Keywords

1 Introduction and Motivation

Parallel web browsing is becoming prominent part of our daily lives. Nowadays, all users access multiple web sites through web browser tabs or multiple browser windows, whether on tablets, smart phones or personal computers. Yet, tabs and windows are insufficient in support parallel web browsing. Previous work has found that a minimum of 57.4% of users’ time is spent switching tabs, where they require continual attention to switch pages [2, 15, 16]. Moreover, the performance of multiple-browser windows in accomplishing parallel web browsing tasks is largely proportional to screen size [25], and overall, they are inferior to interaction techniques that specifically support parallel web browsing [1, 25]. Thus, such shortcomings present multi-faceted challenges in interaction and visualization techniques to successfully support web browsing [24].

As virtual reality (VR) becomes more accessible to people, it is easier to take advantage of the visualization and interaction capabilities enabled by VR. In general, VR can provide advantages in interaction and visualization that surpass experiences on traditional devices, like desktops and tablets [8]. Yet, such potential remain challenging as the visualization and interaction possibilities are still under heavy researches.

In this work, we propose VRowser, which is a virtual reality web browser that utilizes various visualization and interaction methods to support webpage content comparisons, allocation, grouping, and retrieval that are essential of parallel web browsing [15, 25]. Our approach essentially uses parallel web browsing requirements as determining design factors for VRowser. In this paper, our contributions are the following: (1) VR parallel web browsing system that (A) Maintains familiar web browsing metaphors. (B) Leverages VR interaction and visualization capabilities to support parallel web browsing tasks. (2) Presents evaluation results of our proposed platform that investigates user preferences, behaviors and performance in mentioned parallel web browsing tasks.

This paper is organized as follows. Section 2 presents some related researches to our work. Section 3 describes the design and implementation of VRowser. In Sect. 4, we show the results of the evaluation of VRowser. Finally, Sect. 5 concludes the paper.

2 Related Work

2.1 Web Browsing in Virtual Environments

Researchers introduced numerous metaphors to access and visualize web contents in virtual environments (VE). Some work attempted to alter the document metaphor [4, 5] of webpages, such as representing webpages as rooms in a virtual world [11] or driving a car to navigate through webpages [3, 20].

Moreover, the city metaphor [10, 22] portrays webpages as numerous building, which are textured text and other multimedia information, where users are able to navigate these cities using a variety of locomotion methods. These metaphors could bring numerous advantages for specific web browsing tasks, such as revisitation [22] or navigation [10]. In addition, such collectively exploit associative memory techniques to enable faster retrieval of information at specific locations around the city.

In contrary to work that attempts to utilize various metaphors and interaction methods, additional work within VE attempted to maintain the webpages’ document metaphor. The document metaphor is familiar to all users, as it is the metaphor essentially used in all devices to browse the web. Work within this direction mainly textured rectangular 3D objects with webpage contents and integrated various interaction means to deliver the user experience. Earlier work [5, 7] have investigated a method to embed 2D desktop applications within VR, including a web browser, where users could click on various contents using their fingers. Likewise, later works [3, 13] proposed various document-texturing architectures for 3D environments as well as and investigating gesture based interactions with webpages.

2.2 Commercial VR Web Browsing Systems

With the VR revolution that has popularized VR and made it accessible to all users, an expanding number of applications has been made available to experiments with browsing in VR. Standard VR web browsers, such as Oculus browsersFootnote 1, intend to provide web browsing experience similar to that of desktops, albeit with modified input capabilities. Additional community developed browsers, such as those found in smartphone VR enclosures, enable users to browse webpages in various locations, whilst maintaining the overall document metaphor. Yet, such browser mimics the interaction and visualization experience on the personal computer, thus, not taking full advantage of what VR can potentially provide.

Lastly, JauntVR [17] alters the visualization of webpages by showing them as rooms with various 3D interactive contents. Despite its innovation, JauntVR requires customizing a webpage’s code for compatibility, and we believe the final web browsing experience totally deviates from the standard web browsing experience users are familiar with.

All in all, we argue that such web browsers do not provide significant advantage over desktop web browsers, but rather the primitive capability to view webpages within such VR environments in a similar user experience that is found in desktop computers.

