Understanding and supporting multi-session web tasks

Authors


Abstract

We explore how users manage multi-session web tasks (tasks that require more than a single web session to complete). When users are performing multi-session web tasks, they must ‘make sense’ of information that they gather during each of the Web sessions and they must also ‘keep a sense’ of the information between the sessions. We present the results from two studies that examine how users are performing multi-session tasks and outline three prototype tools that we are currently evaluating to help users perform multi-session tasks.

Introduction

Today, people are using the Web as a major information resource to perform complex tasks once reserved for paper sources. For example, previously to plan a vacation, people would meet with travel agents to collect various travel brochures and maps, and to get quotes on flights and hotels. People could look through the brochures at home, plan routes on the maps and compare schedules and costs of flights. They could mark important information and return to it at different times before making decisions. Today, all of this information can be found on the Web through a variety of web sources (e.g., tourism agencies, travel specialty sites, airlines, etc.).

Despite the increased complexity of tasks that can be performed on the Web, the basic tools used by the popular browsers have changed little and, in general, tend to focus on supporting tasks that can be completed in a single web session. There are few browser tools available to support those tasks that span multiple sessions, such as planning a vacation or researching a paper. We define these tasks that span multiple web sessions as multi-session tasks. We define a web session as continuous interaction with a Web browser. A multi-session task is goal-based and requires more than one web session to complete. There is a definable point as to when the task is over, such as a specific date, or an event that occurs (e.g., you finally purchase the computer), or it is abandoned.

From our first two studies, a diary and a field study, we found that users utilize particular browser tools and other applications to help organize and use the information that they find. In this paper, we characterize the type of tasks that users in the two studies defined as multi-session tasks and summarize how users managed the information and the subtasks when they performed their own multi-session tasks. We present three prototype browser tools designed on the basis of those studies to help users organize and make sense of the information found on the Web in order to complete their multi-session tasks.

Background

Previous research has examined the types of tasks and how people perform these tasks on the Web (Choo et al., 2000; Kellar et al., 2007; Sellen et al., 2002). However much of this research has focused on tasks occurring during a single session. We have focused on examining user behavior for those tasks that span multiple web sessions.

While some web tasks can be completed within a single session there are many web tasks that require multiple sessions. For example, Sellen et al. (2002) found that many (40%) information gathering tasks took more than one session to complete and Kellar et al. (2007) reported that 59% of information gathering tasks were repeated. Not only do many web tasks require more than one session to complete, they often involve revisiting web pages. In 1997, Tauscher and Greenberg found that 58% of pages visited by users were previously accessed and in 2001, Cockburn and McKenzie reported that 81% of visits to web pages are previously seen.

A main problem with multi-session tasks is to determine how to help users maintain the information related to the task between web sessions. While there are different ways that people can save information between sessions (e.g., print the web page, save the page or make notes), the two most commonly used browser tools for revisiting information are history lists and bookmarks (Abrams et al., 1998; Cockburn & McKenzie, 2001; Jhaveri & Räihä, 2005).

Many studies have shown (Aula et al., 2005; Bryne et al., 1999; Kellar et al., 2007; Tauscher & Greenberg, 1997) that users tend not to use history lists as several problems exist, such as the lack of page descriptors and confusing organization of previously viewed pages. Bookmarks provide a means to revisit pages, provide groupings of categories and provide storage for important page addresses (Abrams et al., 1998), but there are many disadvantages. Bookmark lists require organization and management. Refinding pages in bookmark lists can be difficult at times if the lists are long and unruly, and the lists often lack a naming description of the marked page (Abrams et al., 1998; Aula et al., 2005; Brown & Sellen, 2005; Jones et al., 2001). Despite these shortcomings people still use bookmarks (Abrams et al., 1998; Aula et al., 2005; Brown & Sellen, 2005; Bruce et al., 2004; Kellar et al., 2007; Tauscher & Greenberg, 1997). In 2004, Bruce et al. reported that 89.72% of 214 respondents used bookmarks frequently. In 2005, Aula et al. also found that 92.4% of experienced web users used bookmarks with 63% of these users having 51 to 300 bookmarks and 16% having more than 300 bookmarks.

