mingana.tex Click here to view linked References 1 2 3 4 5 6 7 8 9 Touching Annotations: A Visual Metaphor for 10 11 Navigation of Annotation in Digital Documents 12 13 14 C.P.Bowers∗, C.Creed, B.R.Cowan, R. Beale 15 16 Human-Computer Interaction Centre, School of Computer Science, University of 17 Birmingham, Birmingham, B15 2TT, UK 18 19 20 21 22 23 Abstract 24 25 Direct touch manipulation interactions with technology are now common- 26 place and significant interest is building around their use in the culture and 27 heritage domain. Such interactions can give people the opportunity to ex- 28 plore materials and artifacts in ways that would otherwise be unavailable. 29 30 These are often heavily annotated and can be linked to a large array of re- 31 lated digital content, thus enriching the experience for the user. Research has 32 addressed issues of how to present digital documents and their related anno- 33 34 tations but at present it is unclear what the optimal interaction approach to 35 navigating these annotations in a touch display context might be. 36 In this paper we investigate the role of two alternative approaches to 37 support the navigation of annotations in digitised documents in the context 38 39 of a touch interface. Through a control study we demonstrate that, whilst 40 the navigation paradigm displays a significant interaction with the type of 41 annotations task performed, there is no discernible advantage of using a 42 43 natural visual metaphor for annotation in this context. This suggests that 44 design of digital document annotation navigation tools should account for 45 the context and navigation tasks being considered. 46 47 Keywords: Document navigation, touch interaction, affordance, interaction 48 techniques, user study 49 50 51 52 ∗Corresponding author 53 Email addresses: [email protected] (C.P.Bowers), [email protected] 54 (C.Creed), [email protected] (B.R.Cowan), [email protected] 55 (R. Beale) 56 57 58 59 Preprint submitted to Elsevier June 2, 2013 60 61 62 63 64 65 1 2 3 4 5 6 7 8 9 1. Introduction 10 11 Arts and cultural organisations are increasingly making use of touch en- 12 13 abled interfaces to allow the general public to interact with precious and 14 rare artefacts. These interfaces have the potential to provide intuitive and 15 engaging experiences when interacting with digital proxies of artefacts or 16 17 manuscripts where the original physical version is inaccessible. This includes 18 situations where restrictions on the number of artefacts constrain scalability 19 and where interactions tend to hold a risk to the user or the artefact (e.g. 20 dangerous substances or rare/precious artefacts). 21 22 In particular, museums and libraries regularly scan ancient manuscripts 23 to make it possible for people to interact with digital versions of these docu- 24 ments (e.g. Turning the PagesTM). These interfaces can provide compelling 25 physical affordances allowing users to browse through documents, work col- 26 27 laboratively, and make comments or raise discussion about the material in 28 an intuitive and familiar manner (Geller, 2006; Terrenghi et al., 2007; Liesa- 29 putra and Witten, 2012). Cultural organisations are particularly interested 30 31 in collecting user generated content from the general public and visualising 32 this in a way that makes it easy to navigate and view the annotations of 33 others. 34 Annotations added to these documents are especially important as they 35 36 help the general public and scholars learn more about the history and con- 37 tent of the documents. For these historical documents, annotation plays a 38 critical role in translation, interpretation and comprehension of the content 39 40 of these documents since they are often written in ancient script that result 41 in difficult and, in some cases, multiple interpretations. Annotations can be 42 part of an ongoing dialog between multiple readers. As such annotations can 43 be gathered over the entire lifetime of a document and can include many 44 45 forms of content including initial elucidations by the original document au- 46 thor through to contemporary scholarly interpretations (Agosti et al., 2007). 47 The nature of historical documents tends to result in the generation of 48 large numbers of annotations that consist of significant amounts of cross- 49 50 referencing both to aspects of the document and to other annotations. In 51 order to gain a full appreciation of a documents content these numerous an- 52 notations must be presented in an intuitive and easily accessible way without 53 54 being overwhelming. 55 Traditionally annotations have consisted of notes hand-written directly 56 onto the original page, appended to the page in the form of a physically 57 58 59 2 60 61 62 63 64 65 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 Figure 1: Use of thumbnails to support navigation between pages 30 31 32 TM 33 attached note (e.g. Post-it notes) or held in a seperate document utilis- 34 ing some suitable form of cross-referencing (e.g. page & line number). For 35 digitised documents, whose origins predate many of the technical innova- 36 tions in interweaving digital knowledge content through hyper-referencing, 37 38 any enhancement must be appended to the original document. However, it 39 remains unclear what the optimal interaction approach is in this context and 40 in particular how users should navigate annotated documents on larger touch 41 42 interfaces. 43 To better support public access to these documents we investigate the 44 use of touch enabled interfaces for navigation of annotated documents. The 45 software developed to support these activities typically focuses on the re- 46 47 quirements of individual researchers working at a traditional desktop using a 48 monitor, keyboard and mouse where users see scanned document pages on a 49 screen and can navigate back and forth between pages by clicking on image 50 51 thumbnails of other pages (Robinson, 2010). Figure 1 demonstrates the use 52 of thumbnail images representing links to other document pages when view- 53 ing an historical text, whilst the current page of the text is displayed as a 54 larger image in the centre. 55 56 A number of studies have looked specifically at document navigation us- 57 58 59 3 60 61 62 63 64 65 1 2 3 4 5 6 7 8 9 ing a traditional interface setup (i.e. ones using a mouse and keyboard), but 10 11 less work has focused on whether these techniques are also effective on touch 12 interfaces where users have the ability to directly manipulate and interact 13 with the document and its annotations (Buchanan and Owen, 2008; Cock- 14 burn et al., 2006). The use of touch interfaces presents the opportunity to 15 16 utilise real world metaphors and affordances that can create a more familiar 17 interaction experience for users, yet it remains unclear whether this is an 18 improved approach over more traditional navigation techniques. 