2.3 Web Workspaces

An information workspace [9] is an environment where users can manipulate and store information or documents of interest. VEs embody the concept of workspaces by utilizing varied interaction and visualization metaphors to enhance the users experience, performance or to achieve other objectives within the workspace concept.

Web workspaces has also been previously investigated in various literature. Such work mainly attempted to combine visualization and interaction techniques as a metaphor to interact with webpages and support various tasks. Card et al.’s research provided insights towards interaction in 3D web browsing environments to support revisitation and bookmarking [23]. Their work included; (i) The webbook: a 3D web browsing environment based on the book metaphor; (ii) The web forager: a 3D information workspace which enables users to aggregate various webbooks. Jhavery et al.’s work presented a web workspace that supported a number of web browsing tasks, especially content comparison [18].

Lastly, two-dimensional workspaces have been investigated. For example, Data Mountain [21] supported collecting and grouping webpages a 3D environment, taking advantage of human spatial memory to recall needed webpages.

3 VRowser

VRowser is a VR web browser system that embodies various design factors and interaction concepts to support parallel web browsing. Our approach extends the concept of web workspaces, such as in Data Mountain [21], which takes advantage of spatial memory cues to group and organize webpages in various workspaces. Our work additionally adds the ability to (1) modify page locations in 3D space (2) enable extended viewing space within immersive VR (3) change the size and orientation of webpages within the 3D VR space.

3.1 Design Factors

Immersive VR:

The utilization of immersive VR could bring several advantages to the interaction experience. In addition to potential advantages of VEs, such as depth cues [22] and spatial memory [12, 21], VR provides an unlimited view space for users to view and manipulate webpages. Such aspect is significant overcoming limitations of limited screen size [14, 25], or limited workspaces of AR that rely on spatial information registration [1]. Lastly, immersion can be utilized to enhance task performance as in previous work [8, 25]. Moreover, 3D user interface visualization and interaction capabilities can be utilized within VR to enhance the web browsing experience and achieve our design objectives.

Maintaining Webpages’ Document Metaphor:

As in previous work [1, 21, 25], we have chosen to preserve the document metaphor to insure a familiar user experience with webpages, both in terms of interaction and visualization. Despite the potential advantages metaphors could bring, such as in [6, 7, 14, 19], we believe numerous shortcomings would arise. For instance, other visualization methods, as in the city or gallery metaphors, could have various limitations in terms of restricting amount of viewable contents or longer accessibility cost [6, 10]. Thus, besides users’ unfamiliarity and various potential shortcomings, such metaphors require conversion mechanisms that transform webpages from documents to the chosen other forms (As in buildings in the city metaphor, or deconstructed-documents as in the gallery metaphor), which is additionally prone to further interaction or visualization issues. As a result, similar to previous work [3], we have chosen to maintain webpages’ interaction and visualization methods.

Segregated Interactions Methods:

VRs and webpages require different interaction methods that suite both webpages and VEs. As such, Jankowski [14] classified interaction with webpages within VEs into: (1) Web Tasks: which consist of interactions with webpages, such as scrolling or navigating to a certain URL. (2) 3D Tasks: which cover activities related to 3D-attribute manipulations, such as scaling or locomotion within VEs. As such, we have to provide suitable interaction methods that suit each task type, whilst also fulfilling web activity needs.

3.2 VRowser Concept

VRowser aims to: (1) Provides an immersive web-space experience for webpage management and revisitaiton. (2) Supports content comparisons across webpages. Thus, our approach intends to leverage VR immersion, visualization and interaction techniques to achieve the above objectives.

VR Web Workspace (VRW):

Our concept of VRW reflects how users group pages of similar interests, whether in tabs and browser windows while browsing or in different folders for bookmarking. A VRW consists of a virtual space where users can freely manage webpages (Fig. 1). Each VRW contains different geographic layout, such as rivers and mountains, where users can place webpages around. Users can navigate VRW as they do with real world locations. Lastly, users multiple VRWs as required and freely transfer webpages across them.

Fig. 1.
figure 1

VRW House (Upper) and Forest (Lower)

Full size image

Web page Management is mainly concerned with 3D web tasks, which are mainly selection, positioning, rotation and scaling [7]. For selection, users ray-cast a hand-beam towards desired webpages to select them. Similarly, selected webpages can be positioned by directly pointing the users’ hand towards the location desired VRW location. Lastly, selected webpages can be rotated and scaled by directly manipulating such attributes using hand gestures; rotating the edge of a page or grabbing both ends of a page and waving-in to scale down and vice versa.