People also perform a variety of other actions to allow for later access to information (Aula et al., 2005; Brown & Sellen, 2005; Bruce et al., 2004; Jones et al., 2001). Users utilize search engines to relocate pages; print or save web pages; email links to themselves or others; and try to remember the URL of an actual page. Users also write URLs on paper, paste a whole page or relevant parts into a separate document, and create a web page of URLs for anywhere access.

There has been some exploration to improve the effectiveness of revisitation tools for web pages, although currently these are not included in standard browsers. For example, Kaasten and Greenberg (2001) integrated the back key, history list and bookmarks into a single tool with a visual representation of how often pages were visited, while WebView (Cockburn et al., 1999) has browser enhancements such as zoomable thumbnail pages integrated with back/forward menus. Session Highlights (Jhaveri & Räihä, 2005) is a browser add-on that creates an immediate storage area, while still using bookmarks for more permanent storage of web pages. It allows users to open several pages at once, compare page content and create and save collections.

These approaches are designed to facilitate re-visitation to a particular web page but not to refind specific information previously accessed within the web page itself. Bryne et al., (1999) noted that locating items on a page (e.g., having to scroll or search) is a time consuming task. Landmarks (MacKay et al., 2005) combine the revisitation of a web page with the refinding of information on the page. Landmarks allow users to mark information on a web page that they may want to return to in a later browsing session and are added in the same fashion as a bookmark. One or several landmarks can be added to a page and they can be renamed and organized into folders. When the user selects a landmark from their bookmark list it will return the page scrolled down to the exact location of indicated text, which is highlighted, for easy recognition.

The Problem

When users are performing multi-session web tasks they must ‘make sense’ and manage the information that they gather during web sessions and they must also keep track of the current status of the information gathered between the sessions. When a user begins a multi-session task they first identify the goal of the task, anticipate the sub-tasks associated with the task and generally determine the end point of the task (e.g., a deadline or an event). During each session, the user may search the Web for more information and choose to integrate this information as they try to complete the task. Users may use a search engine to find new potential documents or they may return to previously viewed information (e.g., open a bookmark, re-search the topic or type in the URL). For each information source, the user needs to utilize their judgment of the usefulness of the information in the context of their task. They must judge whether the data is trustworthy, whether the data is relevant to the task, whether they will need the information in the future and whether they will need to perform any actions relating to the information. Possible actions could include submitting a form within the web page, or if the information is deemed important for the future they could save the information (e.g., print the page, cut/copy to another document, email or bookmark) or they could share the information with others.

What further complicates this type of information management is that users are often performing more than one task on the Web during any given session (Spink et al., 2006), which has most likely increased with tabbed browsing. During multi-tasking users must keep track of streams of information while switching between tasks. Spink et al., (2006) found that even for a single task there are a number of actions that a user may undertake (such as, seeking and search, interactive browsing, retrieving and constructing information).

Additionally, for multi-session tasks users also need to ‘keep sense’ or track the information between sessions. When they return to the task at a later time they must try to remember what state they left the task. For example, users must remember which sub-tasks that they still need to perform, what information they have already seen, and what information that they have saved and where they have stored it.

Methodology

To identify characteristics of multi-session tasks and how people perform these tasks, we conducted two studies: a diary study where participants recorded their own multi-session tasks using a web form and field study where participants used a customized version of Firefox that logged all of their interactions with the browser (MacKay & Watters, 2008).

For both studies, we first met with participants to describe the study, to provide our definition of multisession tasks, and to acquaint them with the necessary study tools (i.e., the web diary or the customized browser for the field study). Participants self-reported their own multi-session tasks using the given study tool for the study period. They also filled in a demographics questionnaire. After both of the studies, we met again with each participant. In this final meeting, they filled in a final questionnaire and performed a semi-structured interview where we discussed one of their multi-session tasks in detail and multi-session tasks in general. We also demonstrated existing tools that might assist them with their multi-session tasks and to get feedback on features that could be helpful for multi-session tasks.

Further details of the methodology of both studies and the customized browser used in the field study can be found in MacKay and Watters (2008).

Diary Study

Twenty-two participants took part in the web-based diary study (18 for a one week study period and four for a two week study period). Participants (12 male and 12 female) were from a university community of students, faculty and staff. Half the participants used Internet Explorer and the other half used Firefox as their main browser.

