Dasher

Guide: Dr. Alan Blackwell | Research Internship | Computer Laboratory, University of Cambridge, UK | June-July 2006
Mobile Research

Dasher is a novel probabilistic text entry interface, using continuous pointing gesture through structured information in a zooming interface. It is widely used by disabled people around the world. It's developed by the Inference Group Cavendish Laboratory University of Cambridge. For downloading Dasher and to know more about it please visit: http://www.inference.phy.cam.ac.uk/dasher/

I redesigned the color mapping for the Dasher UI for English language. My approach was to treat color as an integral part, and such come up with a scheme that makes the Dasher more efficient, reducing the learning required new users, and making it more usable with colours that can work for extensive use.

Important Characteristics

From heuristic evaluation and literature reading I concluded that critical elements and characteristics for the Dasher are:

Information Structure

The information structure is primarily represented through the placement of entities on the screen. I wanted to complement it visually with colour groupings, introducing an implicit way of differentiation for the groups that helps us getting around the short-term memory limitation. The implicit memory is state independent and even the people with amnesia are found to have sound implicit memory.

On left hand side Information Organization is shown on a structural level and the same structure is than represented using colours for visual perception of the information chunks and their levels.

Survey

In the current design colour association was arbitrary. With an objective to find people's perception about the association between colour and character-set, I conducted a survey. The goal was to design a colour mapping, leveraging the finding of the survey, to reduce the cognitive processing required during the fast navigation to choose a character-set in Dasher.

In the survey, I found a very strong correlation between Blue and Numerals, and Black and Capita-Letters. I used the survey response as the basis for deciding group color for each of the character-set.

Designed Color Scheme

Here is the conceptual representation of the proposed colour theme.

All the possible combinations of the foreground and background colours int the proposed colour theme were explored to adjust the perceptible difference between elements.

Observation, Identification and Differentiation

In this dynamic way of entering text one goes through continuous process of thinking about the characters they want to write and then trying to identifying them and differentiating them from others. This is a continuous elimination and selection process where one encounters different elements each having their own identity.

The identity of these elements is always in the system and not alone, so it always interacts with other elements. It can be similar to some elements and at the same time different from others in terms of its characteristics. How strong should the identity be in order to get register depends upon the response threshold limit and if we make it stronger than that threshold value, it may interfere with the others in the system and would demand more attention span from the user which will decrease the efficiency.

For navigation to a character one first has to go to the group to which it belongs. If user makes mistake in choosing a correct group it would lead to a substantial time delay in rectifying the mistake, so the entities here have to be very well different and identifiable. Once the user has chosen a right group next is the task of identifying the particular character. Now, where the user look for the character with in the group (up, down, middle) depends upon the knowledge of the information structure and then how much the particular character is different from its neighborhood elements.

Visual Differentiation in Navigation

Use of slight gradation change in the set of nested boxes with in the two consecutive boxes gives a visual feedback to the user to stay inside the intended box itself and not move to the adjacent box, by creating a Visual Boundary.

Visual Flow

The process of text entry is an extended process, so in the process user come across different coloured nested boxes presenting before user a continuously state changing environment.

The pattern that these boxes form over a long duration is also very important. A pattern with less drastic colour shifts would be less tiring for the eyes and thus would be preferred over the one with frequent drastic shifts in colour gradation, as the user uses the system more often. Also, if the pattern that the coloured boxes form can be related to the actual written text would give user a subconscious feedback about what they are writing. And this association would improve the efficiency over a period of time.

Here are the pattern of the coloured boxes through which navigation takes place for writing "Hello, how are you?" in both the current and proposed colour scheme.

Overall Visual Experience

Our system has an environment, which is always changing and having a lot of identities and elements at all the times. This put a lot of pressure on the number and type of colours used. More the colours, more will be the entities on the screen, increasing the cognitive load associated with interface.

To provide user with a more pleasant and soothing experience, use of warm colours' (yellow and red) is minimized and they are used only with the groups which occur less frequently. Fortunately, the survey data also aligns with this proposition. Also, to achieve a more consistent colour scheme, I opted for the similar hues, wherever possible and appropriate.

Here are the screen shots of the existing and the proposed colour theme. Compare the images and see for the one with the more pleasant and harmonic identities (coloured boxes) when seen as a whole group.