The VR Web Workspace (VRW):

Similar to browser windows that comprise tabs, a VRW is a VR space that contains a numerous webpages allocated throughout the space’s 3D environment (Fig. 1). Each VRW contains different geographic layout. Users can navigate and create VRW and fill them with situate webpages as needed. Lastly, users can create multiple VRWs as required and freely transfer webpages across them.

Web Page Management:

Which consist of 3D web tasks. For selection, users ray-cast a hand-beam towards desired webpages to select them. Similarly, selected webpages can be positioned by directly pointing the users’ hand towards desired VRW location. Rotation and Scaling is done by directly manipulating such attributes using hand gestures; rotating the edge of a page or grabbing both ends of a page and waving-in to scale down and vice versa (Fig. 2).

Fig. 2.
figure 2

A user is grabbing and manipulating the 3D location of a webpage (Upper), pointing to interact with web contents (Lower)

Full size image

Web Tasks:

Dedicated inputs and interaction methods are provided to interact with web contents. In Raycasting [7] a user’s hands are utilized to control pointers, with a dedicated joystick button to mimic left click (Fig. 2). Likewise, scrolling is done with right clicking and physically dragging in a direction to scroll. Lastly, as a prominent problem in VR, we utilize a keyboard for text entry.

3.3 Implementation

  1. A.

    Hardware

    1. 1.

      VR Headset: In this system, we use HTC ViveFootnote 2 for VR head mount display.

    2. 2.

      Interaction space: Our experience is mainly a seated VR experience, where users are expected sit down while engaging in immersive VR. Yet, the existence of positional tracking, through HTC Vive, it is possible to physically walk around the tracked space and interact with contents.

    3. 3.

      Interaction Method: We utilize the HTC Vive’s controller as the main method of interaction. Users hold a controller in each hand, where the left controller is essentially used for locomotion tasks within the VRW, while the right controller is utilized to carry web tasks. The possible interactions in each controller are illustrated in Figs. 3 and 4, and include the following:

      Fig. 3.
      figure 3

      Operation manual of input device. Green indicates 3D Tasks, Yellow indicates Web Tasks. 3D-Tasks. (The “Trigger Button” has various functionalities effecting 3D Tasks as shown in Fig. 4) (Color figure online)

      Full size image
      Fig. 4.
      figure 4

      Description of 3D tasks

      Full size image
      1. (A)

        Scaling: Upon clicking the Trigger button, users may move the controllers closer or further away from one another to make pages bigger or smaller, respectively (Fig. 4-1).

      2. (B)

        Rotation: Upon clicking the Trigger button next to a webpage (Fig. 4-2), webpages can be rotated by directly rotating the joystick in x, y, z axis.

      3. (C)

        Capture and Carry: this feature allows users to select specific webpages to carry with them, after which they can place it in specific locations within the VRW. To execute this interaction, the user aims the controller towards intended webpage and clicks the “Trigger Button” (Fig. 4-3), after which, (4) Users may view captured webpages by clicking the “Grip Button” (Fig. 4-4). To release a webpage, the clicks the grip-button to view available carried webpages and then points and presses the “Trigger Button” at the desired webpage to be released.

  2. (B)

    Software

    1. 1.

      System: Our prototype was developed using Unity 3DFootnote 3 (version 2017.1.1). We additionally utilize supplementary VR PluginFootnote 4 to use HTC Vive with Unity. Additionally, all input is handled by Unity3D input manager, after which each input is allocated specific 3D or 2D task in accordance with our design factors.

    2. 2.

      Web Browsing System: To display web browsers within the 3D environment, we utilize a plug inFootnote 5 that enables us to convert webpages to textures, after which we can utilize such textures on 3D objects. Likewise, we developed our control methods to incorporate mentioned plug-in. We access and retrieve webpages using HTTP communication, which allows our system to be expandable for future iterations that may involve other webpage contents using a similar architecture.

4 Evaluation

4.1 User Study

  1. A.

    Design

    • Objective: Our goal is to investigate impressions and webpage placement strategies within VRowser

    • Participants: The experiment was conducted for 10 college students (9 male; 20 to 25 years old) all of whom have had used VR applications before.