Participants recorded their own multi-session tasks using a web-based diary form. The diary form could easily be filled in at the time that they worked on a multi-session task. The form data was submitted to a secure server accessible by the researchers. At the start of any task likely to take more than one session, participants recorded in their web diary a task name, a short task description, and the anticipated end date (or event that would end the task). For the purpose of this study, participants were told to consider each time that they returned to their browser to visit the Web to be a session. During each session that they worked on a multi-session task (either starting or resuming), they recorded what they worked on for that task during the session, why they stopped working on the task, and what tools and other applications that they used.

We used the data collected during the diary study to help in the design of the logging feature in the customized version of Firefox for the field study.

Field Study

Twenty-four university students (20 male and 4 female) were recruited to participate in a four week field study. Participants were required to use the Web daily, and use Firefox as their main web browser. Participants used a customized browser for all of their web browsing sessions that logged their interactions in the browser (e.g., windows/tabbed opened, navigation types used, URLs visited, actions taken, etc.) for both their regular browsing activity and their own multisession tasks, which we initially called Transient Tasks. This customized browser was Firefox and when installed automatically imported all of the participants' bookmarks and Firefox extensions. The custom browser was password protected to ensure that we only logged the participants' actions and not other users of particular computers.

This browser had a built-in interactive toolbar (see Figure 1) that the participants used to record specific information related to their own multi-session tasks. As with the diary study, this allowed them to record specific information about their multi-session tasks (e.g., the task description, the anticipated end date, what they worked on during the session, why they stopped working on the task, and any outside applications that they used). The customized browser, however, could collect more detailed data from the browser logs, such as the count of browser tools and actions taken, the URLs visited, searches made, how many windows and tabs were opened, which windows/tabs were included in the task and allowed for the user to indicate when they switched between their current multi-session task and other web tasks. Participants could also install the browser on multiple computers and could access all their active multi-session tasks from any installed browser. For this study, we considered each time the browser was opened, as captured in the browser logs, as a new session.

The built-in toolbar enabled participants to indicate when they started working on a multi-session task (‘Start New Transient Task’) and each time they continued to work on multi-session tasks (‘Resume Task’) that included a drop down list of their active multi-session tasks (note, for this study, we used our earlier label ‘transient task’). When participants were working on a multisession task the name of the task would appear on the toolbar highlighted in yellow as seen in Figure 1.

Figure 1.

The interactive toolbar enables user to indicate when they start, stop and resume a multisession task. When a task is active, the name is highlighted in yellow. The user provides further information by indicating which web page/s are ‘included’ or ‘excluded’ from the task. As well, the user can view, manage and submit their browser logs from the toolbar.

We also asked users to let us know which open web pages were included in the task by using the toggle buttons on the toolbar. For example, if they had their web-based email open while working on their multi-session task but it was not related to the task, they would choose to “Exclude this tab or window from the task”. When a tab or window was not included in the active task, the task name in the toolbar would be visible but grayed out. This allowed us to determine which open web pages were actually related to the task to help us better understand how users work on web tasks (e.g., working on one task at a time or having several pages open at a time and switching between the tasks). When the participants stopped the task (“Stop Task” button), another web form appeared in which the participants recorded information on the task that they just worked on.

Each time that the participant used the custom browser, it logged all their interactions (for both multi-sessions and all their other browsing activities). On the toolbar, there were buttons that allowed the user to view their activity log (‘View Log’), delete any URLs they did not want the researchers to view (‘Manage Log URLs’), and to submit their logs (‘Submit Log’) which were sent to secure server. Participants were asked to submit their logs daily. The log contained the following information:

  • Computer (a unique name given by the participant to each computer that the browser was installed)
  • Event Id (a unique number representing each action that was captured in the log).
  • Window Id (during each session, all new windows opened were given a unique id)
  • Tab Id (during each session, all new tabs opened were given an unique id)
  • Date (the date that the action occurred)
  • Time (the time that the action occurred)
  • Name (a descriptor of the action performed)
  • URL (the web page address for any web pages visited)
  • Session Description (specific information related to multi-session tasks, e.g., why the task was stopped for that session).