19 20 In addition, although some effort has been made to explore the impact of 21 touch interface designs on user performance metrics (Marshall et al., 2008; 22 Shaer et al., 2011; Chiu et al., 2011), much of the research still focuses on re- 23 porting high level design solutions and qualitative evaluations. These studies, 24 25 whilst providing valuable insights, do not readily identify the causal impact 26 of interaction design choices on usability related metrics. Those quantita- 27 tive studies that have been performed tend to focus on low-level interactions 28 29 and manipulations (Cockburn et al., 2012). Consequently, little work in this 30 area has been conducted to understand how the mode of interaction with 31 annotations contribute causally to the usability of such interactions. 32 The work presented aims to investigate the usability of differing inter- 33 34 action designs in navigating digital versions of manuscripts in relation to 35 their annotations. This paper explores how the mode of interaction with 36 annotations within these documents impacts on user efficiency and usability 37 judgement when navigating the content of a document using existing annota- 38 39 tions. We start by covering related research on digital document annotation 40 and then describe an empirical investigation comparing a real world interac- 41 tion approach against a more traditional scrolling and thumbnail approach. 42 43 We then provide a discussion of the significant statistical effects found along 44 with suggestions for future research. 45 46 47 2. Related Work 48 49 Supporting the creation and navigation of annotations has been a funda- 50 mental feature of the development of digital document systems. We define 51 the term annotation to mean any additional content that is added to the 52 53 original document content. Annotations can be used to extend and enhance 54 a document in a number of ways. They can be used to summarise/simplify 55 complex content, allow cross-referencing of content, support the capture of 56 57 58 59 4 60 61 62 63 64 65 1 2 3 4 5 6 7 8 9 alternative viewpoints, perspectives or discussions and better support search 10 11 (M¨uller and Maurer, 2011). 12 An annotation can be a mix of any type of media including text, images, 13 and videos. Most research on annotation in digitial documents has focused 14 on the presentation of annotations and how the relationship between the an- 15 16 notation and the document content is supported (Chen et al., 2008; Pearson 17 et al., 2009). Little focus has been placed on how to best support users in 18 navigating existing annotations in relation to the document content (Kim 19 20 et al., 2008). That is, how can users easily find document content that is of 21 interest using the annotations in an efficient manner. 22 Research has focused on the best approach for enabling users to add anno- 23 tations to documents (Brush et al., 2001) rather than how these can support 24 25 document navigation and how these impact usability related variables. For 26 example research by Pearson et al. (2011) studied how a direct manipula- 27 tion interface using lightweight digital annotation tools that reflected real 28 TM 29 world tools for physical documents (such as Post-it notes) support the act 30 of annotating a document. The main aim of such research was to assess the 31 nature of the annotation mechanism itself. The research presented in this 32 paper looks to assess how tools for digital annotation support usability of 33 34 document navigation, specifically in a touch interaction scenario. 35 Digital documents are increasingly being accessed, manipulated, and an- 36 notated digitally on touch devices such as smartphones, tablets, touch en- 37 abled kiosks and touch tables. These interfaces afford new opportunities 38 39 for working with annotated digital documents. In particular, touch interac- 40 tions allow designers to create interfaces that utilise real world metaphors 41 with which users are familiar such as being able to circle and highlight im- 42 43 portant areas of a document through digital ink (Bargeron and Moscovich, 44 2003; Golovchinsky and Denoue, 2002; Olsen et al., 2004) and also to repre- 45 sent annotations through coloured tabs attached to the side of the document 46 (Buchanan and Owen, 2008). 47 48 However, research that has been conducted into navigating digital docu- 49 ments can be useful in understanding the techniques that might be best suited 50 for navigating annotations in touch-based interfaces. A variety of different 51 approaches have been examined and tested including space filling thumbnails 52 53 (Cockburn et al., 2006), camera tilts (Guiard et al., 2006), multi-scale navi- 54 gation (Appert and Fekete, 2006), following figure references (Buchanan and 55 Owen, 2008), zoomable interfaces (Bederson, 2011), and sound cues (Eslam- 56 57 bolchilar and Murray-Smith, 2006). 58 59 5 60 61 62 63 64 65 1 2 3 4 5 6 7 8 9 The use of scrolling to navigate documents is utilised in many popular 10 11 applications (e.g. internet browsers, word processors, etc.) and as such has 12 received much attention. Several researchers have observed that standard 13 scrolling interfaces can have a negative impact on interaction by slowing 14 down user progress due to the additional effort required to operate them 15 16 (O’Hara and Sellen, 1997). Scrollbars can also have a detrimental impact 17 with regard to a user’s memory about the location of different elements in a 18 document. This could negate the potential of their spatial memory to assist 19 20 with navigation of a digital document (Cockburn et al., 2007). 21 A number of researchers have therefore examined how to enhance the 22 scrollbar. Examples include the use of visualisations in the scrollbar (Byrd, 23 1999), navigation techniques for e-book readers (Chen et al., 2008), using 24 25 techniques such as Speed Dependant Automatic Zooming (SDAZ) (Igarashi 26 and Hinckley, 2000), and Rapid Serial Visual Presentation (RSVP) to replace 27 scrolling with page flipping at high speeds (Sun and Guimbreti`ere, 2005). 28 29 Whilst positive benefits have been identified with scroll based approaches it 30 remains unclear how well suited they are to touch screen interactions. 31 It must be noted that the hardware used in these interactions must also be 32 considered. A number of related studies have also demonstrated the impact 33 34 that the form and function of interaction hardware has upon the appropri- 35 ateness of interaction design for navigation of documents. Jones et al. (1999) 36 demonstrate that the size of a display can impact on appropriateness of in- 37 teraction design with regard to scrolling and paging. Smith and schraefel 38 39 (2004) describe how navigation performance can be improved by designing 40 scrolling for touch interaction. However, no work to date has looked at nav- 41 igation of annotation or compared new interaction techniques such as these 42 43 with traditional interaction approaches on touch surfaces. 