    • Conditions: The user study was carried on two different VRWs, a House and Forest. Each VRW has varied geographical attributes (Fig. 5). While the House includes various rooms and furniture, the forest only includes trees and a simple road. The user study comprised the same set-up described in the implementation section, where we utilized a seated VR experience for this user study (Fig. 6).

      Fig. 5.
      figure 5

      A heat map that show the concentration of where webpages where allocated each VRW, inside the house (Upper) and the forest (Lower)

      Full size image
      Fig. 6.
      figure 6

      A user wearing the HTC Vive and using VRowser while seated.

      Full size image

    • Scenario: The study included 10 several genres webpages are arranged in VRW. Thus, after a brief familiarization session, participants were placed within each VRW, starting with the House and then the Forest, and were given 10 webpages to arrange. Next, each participant was given 5 min to check the contents of the webpages, after which they were given 10 min to arrange the pages to their liking within the VRW.

      The pages used were the following:

We diversified the type of webpages to cover a wide range of user interests, and so we can observe how different types of pages may affect the placement at various VRW locations.

Lastly, participants took a questionnaire and a semi-structured interview.

Collected Data:

We collected the location of each webpage within the two VRWs. And, after the experiment we interviewed participants about VRowser. The contents of the question asked the degree of satisfaction with the system and which VRW was superior.

  1. B.

    Results and analysis

Web Page Placement

Figure 7 shows the heat map of the placement of webpages in each VRW. In VRW House, many participants placed webpages around featured things in the room. For example, bed is the most prominent object in the bedroom. Also, most webpages were placed in front of the TV in the living room. We thought that this result was generated as participants usually watch webpages on a display. The second most were placed on the bedroom bed. We thought that there were many participants who are interested because the bed stands out from objects in other rooms.

Fig. 7.
figure 7

A heat map that show the concentration of where webpages where allocated each VRW, inside the house (Upper) and the forest (Lower)

Full size image

From the interviews, we asked participants about specific page arrangement within VRW. In VRW-House, many participants placed the webpage in association with the characteristics of the VRW. For example, since the bedroom is a private space, most participants placed web site under the category “Hobbies” in the bedroom.

Meanwhile, in the VRW Forest, the placement strategies where quite different from each participant to another. Although participants attempted to utilize a similar strategy as to the House, we believe that the lack of distinguishing map characteristics, such as unique tree arrangements or objects, were a challenge. Thus, participants utilized the scarcity of landmarks by placing many pages on one distinguishing landmark, such as on an easily found tree or where the road forked. Some participant utilized tree condensation as an indicator of amount of information of a placed webpage.

All in all, we conclude that all participants preferred the House VRW over the Forest VRW. Their preference was based on the fact that the house was a place that is easy to navigate in, as it included a variety of objects that acted as landmarks to situate and retrieve placed webpages. In the contrary, the forest lacked sufficient amount of distinct characteristics that helped in navigation or placing webpages.

On the other hand, many participants placed webpages around trees in the VRW Forest. However, as in Fig. 7, the webpage placement was almost randomized and we could not conclude specific placement preferences among users. Based on the above results, we thought that VRW with familiar and noticeable locations and landmarks, like the house, was probably better suited for use as VRW.

Interview Result

Table 1 is a summary of the contents of the interview. First, various participants indicated that the implemented VRW locomotion method was quite slow, elaborating that it could be faster. Thus, we a better locomotion method is required for future iterations, such as warp to arbitrary places quickly.

Table 1. Interview summary
Full size table

5 Conclusion

In this paper, we proposed VRowser, which is a virtual reality web browser that utilizes various visualization and interaction methods to support webpage content-comparisons, allocation, grouping, and retrieval that are essential of parallel web browsing. Our approach essentially uses parallel web browsing requirements as determining design factors for VRowser. VRowser shows that virtual reality parallel web browsing system that maintains familiar web browsing metaphors and leverages virtual reality interaction and visualization capabilities to support parallel web browsing tasks.

We believe our evaluation results are promising to pursue further research. Although participants showed strong preference towards VRWs that comprise a larger number of landmarks over those with smaller one, the type of landmarks that a VRW comprise requires further investigation. Moreover, the size of VRW, and how it relates to the number of available landmarks also requires deeper investigations, as such outcome would facilitate creating efficient VR workspaces for future immersive systems.