Limitations of the studies

For both the diary and field studies we used a ‘convenient sample’ of mainly university students. This provided us with an experienced set of web users who are often early adaptors of technologies and who are heavy users of the Web. Our population and sample numbers are comparable to other studies (Brown & Sellen, 2005; Choo et al., 2000; Kellar et al., 2007; Sellen et al., 2002) which also reported on similar user sets and number of participants to acquire a deeper understanding of how experienced web users perform web tasks.

Asking participants to record their activities with multi-session tasks (the web-based diary for the diary study and the customized toolbar in Firefox for the field study) may have influenced their behaviour. However, we felt that this was necessary in order to gather the relevant data relating specifically to multi-session tasks. When the participants were asked in the final interview of the field study whether using the customized browser influenced their web browsing behaviour most noted that it did not. Others mentioned that the interactive toolbar kept them on task when performing multi-session tasks.

Results

Previously we analyzed the data collected from these studies to describe the user behaviour when performing these tasks and specific task characteristics (MacKay & Watters, 2008). In this work, we examine the studies in the context of making sense of the information while working on a multi-session task and again when returning to the task between sessions. We examine task switching within sessions containing multi-session tasks, the reasons that tasks are stopped and continued in a later session, and whether participants tend to keep information collected during a multi-session task for future use. We focus on the browser tools used in both studies specific to revisiting information (e.g., using bookmarks) or for keeping information between sessions (e.g., coping to a document). We also introduce three prototype tools developed from the results of these studies.

Multi-session Task Characteristics

There were 235 multi-session tasks recorded over both studies (85 in the diary study and 150 in the field study). On average, the participants worked on their multi-session tasks 204 times during the diary study (an average of 2.4 times per task) and 391 times during the field study (an average of 2.61 times per task).

When examining the task names and descriptions of the tasks that participants named as multisession, we found similar themes appearing in both studies. We were able to classify eight general categories (MacKay & Watters, 2008): school work (67 of multi-session tasks), general topic search (66), research (25), travel/tourism (21), projects (19), action-based (18), shopping (14), and status checking (5). Several of these tasks contained sub-tasks that could be classified according to the web task categories of Kellar et al., (2007): information gathering (228 of the sub-tasks), fact finding tasks (119), transactions (46), communication (21), and maintenance (14). Browsing was not included as it is considered a serendipitous act (Kellar et al., 2007) not performed for the specific purpose of the multi-session task.

Multi-session Task Sample

Table 1 presents a ‘school work’ multi-session task to demonstrate how varied and complex these tasks can be. The task end date was based on a deadline, and the participant worked on the task three times over four days. When examining the log data for this task, it showed that the participant would start at a main page (such as, the computer science home page or www.google.com to perform a search), ‘follow links’ off the page and then ‘go back’ to the main page to follow a link to a new page.

The customized browser could be installed on more than one machine and participants could access their multi-session tasks from any of the installed browsers. This participant installed the browser on two different computers: one at the library and one at their workplace. They worked on the first session in the library for 28 minutes and worked on the second and third session at their office. The second session had a gap about half way through the session where no actions or tools were logged by the browser. While the second session started at 7:00pm, the last session was conducted during the middle of the day and the gaps in the logged data during this session (46 minutes and 29 minutes) may indicate that the participant was interrupted with other ‘work’.

Table 1. A ‘School Work’ multi-session task recorded in the field study
Task NameAssignment for Computer Networks
DescriptionAn assignment of Adv Topics in Computer Networks related to Wireless Technology
Task TypeSchool Work
Task Length4 days
Num. of Visits Time of Sessions3 visits
Session 1Session 2Session 3
28 min.126 min.127 min.
LocationLibraryOfficeOffice
Task End ReasonDeadline
Sub-tasksflow control techniques, assignment 3 for wireless networks, and assignment of computer networks (one per session)
Other Applications UsedWord (3 times) and email (3 times)
Tools/Actions logged in browserFollowed link (125 times), go back (66), submitted form (e.g., submit search terms, 14 times), user typed URL (12), opened new tab (5), copy to (3), paste from (1), opened bookmark (1)

Multi-tasking within Sessions

We had anticipated that while participants were working on their multi-session tasks that they would also be switching at times to other tasks. We were unable to capture this in the diary study, but we designed the logger in the custom browser to capture multi-tasking within multisession tasks. In order to explore how users were multi-tasking we asked participants when they were working on a multi-session task to indicate on the interactive toolbar if a particular web page in a tab or window was a part of the active task or not (see Figure 1). By default, whenever a participant started a multi-session task their current tab or window was automatically ‘included’ into the task and any tabs or windows opened from an included page was also considered included. Other tabs/windows opened previous to the start of a multi-session task were considered to be excluded unless explicitly denoted as ‘included’ by the participant.

