DormMate - GR6 - User Testing

Design    

The main page is the default landing page for users. It's main use is to keep track of the current active tasks.

Figure 1: An overall view of the main page.

Users are able to receive what is essentially a news feed of recent changes by looking at the "Recent Activity" column on the left. On the right, users may look up specific tasks and edit them as necessary.

Figure 2: When a task row is clicked, additional information about the task is displayed.

Figure 3: A close up of the notifications section.

Figure 4: After a task is created or edited on the Task Creation Page, that task will be highlight with a border for 5 seconds upon returning to the main page.

Figure 5: Showing off the possible filters that can be used.

Implementation

Design:

The form contains standard information that the dorm manager need to fill out when a resident moves in or moves out of the dorm. There are two main tasks including Move In and Move Out. The information on the page changes dynamically depending on the type of the task. This page will be visited very frequently by the user, so it's designed to be efficient for filling out information. The Useful information panel will contains auto-fill information that help the user speed up the filling out task. One of the important decision for this page is to fit the two panels into two scroll views without overflow the information to the whole page. The user can choose to scroll each panel separately without moving the whole window. It's because we want to user to make good use of the auto-fill function, so they can scroll the left panel up and down while keeping the right panel statically in the view port.
Figure 1: The start new task page.
One of the importance decision that effect the design is the affordances of this right panel. Many users mentioned that there wasn't enough affordances to show the auto-fill function of the information on the right. Therefore, we have decided to put a place holder on the input where the auto-fill information supposed to fill in when the user hovers over those information. Figure 2 shows the result of the auto-fill functionality. We chose this as our implementation for its simplicity over other methods such as having a tutorial pop-up or changing color or boxing around the auto-fill result.
Figure 2: Auto-fill information when hover over item.
Another important feature is when the user submits the form, if there are some missing fields, the user will receive a pop up noticing them about these fields. The user can have the option to go back and fill out those missing fields or save the form any way. Figure 3 demonstrates this functionality. This confirmation dialog will help prevent user from making mistake when submitting form. We choose this decision instead of preventing the user from submitting the form at all and have the missing fields highlighted at the beginning of the form because the user might want to leave those fields blank to fill in later. In addition, different dorms might have different requirement for required field, so providing them with confirmation dialog will help them decide on their own which field they want to keep.
Figure 3: Missing fields notification when submitting the form
In order to follow the CRUD model presented in class, we decided to let the user edit the task. The page will be pre-filled with the most recent updated information. Figure 4a and 4b demonstrates the edit task functionality of the website.

 
---------------> 
Figure 4: Edit Task functionality

Implementation

This page follow the model view control (MVC) model. The database for MIT directory search is kept in a separate javascript file since this field is static, and the information of the resident is stored on the Parse database since it will be dynamically changed over time. We utilize the convenience of Jquery UI in creating dialog and drop-down menu. We also make use of twitter bootstrap for organizing the layout. All of the controller for listening to action are added separately in order to follow the MVC model and to keep the code modular.

In addition, since there are different information presented for each type of task (Move In versus Move Out), we decided to dynamically repaint the page instead of reload the page from scratch. This will help the user preserve some of the information that are shared between these two tasks.

The information about vacant room are update dynamically through parse. For example, if a person moves into a room, we update the current list of resident of that room and remove it from the list of the available room.

MIT directory is design to match with the current MIT directory on web.mit.edu/people to help user with recognition because most users are familiar with this interface already. After the information is generated, the action listener will be attach to the result to auto-fill the form.

The list of suggested tasks are static data collected from the suggestion of the users. We also decided to add the default task button to speed up the filling process.

The back end of this page is supported by Parse. There are three objects that currently present on the server: Resident, Task, and Room (See Figure 5). Each of this object will have a reference to the other objects so that whenever one object changes, it will also change the other. After the user submitting the task to the server, a task object will be created with each attribute corresponding to a field in the form. Then, the room and the resident will also be updated on the server to reflect the task i.e if a resident Move In to a room, the room becomes occupied and remove from the available room and the resident gets added to the system.

For the updated task, we also have to make sure that we do not created duplicate object on the server since Parse doesn't automatically check for that.

Since there are three main pages sharing the same source of information, we need to make sure that changes on one page will be reflected on other pages too. For example, if a person moves out, the main page will show that as a pending task and the floor plan also have to indicate that the room is now vacant. All of the information uploading to and downloading from the server are text-based, so the retrieving time is almost instantaneous which enables the user to see the current update from the server at no time. In addition, to respect the anonymity of 6.813, we also use fake data for our database which introduce some inconsistency among information. However, this doesn't present any usability problem for the user.
Figure 5: The current state of the parse server.

Design

The Floor Plan page was designed to address an issue that dorm management staff had noted about having difficulty determining positioning information about rooms on both macro- and micro-scales. As shown here, the page includes the navigation bar present throughout our entire interface for consistency, safety, and user control; the rest of the page is used for addressing the problem just mentioned. In the center we placed a large image of the floor for visibility, and because it is the most important part of this section of our interface - on the right side of that we horizontally aligned tables of useful on-hand information; on the left we placed extra UI elements (also horizontally aligned) for helping to navigate to other parts of the building besides the current floor; and above we placed a search bar for helping to navigate the floor plan.