44 45 3. Research Aims 46 47 As discussed in the previous section interweaving digital knowledge con- 48 49 tent into existing digitised documents, in the form of digital annotation, is a 50 growing challenge for heritage and cultural organisations wishing to provide 51 access to such materials. Researchers have begun to address a number of 52 53 these issues, in particular with regard to suitable presentation mechanisms 54 for annotation content. At the same time researchers have been investigat- 55 ing suitable interaction mechanisms to support digital document navigation. 56 However, apart from Liesaputra and Witten (2012) examining how anno- 57 58 59 6 60 61 62 63 64 65 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Figure 2: Visual metaphor of annotation in the form of tabs 26 27 28 29 tation can promote efficient and effective navigation via key word search, 30 limited research has investigated how annotations impact upon navigation, 31 32 especially using a direct navigation paradigm. 33 This work undertakes a comparative study of two common annotation 34 presentation paradigms in the context of a touch enabled digital document 35 interface. The aim is to identify how the manner in which they are pre- 36 37 sented might impact upon the efficiency of document navigation via those 38 annotations. The first (Tab condition) will exemplify the use of real-world 39 metaphors and natural affordances to facilitate navigation (Pearson et al., 40 41 2011; Liesaputra and Witten, 2012). The second (Scrollbar condition) will 42 draw from traditional document navigation interfaces (Buchanan and Owen, 43 2008; Cockburn et al., 2006). 44 45 3.1. Tab condition 46 47 The Tab condition utilised a real-world metaphor of annotation within a 48 document by placing tabs along the fore-edge of the book (figure 2). This 49 acts as a visual reference point for the annotation with the aim of providing 50 51 a clear affordance in terms of navigating to an annotation. 52 The location of each tab on the fore-edge reflects the position of its rep- 53 resentative annotation with respect to its position within the book. Tabs are 54 55 arranged in order down the page, so that one tab does not occlude another. 56 In addition, tabs on the left fore-edge of the documents facing pages reflect 57 58 59 7 60 61 62 63 64 65 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Figure 3: Navigation design utilising a scrollable list of annotations 26 27 28 29 annotations that are prior to the currently selected annotation. Those to 30 the right represent subsequent annotations following the current annotation 31 within the document. 32 33 The book can be turned to the marked page by touching the tab, where- 34 upon the relevant page is displayed along with the annotation and its tie 35 indicating precisely what text it is attached to. 36 37 3.2. Scrollbar condition 38 39 The Scrollbar condition provides a navigation mechanism for the annota- 40 tion that is more akin to navigation within a traditional desktop interface. In 41 42 this case the tabs used in the previous condition are replaced with a scrollbar 43 (figure 3). 44 The scrollbar utilises a scrolling list containing an overview of the annota- 45 tion state of the document. The overview consists of a brief description of the 46 47 annotation and a thumbnail of the relevant document page with annotated 48 contents highlighted (figure 4). 49 The scrollbar is manipulated through the touch interface using vertical 50 51 swipe gestures which are aligned so that the content moves with the users 52 finger. If the user taps an annotation it opens the document at the location 53 of the relevant annotation. 54 From the previous section we identify two factors that might influence the 55 56 performance of annotation navigation: how the content of the annotation is 57 58 59 8 60 61 62 63 64 65 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 Figure 4: Details of overview in scrollable list 32 33 34 35 presented and how the relationship between the annotation and the docu- 36 ment content is supported. Two sets of navigation tasks are defined which 37 38 explicitly account for these two factors. Three tasks require the annotation 39 to be found based on the content of the document, such as annotations that 40 refer to particular images or text. The other three tasks focus on the loca- 41 tion of the annotation with respect to the contents of the document. The 42 43 tasks are detailed in section 4.4. The tasks form two fundemental types of 44 navigation and are used to undertake a comparitive task-based study. We 45 aim to test the following hypotheses: 46 47 48 Hypothesis 1. There will be a significant interaction effect between task 49 type and annotation navigation paradigm with respect to task efficiency. 50 51 Since the Scrollbar condition will provide more detail with respect to 52 53 the content itself it is expected that users would be able to more quickly 54 identify whether an annotation is relevant to the task if the task is related 55 to the content of the document rather than the location. Additionally we 56 hypothesise that users will be more efficient in the Tab condition where the 57 58 59 9 60 61 62 63 64 65 1 2 3 4 5 6 7 8 9 task is dependent upon location of annotations relative to each other or to 10 11 the document overall. 12 13 Hypothesis 2. Participants in the tab condition will rate the usability of 14 the tab condition significantly differently to those in the scrollbar condition. 15 16 17 The visual metaphor using tabs to indicate the location of annotations 18 more closely matches the affordances that a user might expect in the context 19 of annotation within a physical document. 20 21 22 4. Method 23 24 4.1. Participants 25 26 33 participants took part in the study and were recruited from the body of 27 staff and students at the University of Birmingham using an email request for 28 participation. This email was posted to all departments across the university 29 £ 30 to gather a wide range of participants. A 5 gift voucher was given to each 31 participant as an honorarium for participation. 32 33 4.2. Conditions 34 35 The experiment consisted of the two annotation navigation conditions 36 for the document as described in section 3. These conditions make up the 37 between participant variable Annotation Navigation in the analysis. 38 39 40 4.3. Materials 41 Although access to ancient documents in a culture and heritage context 42 formed the initial inspiration for the research, we wished to generalise our 43 44 findings more widely to the display of annotated documents using touchta- 45 bles. The material chosen for the evaluation was Alice’s Adventures in Won- 46 derland by Lewis Carroll. This document is freely available as a set of high 47 48 resolution page images. Annotations were generated by the authors to ac- 49 count for the experimental conditions. Due to the number of different ver- 50 sions of the book, each with different structures, it was felt that any famil- 51 iarity a user might have with the story would not confound measurements of 52 53 task performance made in this research. 