By having the participant indicate include/excluded tabs/windows we were able to tell when a user switched out of the active multi-session task (by ‘excluding’ a particular page or by returning to a page that was already excluded). As well we could also track when they returned to the active multi-session task (by either ‘including’ an excluded page or by returning to an ‘included’ page from an excluded one). In total, we logged 627 task switching actions (MacKay & Watters, 2008) during the field study.

Reasons for Tasks that require Multiple Sessions

In both studies, we asked the participants to record why they stopped working on their multisession task for that session to help in understanding why these tasks spanned several sessions. Participants gave several reasons why they stopped and many of the reasons overlapped. Based on the responses we were, however, able to categorize the reasons (see Table 2).

The two most common reasons given were that the participant finished what they wanted to accomplish during the session (e.g., they found the information that they needed) or that the task itself was finished. Other reasons included that they needed a break or that they needed to do other things (e.g., have supper), that they were tired, that they needed sleep or were done for the day, that they were physically going somewhere (e.g., going to work, a meeting or class), that they were waiting on some information that was pertinent to the task (such as waiting for better price offers or confirmation emails), or that they needed to do further work on the task at a later time.

Table 2. Participants reasons to stop working on a multi-session task
Reasons to StopDiary StudyField StudyTotal
Finished sub-task6634100
Task Completed168399
Take a break/do other things135265
Sleep/tired/done for the day144054
Physically going somewhere142135
Work on something else112334
Wait on information/status checking20323
Need to do more work13417
Problems with computer01313
Need time to decide707
Got/Get information from another source325
Ran out of time044
Frustrated/overwhelmed303
Bored123
Couldn't find information123
Couldn't complete task022
Don't have all the information011

There were some categories that were unique to the diary study and others unique to the field study. For example, in the diary study some participants recorded that they needed to make a decision before they would continue the task. In the field study, some participants indicated that they had computer problems (e.g., that their laptop battery was low, or that the internet was down).

Keeping Information between Sessions

In the diary study, participants could indicate for each session working on a multi-session task which tools and actions that they did. In the field study logs, we were able to capture not only when participants used a browser tool or performed a particular action during a session but the actual number of times for both multi-session tasks and all other web activities. Table 3 summarizes the actions/tools recorded in the diary study and the actual counts inform the logs in the field study.

Table 3. Browser Tools/Actions used in both studies
Browser Tool/ActionDiary StudyField Study
MS tasksOther Web ActivitiesTotal Counts
 ##% of Total#% of Total#
Bookmark Actions5835722%127478%1631
  Opened BM2625418%118782%1441
  Created BM238555%6945%154
  Created BM Folder92100%00%2
  Managed BM1647%1853%34
Use History161248%1352%25
Print Page171338%2162%34
Save Page18529%1271%17
Find in Page376237%10763%169
Look for next in Pagen/a32256%25544%577
Copy/paste Actions4050247%55953%1061
  Copy 39051%37349%763
  Cut 318%1482%17
  Paste 10939%17261%281

Specifically these actions and tools represent those used to help participants keep information between web sessions. Similar to other research (Abrams et al., 1998; Aula et al., 2005; Bruce et al., 2004; Kellar et al., 2007; Tauscher & Greenberg, 1997), our studies showed that the popular tools for keeping and revisiting information between sessions are bookmarks and to a lesser extent history lists. Participants recorded using these tools in both the diary and field study. Diary participants recorded using bookmarks 58 times (26 times to open a bookmark, 23 times to add a bookmark, and 9 times to create a bookmark folder) and recorded using their history list 16 times. While the logged use of bookmarks and related actions are much higher in the field study, the use of history lists is actually less for field study participants and their multi-session tasks.

During the field study, participants opened more bookmarks while performing other web activities (1187 times) compared to multi-session tasks (254). However, more bookmarks were created for multi-session tasks (85 times) than other web activities (69) and participants managed their bookmarks for both multi-session tasks and other web activities almost equally (16 and 18 respectively). History lists were used less often than bookmarks with only 12 counts for multisession tasks and 13 for other web activities.

