P., Levialdi, S., Labella, A., Panizzi, E.,

Trinchese, R., and Gigli, L. (2006) MADCOW: a

visual interface for annotating web pages. In

Proceedings of the Working Conference on Advanced

Visual interfaces (Venezia, Italy, May 23 - 26, 2006).

AVI '06.

Bouvin, N. O. (1999). Unifying strategies for Web

augmentation.Proceedings of ACM Hypertext’99, p

91-100, 1999.

Brusilovsky, P. (2001) Adaptive hypermedia. User

Modeling and User Adapted Interaction 11 (1/2), 87-

110. Claypool, M., Le, P., Wased, M., and Brown, D.

(2002) Implicit interest indicators. In: Proceedings of

6th International Conference on Intelligent User

Interfaces, pp. 33-40.

Cockburn, A. and B. McKenzie. (2001) What do Web

users do? An empirical analysis of Web use. Int. J. of

Human-Computer Studies, 54(6): 903-922, 2001.

Farzan, R. and Brusilovsky, P. (2005). Social navigation

support through annotation-based group modeling. In:

Proceedings of 10th International User Modeling

Conference, pp. 463--472.

Farzan, R., Brusilovsky, P. (2006). AnnotatEd: A Social

Navigation and Annotation Service for Web-based

Educational Resources. In: Proc. of E-Learn 2006,

Honolulu, HI, USA, October 13-17, 2006, AACE

2794—2802

Halasz, F. G. (1991). "Seven issues": Revisited. Closing

plenary address. In Proceedings of ACM Hypertext '91

Conference, San Antonio, Texas, December 18, 1991

Karger, D., Katz, B., Lin, J. & Quann, D. (2003), Stickey

notes for the semantic web, in ‘Proc. Intelligent User

Interfaces 2003’.

Kawase R., and Nejdl W. (2009a) A Straightforward

Approach for Online Annotations: SpreadCrumbs -

Enhancing and Simplifying Online Collaboration.

WEBIST, page 407-410. INSTICC Press, (2009)

Kawase, R., Herder, E. and Nejdl W.. (2009b) A

Comparison of Paper-Based and Online Annotations

in the Workplace. Learning in the Synergy of Multiple

Disciplines, Proceedings of the EC-TEL 2009, volume

5794 of Lecture Notes in Computer Science,

Berlin/Heidelberg, Springer, October 2009.

MacMullen, W. J. (2005). Annotation as Process, Thing,

and Knowledge: Multi-domain studies of structured

data annotation. SILS Technical Report TR-2005-02.

UNC School of Information and Library Science.

Marshall, C. (1997). Annotation: From Paper Books to the

Digital Library. Proceedings of the 1997 ACM

International Conference on Digital Libraries (DL 97).

Marshall, C. (1998). Toward an Ecology of Hypertext

Annotation. Proceedings of the Ninth ACM

Conference on Hypertext and Hypermedia (Hypertext

98).

Millard, D. E. and Ross, M. (2006). Web 2.0: hypertext by

any other name?. In Proceedings of the Seventeenth

Conference on Hypertext and Hypermedia (Odense,

Denmark, August 22 - 25, 2006). HYPERTEXT '06.

O’Hara, K., Sellen, A. (1997). A Comparison of Reading

Paper and On-Line Documents. Proceedings of the

1997 ACM Conference on Human Factors in

Computing Systems (CHI 97).

Vitali, F. & Bieber, M. (1999), ‘Hypermedia on the web:

What will it take?’,ACM Computing Surveys 31 (4es),

Article No. 31

Wang, S. 2005 Annotation Persistence Over Dynamic

Documents. Doctoral Thesis. Massachusetts Institute

of Technology.

X. Fu, T. Ciszek, G. M. Marchionini, and P. Solomon

(2005). "Annotating the web: An exploratory study of

web users needs for personal annotation tools," in The

68th Annual Meeting of the American Society for

Information Science & Technology (ASIS&T),

Charlotte, NC, USA, 2005.

Zellweger, P., Mangen, A., Newman, P. (2002). Authoring

fluid narrative hypertexts using treetable

visualizations. Proceedings of ACM Hypertext 2002.

WEBIST 2010 - 6th International Conference on Web Information Systems and Technologies