Accelerating Alfresco search experience with sunburst controls

Context of the challenge

As the subject of his thesis at KU Leuven, Martijn Millecamp investigated new ways to improve the Alfresco search experience. A number of options were evaluated, and user tests helped us to pick a sunburst representation as very helpful and productive. Traditionally, facets are represented in lists, but leveraging the graphical qualities and ease of use of a sunburst makes Alfresco search more enjoyable and productive.


Nowadays, more and more people spend an increasing amount of time in front of their computer screens. In addition, they spend more and more of this time organizing and searching for files. A user of alfresco is no different. Due to the rise of the internet, digital newspapers, digital cameras, smartphones, and other technological gadgets, we are bombarded by a variety of digital documents. This abundance of documents has made it much more difficult for users to retrieve the files they need. A user-friendly interface can really help people to find their files more efficiently.

Search strategy

Research has shown that computer users search for files using two different strategies: teleporting and orientation.

Teleporting is defined as searching for a document in a single step. In search interfaces, this strategy is often supported by a search box, which allows users to enter a search term and obtain a list of documents that may be relevant to their search.

Orientation, on the other hand, involves several steps. A concrete example of this search strategy is a faceted search: users apply different filters to create a long binary query, which allows them to reduce the subset of results in an iterative manner. In modern search interfaces, this is usually supported by a drop-down menu with filters on the left side of the screen. Users can then open these drop-down menus and select a filter to narrow down the results.

Although teleporting is still supported more commonly than orientation, research shows that users prefer orientation over teleporting, since orientation results in a lower cognitive load. This is because users don’t have to know all the properties of the file in advance, but can remember additional elements while using the filters. In a systematic search users are also more aware of the context of the file and, consequently, more confident about the results.




A first attempt to support users with a new visualization technique was to present files in a graph. Each document was displayed as a page, and the folder in which they were stored was the parent node. This gives users a quick overview of the hierarchical structure. When they hover the mouse pointer over a node, a link to similar documents or documents that were created around the same time appears. To highlight the time properties of a document, the first interface provided a timeline, allowing the user to select all the documents from a specific period of time (e.g. a day, a week, a month or even a year). The different filters were also visualized in graphs, as shown in the figure below.

Evaluation & Results

This first paper prototype was evaluated by five fellow students, who were all following a different course programme. The evaluation mainly showed that a graph can become very complicated, especially if there is a large number of documents. However, all users mentioned that searching with filters feels very natural and matches their way of thinking.



Since graphs are difficult to apply to large collections of documents, we implemented a second prototype. This digital prototype was a 3D globe on which documents were visualized. A 3D interface has the advantage that more documents can be visualized in a clear way. However, research has shown that navigation in such an interface is often confusing and not very user-friendly.


To support easy 3D navigation, we decided to use a globe interface. All users are familiar with this kind of navigation, and can always find out where they are. Similar documents can be clustered and combined in a country or continent. This technique is sometimes referred to as spatial-semantic mapping and has been proven to make searching easier. During a search, a bar is placed on top of each document and the height of the bar shows the document’s relevance to the search.

Evaluation & Results

This digital prototype was evaluated by a data visualization expert, who reported that, while navigating the globe was relatively easy, it was difficult to search for documents, because there is no point where you have an overview of the entire globe, and you can never see all files together. This makes it very difficult to compare the height of the bars or even to spot a relevant file (because there may be a relevant document on the other side of the globe).


sunburst final version

Motivation & design

Since the 3D interface was not very successful and graphs can quickly become complicated, we increased our focus on supporting the search process in the next visualization. In this iteration we focus on visualizing the different filters and no longer on any other visualization of the files. To visualize files, we return to the traditional list, since all users are familiar with it.

For the visualization of the filters, we use a space-filling technique called sunburst. This technique is already applied in existing applications to visualize hard disks (DaisyDisk and Baobab).


We also organized and visualized the filters hierarchically in a sunburst in the same way. When you click on a segment, the segment will ‘unfold’, so the segments under the selection will cover the entire circle. To support the users, selections are also displayed as buttons at the bottom of the list of results. Users can click these buttons or click in the middle of the sunburst to undo their selection. You will find an example in the figure below: ‘Image’ is selected and can be unselected by clicking in 1 of the 2 frames.

The number of files in a segment is visualized by the angle of the segment (for example, the image shows that there are more .png files than .jpg files). For more details, users can hover over a segment to display a tool tip that shows the exact number of documents. This way, users know how many documents meet the selection criteria before they even select a segment (this is called a ‘query preview’). This also means that users will never have an empty result list.


Evaluation & Results

The evaluation of the design comprised three parts, in which 10, 5 and 12 people took part respectively. Users had to perform specific tasks during each evaluation. In the first part, we asked the users to perform the tasks as quickly as possible. In the second part, we asked the users to think aloud to give researchers a better understanding of the test users’ thoughts. After these tasks, the users completed a SUS questionnaire. This 10-question survey measures the user-friendliness of an interface and can be used to compare the different prototypes.

In the first evaluation, one prototype was tested to establish if a sunburst prototype is an intuitive and user-friendly visualization technology. Both the questionnaire and the results of the thinking aloud exercise suggest that this technique is not only very attractive from a visual point of view, but also easy to use and intuitive.

In the second evaluation we adapted the interface slightly and compared it to another prototype that uses a standard drop-down menu to select filters. This interface is shown below. 

This evaluation mainly focused on the time that was required to solve a similar task in both prototypes. The results showed that users were able to find files more quickly in the sunburst interface than in the more traditional interface. Additionally, the results of the questionnaire show that this interface is not only faster, but also more user-friendly! (average of 65.2 compared to 93.2)

In the third and final evaluation, we compared the interface presented at the beginning of this chapter with the traditional Fred (Alfred Desktop) interface. In this evaluation we mainly wanted to test the graphic approach and compare it to an existing commercial search application. We measured the time a user spent on completing a task, but also the number of clicks and the number of documents a user had obtained after filtering. Our prototype outperformed Fred across all categories. Logically, the results of the SUS questionnaire also show that our prototype is more user-friendly than Fred. During the thinking aloud test and the subsequent semi-structured interview, users stated that this was mainly related to the more graphic approach. By making the filters more visually attractive, they are found and used more quickly, while users need fewer clicks to find the files they need.

One critical remark that has to be considered when interpreting these results is that Fred supports more functionality than our own prototype. This is certainly one of the reasons why our own prototype is faster and more user-friendly than Fred.

Add new comment

The content of this field is kept private and will not be shown publicly.