54 The main document view is dominated by a centrally positioned image 55 of the content of the document displayed in the form of facing pages. This 56 accurately represents how the content of a physical book would appear to 57 58 59 10 60 61 62 63 64 65 1 2 3 4 5 6 7 8 9 No. Task type Task description 10 11 1 Locate the first annotated image of Alice. 12 2 Content Locate the first annotated image that includes 13 the rabbit. 14 3 Locate the third annotated image of Alice. 15 16 4 Find the first annotation about the mad hatter 17 Location (around the middle of the book). 18 5 Find the annotation about Alice eating a mush- 19 20 room (in the first third of the book). 21 6 Find the annotation about Alice in the jury box 22 (in the final third of the book). 23 24 Table 1: The annotation navigation tasks presented to the participants 25 26 27 28 a user, leading to a more natural metaphor for interaction (Pearson et al., 29 2011). 30 31 Navigation of the document content was constrained under experimental 32 condition to ensure that the only way for users to browse to the relevant 33 location in the document was by the explicit annotation navigation condition. 34 35 Therefore, no page turn gestures or buttons were provided. 36 37 4.4. Tasks 38 Two sets of navigation tasks are defined representing two fundamental 39 40 type of navigation. The first set of three tasks require the annotation to be 41 found based on the content of the document. The second set of three tasks 42 focuses on the relationship between the annotation and the location of the 43 44 relevent content within the document. 45 The 6 tasks presented to users are defined in table 1. In order to avoid 46 ordering effects the task orders were counterbalanced across the participants 47 using a balanced Latin Square approach. The tasks and related annotations 48 49 were pre-determined and formed two distinct forms of Task Type. 50 51 4.5. Dependent Variables 52 53 Efficiency was quantified using response times for the tasks, measured as 54 the time taken from the point at which the task was started by the user to 55 the time at which they completed the task. 56 57 58 59 11 60 61 62 63 64 65 1 2 3 4 5 6 7 8 9 Usability was measured using the Software Usability Scale (SUS) (Brooke, 10 11 1996). SUS has been used and researched in multiple studies (Bangor et al., 12 2008; Brooke, 1996; Lewis and Sauro, 2009) and consists of a 10-item scale 13 that provides a subjective assessment of usability. In previous studies it has 14 shown high internal reliability (Bangor et al., 2008). The questionnaire was 15 16 administered using a Likert Scale from Strongly Disagree (1) to Strongly 17 Agree (5). Negatively worded items were reverse scored and the scores for 18 each item were summed to gain an overall usability score. This was then 19 20 converted into a percentage. The scale therefore ranges from a minimum of 21 0 to a maximum of 100. 22 23 4.6. Procedure 24 25 The experiment was conducted in a small room (approx. 3m x 2.5m) 26 with controlled lighting to reduce reflection and eyestrain when interacting 27 with the touchtable. The interface was presented on a 22-inch multi-touch 28 29 capable display positioned horizontally at a height of 36-inches. 30 On arrival participants were provided with a short overview of what they 31 were required to do during the experiment. They were then asked to read 32 through each of the annotations in the interface and get a general sense of 33 34 the content contained within the annotations. This was an important initial 35 task as it enabled participants to become familiar with the touch interface 36 prior to undertaking the actual experiment. To avoid the interfering effects 37 of users recalling the order, or content, of annotation a total of 12 separate 38 39 annotations were applied to both text and illustrations within the document. 40 Once participants had been through each of the annotations they informed 41 the experimenter who then reset the interface and started the experiment. 42 43 At the start of the study participants were assigned to either the Tab or 44 Scrollbar condition. A task was then presented to the user that involved them 45 having to locate an annotation within the document. Once the participant 46 was ready to start the task they tapped a “Start Task” button and began 47 48 searching for the annotation. Once participants had started a task a “Fin- 49 ished Task” button was displayed at the top of the interface that they were 50 asked to tap, once they had located the appropriate annotation. They were 51 allowed to select and search through as many annotations as they wanted 52 53 when performing their search (i.e. they did not have to guess the correct 54 answer at the first attempt). When the “Finished Task” button was selected 55 the next task was immediately displayed. This process was then repeated for 56 57 the further tasks. 58 59 12 60 61 62 63 64 65 1 2 3 4 5 6 7 8 9 Annotation Navigation Task Type N Mean (secs) S.D. (secs) 10 Location 16 29.20 17.60 11 Tab 12 Content 16 34.75 17.49 13 Location 17 36.07 26.68 Scroll 14 Content 17 17.65 8.40 15 Location 33 32.74 22.66 16 Total 17 Content 33 25.94 15.97 18 19 Table 2: Mean task times by conditions 20 21 22 23 After participants had finished all of the tasks they were then asked to 24 complete the SUS questionnaire. Upon completion of the questionnaire they 25 £ 26 were debriefed and received their 5giftcard. 27 28 5. Results 29 30 5.1. Task Efficiency 31 32 The descriptive statistics for the sample in terms of amount of time taken 33 to complete the task (in seconds) are displayed in table 2. The effect of Task 34 35 Type (Content or Location Tasks - within participants) and Annotation Nav- 36 igation (Tab or Scroll - between participants) on the time taken to complete 37 the task (Efficiency) was analysed using a 2x2 mixed design ANOVA. There 38 was no significant main effect of Task Type [F (1, 31) = 2.68,p > 0.05] or 39 F , . ,p > . 40 Annotation Navigation [ (1 31) = 0 97 0 05] on the time participants 41 took to complete the tasks. 42 In support of our first hypothesis we observed a significant interaction 43 between Task Type and Annotation Navigation on the time taken to complete 44 45 the tasks [F (1, 31) = 9.29,p<0.01]. The interaction is presented graphically 46 in Figure 5. 47 In order to explore this interaction effect further a paired samples t- 48 49 tests were performed for each Annotation Navigation condition. In the Tab 50 condition there was no significant difference in efficiency between the Content 51 (M =34.75,S.D =17.49) and Location (M =29.20, S.D. =17.60) task 52 conditions [t(15) = 0.995,p>0.05]. This means our prediction that the Tab 53 54 condition better supported navigation tasks that relate to annotation and 55 their relative positions to other annotations, or the overall document, was 56 unsupported. 57 58 59 13 60 61 62 63 64 65 1 2 3 4 5 6 7 8