While bookmarking actions were reported in the diary study and logged during the field study, when participants were asked about bookmark use in the post-study interviews, the responses were mixed. About half of the participants from both studies regularly use bookmarks. Other participants either didn't use bookmarks in general or for multi-session tasks specifically. One participant stated with respect to the use of bookmarks that “… I don't put a lot of weight on bookmarks – dead links need to be organized, can be difficult to manage, not a good organizational system…”. Others felt that multi-session tasks were temporary and that they did not want to clutter their bookmark list with short lived bookmarks.

In terms of copying and pasting information, we only captured ‘paste to’ and ‘paste from’ in the diary study to represent all the actions (cut, copy and paste). The number of ‘pastes’ in the diary study were high in comparison to the other actions recorded and we can see that in the field study that participants relied heavily on cut or copy/paste actions for multi-session tasks. The copy action was most often used, that would seem to indicate that participants more often copied something in their browser (e.g., web page information) to an outside source. Each time that a participant performed a cut, copy or paste a popup dialog box would appear in the browser asking the participant to record where they were copying to and where they pasted from as the logger was unable to track actions outside of the browser. Participants could choose to type this additional information or to ignore the request. We were able to gather some responses such as “Paste from a website”, “Paste from word”, “Copy to website”, “Copy to refworks”, and “Copy to search engine”.

Other tools that participants recorded using in both studies to keep information for future sessions was printing and saving web pages. Participants performed finds within pages to locate specific information within a page both to identify new information and to refind information within a previously seen web page. In the field study the ‘look for next’ action was performed more frequently for multi-session tasks (322 times) than for other web activities.

Post-study Interviews

During the post-study interviews, we asked participants whether they kept information that they found from their multi-session tasks for use in possible future tasks and if they did, how they saved it. Most participants (20 out of the 46 participants) replied that it depended on the task and the type of information whether they saved the information for a later task. One participant said “… it depends on the task. For example if I buy online I may put receipt in a PDF file and store it”.

Thirteen participants said that they keep the information although three clarified that this was for a set period of time (e.g., to keep a bibliography for a semester). Seven participants commented that they keep some of the information from a task and six said that they never keep the information. The most common way that participants said that they kept information was to save the web data on their computer. They also cut and pasted information into an electronic document form (e.g., Word or Notepad) which was further supported by the field study logs. Others used bookmarks, although some noted that they worried that saving pages cluttered their bookmark list.

During the post-study interviews, we demonstrated up to five browser tools that we thought might be useful when working on multi-session tasks. This demonstration of the tools also prompted participants into a discussion on the types of tools or features of tools that they thought would be helpful in performing multi-session tasks. The tools we demonstrated included tabs (to diary study participants who had never used tabs), session saver, landmarks, page annotations and Google Notebook (www.google.com/notebook) to field study participants only.

We demonstrated tabs before the official release of the most recent version of Explorer which includes tabs. While half the diary participants used Explorer as their main browser, only six of these had never used tabs. After demonstrating tabs (on Firefox), four of the six believed that tabs would be useful for multi-session tasks. One participant commented that it would be “…especially [useful] if doing compare and contrast activities”. Of the two other participants who had never used tabs, one was unsure if tabs would be useful for multi-session tasks, but felt that they would need time to get used to using tabs before making a final decision. The other felt that they would still prefer to cut and paste into another document.

Session saver is a Firefox extension that saves all the active tabs when the browser is closed and reloads the tabs when the browser is restarted. Overall, participants were receptive to session saver (36 out of 46 indicated that they would or might use it). Participants thought it would be useful for large tasks with lots of web pages that lasted longer than a day, although it was also noted that it won't return you to the exact state where you left off and that not all tabs saved would necessarily be relevant to the task.

Landmarks (MacKay et al., 2005) allow users to highlight information within a web page that are then saved similar to a bookmark except when the web page opens again it automatically scrolls to the user selected information (text or images). Participants liked this tool (34 of the 46 indicated that they would use the tool, and 11 mentioned that would use it sometimes). Participants thought it would be useful to refind information in long pages (less searching within pages and scrolling), to mark several landmarks within a page, and while some thought it would clutter up their bookmark lists, more participants believed it would them help remember why a page was saved.