9 

10

¥

11 ¦

¡

12  

13 14

15 

 

16 





  17 

18 



19 



 

20   21  22

23   24 25

26 

27

£¤ £¥ £ ¥

¢ ¦§¨§ ©¡ ¦ ¡¢ ¦§¨§ 28 ¡¢ 29 30 31 Figure 5: Mean & Standard Error of task completion time for Annotation Navigation 32 conditions by Task Type. 33 34 35 36 There was however a significant difference between the Content (M = 37 17.65, S.D. =8.40) and Location (M =36.07, S.D. =26.68) task conditions 38 in the Scroll condition [t(16) = −3.33,p < 0.01]. Participants completed 39 40 the content tasks significantly faster than the location tasks when using the 41 Scroll condition. This lends support for our hypothesis that there will be a 42 significant difference in the Scroll conditions between the tasks. 43 It seems that the Scroll condition was felicitous for the content-based 44 45 tasks compared to the location based tasks, whereas there was no significant 46 impact of the Tab condition on the speed of task completion in the different 47 task conditions. 48 49 50 5.2. Subjective Usability 51 Reliability analysis highlights that the Software Usability Scale (SUS) 52 (Bangor et al., 2008) reached an acceptable level of reliability for a psycho- 53 α . 54 metric scale (Cronbach =074) (Kline, 2000) although this is lower than 55 alpha values reported in previous research (Lewis and Sauro, 2009). The de- 56 scriptive statistics for the sample in terms of the usability scores are displayed 57 58 59 14 60 61 62 63 64 65 1 2 3 4 5 6 7 8 9 Annotation Navigation N Mean (%) S.D. (%) 10 11 Tab 16 85.78 12.57 12 Scroll 17 83.09 7.73 13 14 Table 3: Mean usability scores by conditions 15 16 17 18 in table 3. 19 A Welch two samples t-test was used to analyse the effects of the Annota- 20 21 tion Interface on participants usability rating. The Welch’s t-test adjusts the 22 degrees of freedom to correct for potential unequal variance between sam- 23 ples. There was no significant effect of Annotation Navigation on usability 24 [t(24.65) = 0.74,p > 0.05]. Participants therefore did not rate the inter- 25 26 faces significantly differently in terms of usability. Our second hypothesis 27 was therefore not statistically supported. The means and standard errors 28 are displayed graphically in figure 6. 29 30 31 6. Discussion 32 33 Our results indicate that a comparison between a visual metaphor and a 34 more traditional scroll based approach to document navigation in a touch in- 35 36 teraction environment exhibits a significant interaction with the type of nav- 37 igation task undertaken on task efficiency. People were significantly quicker 38 completing the content tasks compared to the location task using the Scroll 39 40 condition. There was no significant difference between task efficiency in the 41 Tab condition. 42 When discussing navigation of annotation it is important to consider 43 the context of use. In our case we utilise annotations that contain cross- 44 45 references relative to both the content of the document and location within 46 the document. 47 For content-based tasks, where position of the annotation is provided 48 with respect to other content of the document, it is clear that immediate 49 50 access of the annotation content is important in supporting efficient navi- 51 gation. In this case the scroll bar provided an easily accessible overview of 52 each annotation. It therefore appears that some of the benefits of scrollbars 53 54 found in evaluations using traditional interfaces (Byrd, 1999; Chen et al., 55 2008; Igarashi and Hinckley, 2000; Sun and Guimbreti`ere, 2005) also apply 56 in touch-table interfaces (in the context of this study). 57 58 59 15 60 61 62 63 64 65 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 Figure 6: Mean & Standard Error of Software Usabiity Scale score for Annotation Navi- 35 gation conditions. 36 37 38 39 The benefit of a more natural tab based visual metaphor for annotation 40 41 navigation in the context explored was less clear. The results indicate that 42 participants in either annotation navigation conditions were able to navigate 43 to the relevant part of the document with similar efficiency for location based 44 45 navigation tasks. Additionally, there was no significant difference between 46 the annotation navigation approaches in terms of user perception of usability. 47 Our findings suggest that the Tab condition leads to a less efficient interaction 48 within the content tasks than the Scroll condition but that there is relatively 49 50 little difference between their efficiency in the location tasks. 51 We conclude that there is no discernable benefit from attempting to pro- 52 vide a more natural interface for navigation in this context and that more 53 traditional models of navigation utilised in desktop interfaces are quite ap- 54 55 propriate when modified to suit touch input. From our interaction it seems 56 that a more natural interface in fact impacts negatively on efficiency when 57 58 59 16 60 61 62 63 64 65 1 2 3 4 5 6 7 8 9 completing location based tasks. This is in contrast with seminal ideas on 10 11 metaphor and affordances (Norman, 1988) that are typically adopted in an 12 attempt to make interfaces more intuitive and simple to use. 13 This work demonstrates that context is all important when considering an 14 appropriate interaction design for digital document navigation. Not only is 15 16 the nature of the hardware platform of concern but perhaps more importantly 17 the nature of the annotations being presented and how they are being used 18 to navigate the document are important considerations. 19 20 21 6.1. Limitations and Further Work 22 This work attempts to fairly reflect the workflow of a research or non- 23 expert user navigating annotations within the context of a touch-table device 24 25 within a controlled study. However, there are a number of limitations of this 26 work that need to be considered. 27 One distinction between expert/non-expert users would be the length of 28 29 time spent interacting with such systems. Our experiment did not identify 30 the users experience with touch interactions before the experiment, although 31 it can be assumed that the amount of interactions with a touchtable in this 32 context is low. All participants were also given a practice session to acclima- 33 34 tise to the interface therefore the influence of such experience effects on task 35 efficiency are likely kept to a minimum. Our navigation task and annotations 36 were also relatively simple compared to the more complex cross-referencing 37 that is prevalent in the annotations of ancient manuscripts (Baechler and 38 39 Ingold, 2010; M¨uller and Maurer, 2011). Further work should aim to address 40 this limitation and explore: 41 42 • The scalability of these approaches. This is especially important, as 43 44 typically the number of annotations for a heavily researched manuscript 45 will be very large. 46 47 • Support for semantic and hierarchical structure within the annotation 48 such as topic or themed groupings. This will not only allow more 49 50 complex annotation structure and cross-referencing but also assist in 51 the retrieval of documents. 52 53 A further study should determine whether a touch interface plays a signif- 54 55 icant role in the effect presented in this paper. Currently the study does not 56 clarify whether these observed results are particular only to touch interfaces 57 58 59 17 60 61 62 63 64 65 1 2 3 4 5 6 7 8 9 or observed in other platforms such as small touch screen devices (such as 10 11 smart phones or tablet devices) or more traditional desktop environments. 12 The constraints of these platforms in terms of size and the modality differ- 13 ences between desktop and touch interactions may impact specifically on the 14 effect we have seen. It could be of value to identify and explore whether this 15 16 effect reflects across platforms and modalities. 17 The reliability of the SUS scale also highlights a reduction in internal 18 consistency in this context. It may be that when assessing such interactions 19 20 as the ones being investigated in this research the SUS is not as reliable 21 in measuring the construct of usability within a uni-dimensional scale as 22 mentioned in other research. Although the questionnaire achieved a good 23 reliability co-efficient, further research on touch table interaction should use 24 25 the SUS to help identify whether this reduction in reliability is an issue of 26 the interaction method being assessed. This would suggest refinement of the 27 metric to more accurately reflect touchtable interactions could be needed. 28 29 Given the context of a touch-table environment it would also be valuable 30 to investigate how navigation approaches impact on collaboration between 31 users and how navigation techniques can support referencing between docu- 32 ments. Touchtables naturally provide an opportunity for collaboration and 33 34 co-discovery, although recent research suggests users find it difficult to fully 35 embrace collaborative working on touchtables (Marshall et al., 2008). It 36 would be interesting to identify how navigation of annotation could be de- 37 signed to impact and encourage co-discovery and how this could impact task 38 39 efficiency and subjective usability. Recent research on interfaces used to sup- 40 port co-reading (Pearson et al., 2012) could be a useful starting point for 41 such research. 