For page annotations we demonstrated MyStickies (www. mystickies.com). This application allows users to add notes within web pages that look similar to paper ‘post-it’ notes that are organized on a main page for easy viewing of the notes. This was the least popular tool with participants. Eleven of the 46 participants indicated that they would not use it and 21 participants said maybe they would use it but infrequently. Participants commented that it would be good for a large research project to help organize the information and for comparing information, but many thought that it seemed like a lot of work and some were leery of using an outside site to store their personal information.

With Google Notebook users can easily add snippets of different web pages into their own ‘notebook’ and can make their notebook available for others to add to it and search on it. Many participants thought that this would be a good tool for a group activity or collaboration, and liked that it kept everything in one place, but many did not think it would be very useful for multisession tasks. Six of the 24 field study participants indicated that they would use it and 12 thought they might use it.

Suggestions & Browser Tool Prototypes

Participants made several suggestions for helpful features for browser tools. Suggestions from participants ranged from different bookmark improvements (e.g., have expiry dates attached to bookmarks) to having a task manager. Surprisingly, field study participants often commented during the final interview how they liked the interactive toolbar in the customized browser to help them with their multi-session tasks. There were three common features that participants most often mentioned: to keep a list of current multi-session tasks, to be reminded of the current multi-session task that they are working on during the session (to help with multi-tasking) and to help manage and organize information between sessions for the task that is more useful than the common tools (e.g., bookmarks).

Using these suggestions we have developed three prototypes that contain different levels of features to help users cope with information that they gather during web sessions and to keep track of the information between the sessions. Keeping the comments of the participants in mind, we designed the prototypes tools with an interactive toolbar similar to that used in the field study with some of the same features (e.g., highlighting the active multi-session task name in yellow). All the prototype tools include a logging feature that will allow us to record all the actions and tools that the participants use while they evaluate the prototype tools.

Prototype 1

Prototype 1 simply keeps track of active multi-session tasks and reminds users when they are currently working on a multi-session task during a session. When the user is working on a multisession task, the name of the task is displayed in the toolbar highlighted in yellow. Between sessions, the user can access their list of active multi-session tasks through a drop down box (see Figure 2).

Figure 2.

Drop down box that lists all active multi-session tasks.

Prototype 2

Prototype 2 (Figure 3) contains the same features as the first prototype with an additional function that allows the user to include and exclude web pages from the active multi-session task. The included tabs are grouped together and are also highlighted yellow like the task name. It also saves the web pages that the user indicates are included in the multi-session tasks and these are automatically re-opened when the user resumes the task in a later session.

Figure 3.

The current multi-session task and included tabs are highlighted in yellow. The included tabs are also grouped together. In the next session the included web pages will be reopened. The buttons that reference ‘Log’ are used only for testing (to view the log, delete URLs from the log and submit the log.

Prototype 3

Prototype 3 contains all the features of the first two prototypes with supplementary features to further help users in resuming their task. In addition to just saving the ‘included’ web pages, it allows the user to manage which pages to save for later sessions. When the user stops an active task a Save Web Pages popup box (Figure 4) appears that lists all the web pages that are currently open in the browser. It has checks beside those pages that have been ‘included’ in the task. The user can save the automatically checked pages or they can unselect checked pages and/or check unselected pages. It also automatically landmarks the saved web pages so that the user will be returned to the exact location in the page where they left off when the pages are reopened in a future session. Prototype 3 includes a new feature that helps the user to manage the saved web pages when the multi-session task is finished. They have the option to delete the saved pages or to collect all of the pages as a bookmark.

Figure 4.

The Save Web Pages popup box appears when the user Stops the task. It lists the web pages associated with that task with ‘included’ pages already checked. However, users can chose to select and unselect any pages that they want saved for later sessions.

We are currently in the process of evaluating all the prototypes with participants in a field study.

Conclusions

When people perform multi-session tasks they need to make sense and organize large amounts of information to help perform the task over a period of time. They also need to keep a sense of where they left the task off each time that they restart or continue the task. New browser tools should consider these requirements as part of their design, in particular for tasks that may require multiple sessions in order to better support people when they perform these tasks.

Ancillary