42 43 Additionally further work should look to identify ways of developing in- 44 terfaces to facilitate user generated content on such systems. Although we 45 supplied all annotations for users to search in a cultural and heritage context 46 one can envisage people potentially having a desire to contribute content 47 48 and annotations that could be valuable to the interpretation of the docu- 49 ment. Whether in a museum or academic collaboration context, designing 50 an interface to facilitate contribution is something that could be of value in 51 keeping annotations of these documents current and user led. 52 53 54 55 56 57 58 59 18 60 61 62 63 64 65 1 2 3 4 5 6 7 8 9 7. Conclusion 10 11 To conclude, our research suggests that the type of navigation design 12 13 impacts efficiency of task completion, yet this is dependent on the type of 14 task being conducted. We found that there was no significant difference in 15 task efficiency when completing the tasks using a more real world tab nav- 16 igation condition. However a scroll navigation interface was felicitous when 17 18 completing content-based tasks compared to location based tasks. Therefore 19 when navigating annotations on touchtables, a scroll based interaction seems 20 to bring benefits in content based task completion. As mentioned although 21 we found this in a touchtable-based interaction, further research is needed 22 23 to gather whether such an effect is relevant to other touch based platforms 24 and other modalities of interaction. This paper also highlights the benefits 25 of creating more natural affordances on touchtable based interactions, within 26 27 an annotation navigation context, may not be universal. 28 29 8. Acknowledgments 30 31 This work was undertaken with support of JISC UK. 32 33 34 References 35 36 Agosti, M., Bonfiglio-Dosio, G., Ferro, N., 2007. A historical and contemorary 37 study on annotations to derive key features for systems design. Int J Digit 38 Libr 8, 1 – 19. 39 40 Appert, C., Fekete, J.-D., 2006. Orthozoom scroller: 1d multi-scale navi- 41 42 gation. In: Proceedings of the SIGCHI conference on Human Factors in 43 computing systems. CHI ’06. ACM, New York, NY, USA, pp. 21–30. 44 45 Baechler, M., Ingold, R., 2010. Medieval manuscript layout model. In: Pro- 46 ceedings of the 10th ACM symposium on Document engineering. DocEng 47 ’10. ACM, New York, NY, USA, pp. 275–278. 48 49 Bangor, A., Kortum, P. T., Miller, J. T., 2008. An empirical evaluation 50 51 of the system usability scale. International Journal of Human-Computer 52 Interaction 24 (6), 574–594. 53 54 Bargeron, D., Moscovich, T., 2003. Reflowing digital ink annotations. In: 55 Proceedings of the SIGCHI conference on Human factors in computing 56 57 systems. CHI ’03. ACM, New York, NY, USA, pp. 385–393. 58 59 19 60 61 62 63 64 65 1 2 3 4 5 6 7 8 9 Bederson, B. B., 2011. The promise of zoomable user interfaces. Behaviour 10 11 and Information Technology 30 (6), 853–866. 12 13 Brooke, J., 1996. SUS-A quick and dirty usability scale. Usability evaluation 14 in industry, 189–194. 15 16 Brush, A. J. B., Bargeron, D., Gupta, A., Cadiz, J. J., 2001. Robust anno- 17 18 tation positioning in digital documents. In: Proceedings of the SIGCHI 19 conference on Human factors in computing systems. CHI ’01. ACM, New 20 York, NY, USA, pp. 285–292. 21 22 Buchanan, G., Owen, T., 2008. Improving navigation interaction in digital 23 documents. In: Proceedings of the 8th ACM/IEEE-CS joint conference on 24 25 Digital libraries. JCDL ’08. ACM, New York, NY, USA, pp. 389–392. 26 27 Byrd, D., 1999. A scrollbar-based visualization for document navigation. In: 28 Proceedings of the fourth ACM conference on Digital libraries. DL ’99. 29 ACM, New York, NY, USA, pp. 122–129. 30 31 32 Chen, N., Guimbretiere, F., Dixon, M., Lewis, C., Agrawala, M., 2008. Nav- 33 igation techniques for dual-display e-book readers. In: Proceedings of the 34 twenty-sixth annual SIGCHI conference on Human factors in computing 35 systems. CHI ’08. ACM, New York, NY, USA, pp. 1779–1788. 36 37 38 Chiu, P., Liao, C., Chen, F., 2011. Multi-touch document folding: gesture 39 models, fold directions and symmetries. In: Proceedings of the 2011 annual 40 conference on Human factors in computing systems. CHI ’11. ACM, New 41 York, NY, USA, pp. 1591–1600. 42 43 44 Cockburn, A., Ahlstr¨om, D., Gutwin, C., Mar. 2012. Understanding perfor- 45 mance in touch selections: Tap, drag and radial pointing drag with finger, 46 stylus and mouse. Int. J. Hum.-Comput. Stud. 70 (3), 218–233. 47 48 Cockburn, A., Gutwin, C., Alexander, J., 2006. Faster document navigation 49 50 with space-filling thumbnails. In: Proceedings of the SIGCHI conference 51 on Human Factors in computing systems. CHI ’06. ACM, New York, NY, 52 USA, pp. 1–10. 53 54 Cockburn, A., Kristensson, P. O., Alexander, J., Zhai, S., 2007. Hard lessons: 55 56 effort-inducing interfaces benefit spatial learning. In: Proceedings of the 57 58 59 20 60 61 62 63 64 65 1 2 3 4 5 6 7 8 9 SIGCHI conference on Human factors in computing systems. CHI ’07. 10 11 ACM, New York, NY, USA, pp. 1571–1580. 12 13 Eslambolchilar, P., Murray-Smith, R., 2006. Model-based, multimodal inter- 14 action in document browsing. In: Renals, S., Bengio, S., Fiscus, J. (Eds.), 15 Machine Learning for Multimodal Interaction. Vol. 4299 of Lecture Notes 16 17 in Computer Science. Springer Berlin / Heidelberg, pp. 1–12. 18 19 Geller, T., 2006. Interactive tabletop exhibits in museums and galleries. Com- 20 puter Graphics and Applications, IEEE 26 (5), 6 –11. 21 22 Golovchinsky, G., Denoue, L., 2002. Moving markup: repositioning freeform 23 annotations. In: Proceedings of the 15th annual ACM symposium on User 24 25 interface software and technology. UIST ’02. ACM, New York, NY, USA, 26 pp. 21–30. 27 28 Guiard, Y., Chapuis, O., Du, Y., Beaudouin-Lafon, M., 2006. Allowing cam- 29 era tilts for document navigation in the standard GUI. In: Proceedings 30 31 of the working conference on Advanced visual interfaces - AVI ’06. ACM, 32 New York, New York, USA, pp. 241–244. 33 34 Igarashi, T., Hinckley, K., 2000. Speed-dependent automatic zooming for 35 browsing large documents. In: Proceedings of the 13th annual ACM sym- 36 37 posium on User interface software and technology. UIST ’00. ACM, New 38 York, NY, USA, pp. 139–148. 39 40 Jones, M., Marsden, G., Mohd-Nasir, N., Boone, K., Buchanan, G., 1999. 41 Improving web interaction on small displays. Computer Networks 31 (11), 42 43 1129–1137. 44 45 Kim, J.-K., Farzan, R., Brusilovsky, P., 2008. Social navigation and anno- 46 tation for electronic books. In: Proceedings of the 2008 ACM workshop 47 on Research advances in large digital book repositories. BooksOnline ’08. 48 ACM, New York, NY, USA, pp. 25–28. 49 50 51 Kline, P., 2000. A psychometrics primer. Free Association Books. 52 53 Lewis, J. R., Sauro, J., 2009. The factor structure of the system usability 54 scale. In: Proceedings of the 1st International Conference on Human Cen- 55 tered Design: Held as Part of HCI International 2009. HCD 09. Springer, 56 57 Berlin, Heidelberg, pp. 94–103. 58 59 21 60 61 62 63 64 65 1 2 3 4 5 6 7 8 9 Liesaputra, V., Witten, I. H., 2012. Realistic electronic books. International 10 11 Journal of Human-Computer Studies 70 (9), 588 – 610. 12 13 Marshall, P., Hornecker, E., Morris, R., Dalton, N. S., Rogers, Y., 2008. 14 When the fingers do the talking: A study of group participation with 15 varying constraints to a tabletop interface. In: Proceedings of the 3rd 16 17 IEEE International Workshop on Horizontal Interactive Human Computer 18 Systems. TABLETOP’08. IEEE, pp. 33–40. 19 20 M¨uller, H., Maurer, H., 2011. How to carry over historic books into social 21 networks. In: Proceedings of the 4th ACM workshop on Online books, 22 23 complementary social media and crowdsourcing. BooksOnline ’11. ACM, 24 New York, NY, USA, pp. 25–34. 25 26 Norman, D. A., 1988. The Psychology of Everyday Things. Basic Books. 27 28 O’Hara, K., Sellen, A., 1997. A comparison of reading paper and on-line 29 documents. In: Proceedings of the SIGCHI conference on Human factors 30 31 in computing systems. CHI ’97. ACM, New York, NY, USA, pp. 335–342. 32 33 Olsen, Jr., D. R., Taufer, T., Fails, J. A., 2004. Screencrayons: annotating 34 anything. In: Proceedings of the 17th annual ACM symposium on User 35 interface software and technology. UIST ’04. ACM, New York, NY, USA, 36 37 pp. 165–174. 38 39 Pearson, J., Buchanan, G., Thimbleby, H., 2009. Improving annotations in 40 digital documents. In: Proceedings of the 13th European conference on 41 Research and advanced technology for digital libraries. ECDL’09. Springer, 42 43 Berlin, Heidelberg, pp. 429–432. 44 45 Pearson, J., Buchanan, G., Thimbleby, H., 2011. The reading desk: support- 46 ing lightweight note-taking in digital documents. In: Proceedings of the 47 15th international conference on Theory and practice of digital libraries: 48 49 research and advanced technology for digital libraries. TPDL’11. Springer- 50 Verlag, Berlin, Heidelberg, pp. 438–441. 51 52 Pearson, J., Owen, T., Thimbleby, H., Buchanan, G. R., 2012. Co-reading: 53 investigating collaborative group reading. In: Proceedings of the 12th 54 55 ACM/IEEE-CS joint conference on Digital Libraries. JCDL ’12. ACM, 56 New York, NY, USA, pp. 325–334. 57 58 59 22 60 61 62 63 64 65 1 2 3 4 5 6 7 8 9 Robinson, P., 2010. Virtual manuscript room (vmr). Final report, Joint In- 10 11 formation Systems Committee. 12 URL http://www.mingana.bham.ac.uk/ 13 14 Shaer, O., Strait, M., Valdes, C., Feng, T., Lintz, M., Wang, H., 2011. 15 Enhancing genomic learning through tabletop interaction. In: Proceedings 16 17 of the 2011 annual conference on Human factors in computing systems. 18 CHI ’11. ACM, New York, NY, USA, pp. 2817–2826. 19 20 Smith, G. M., schraefel, m. c., 2004. The radial scroll tool: scrolling support 21 for stylus- or touch-based document navigation. In: Proceedings of the 22 23 17th annual ACM symposium on User interface software and technology. 24 UIST ’04. ACM, New York, NY, USA, pp. 53–56. 25 26 Sun, L., Guimbreti`ere, F., 2005. Flipper: a new method of digital document 27 navigation. In: CHI ’05 extended abstracts on Human factors in computing 28 29 systems. CHI EA ’05. ACM, New York, NY, USA, pp. 2001–2004. 30 31 Terrenghi, L., Kirk, D., Sellen, A., Izadi, S., 2007. Affordances for manipula- 32 tion of physical versus digital media on interactive surfaces. In: Proceed- 33 ings of the SIGCHI conference on Human factors in computing systems. 34 35 CHI ’07. ACM, New York, NY, USA, pp. 1157–1166. 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 23 60 61 62 63 64 65