The choices for alignment were mainly for the sake of aesthetics, since separation of the UI elements already seemed clear to us and users; however, positioning was important. We placed the search bar on top so that it'd be very visible, static on-hand information on the right since it fell into the same category and would be easer to process, and the floor switcher on the left since we wanted it to be both visible and because it was dynamic content.

Figure 1: A screenshot of the entire floor plan page. You can see each of the elements described on the image.

The image in the center shows current floor of the floor plan of the dorm being managed*;* it is annotated with colored labels (described with the legend) so that the user can gather information from quick glances as color is a good visual differentiator. Because floor plans might be large and difficult to understand at a glance, we also made sure to have the image be both pannable and zoomable, so that the user could examine particular areas of a floor more carefully if they wanted; we decided to make the interface work similarly to Google maps both for the sake of external consistency and efficiency (shoutout to Google). The zooming is done by moving the slider on the top left of the image, and the panning is done by either holding the mouse down over the arrows in the top left, or by using the mouse to do real-time panning.

Figure 2: A closeup of the image of the pannable, zoomable floor plan.

In addition to the other wonderful features of the floor plan we've already discussed, we also made the floor plan clickable, with click events over rooms generating dialog boxes (top left corner on center of room) displaying information about the room. The current interface had information only about the room number and the residents staying in the room, but it is likely in the future more relevant information will be included to help the management staff with making rooming decisions. From the dialog, a user can create a new task by clicking on the large button present on the dialog, and can also exit out of the dialog by clicking on the X-button. The dialog was created this way for external consistency; for safety, we made sure that multiple dialog boxes could not pop up on the screen and crowd each other - we also made sure that the dialog scales and moves with the image when panned or zoomed.

On the right you can see a floor-switcher; we created this piece of dynamic content for the sake of helping users navigate floors in a floor plan (as indicated by the name), since we felt that there was only space to put one floor in the viewport at a time. The selected floor is highlighted differently, and each floor is represented by a text box which has the affordances of a button, for external consistency. We chose the colors for the sake of unity with the rest of the interface, as well as because the blues we chose for selection, highlighting, and deselection were all discernible. 

Figure 3: Screenshots of the dialog which pops up on clicking the floor plan (left) and the floor switcher (right).

So that the user would be able to discern the information provided by the color labels on the floor plan, we have a large legend on the right side - since color doesn't inherently indicate any meaning in floor plans or our context, our heuristic evaluators and users found that something like this was necessary (information scent).

For efficiency, we also included the floor statistics table - horizontally and vertically aligned with a large header and line width mostly for the sake of aesthetics and readability. We included the floor statistics table below the legend due to information scent - when viewers see the necessary legend, their eyes easily move to the table below.

Figure 4: Zoomed screenshots of the legend and floor statistics table.

Lastly, we have a search bar on this page for helping the user navigate the floor image efficiently. The search bar has the typical affordances and benefits of search bars, like autocomplete for safety, as well as border highlighting to indicate selection. The user can search by room number, or name of resident - on selection, the floor plan automatically pans so that the room selected (or room of the resident) is close to the top left corner of the image (next to borders so it's visible), and also pops out the dialog for the room (for visibility of the changes).

Figure 5: The search bar for the floor plan.

Implementation

Because the UI elements on this page are fairly simple, they were made almost exclusively with personal HTML and CSS, though the navigation bar, as with the other pages, was made with Bootstrap, and autocomplete was done with JQueryUI. All dynamic content was implemented with JavaScript, with JQuery being the only external library used.

Upon opening, the page stores all the data on rooms after calling the Parse backend, and stores it in an array; at this point, all arrays and objects needed for operations like searching and simple looping are created by looping through this room array and manipulating the information stored in it. After this has been done, all operations which might need to be performed on the page are available; before this, nothing can be done (for the sake of safety, otherwise error messages get thrown). We decided that doing a large amount of local processing of the data obtained from the backend for convenience was better because it was faster than querying the Parse server.

The visibility of the floor plan, along with its color labeling, and the ability to pan and zoom it were implemented using two canvases - one was used as an invisible buffer, and the other just drew the image in the other canvas with the appropriate transformation based on the continually-updated scaling and panning parameters. This was required for appropriate efficiency, since otherwise there would be large amounts of clearing and stroke-level redrawing being done extremely often, which would be visibly laggy. The parameters were updated with mouse press events. Mouse clicks had listeners which checked the coordinates of the click, transformed them, and compared them to the coordinates of other rooms to see which room the click corresponded to - this then helped to generate the dialog.

The floor switcher was just implemented in HTML and CSS as a list of buttons, with JQuery click events being registered by the system and changing the floor image rendered in the viewport, as well as the arrays used to generate the labels for the rooms. CSS was used to ensure it didn't have the same look as normal button HTML elements, since switchers typically have a different look or feel. 

The legend was implemented as another Canvas, for the reason that drawing the example color labels with the HTML5 Canvas was easy and still very efficient.

The floor statistics UI element was just implemented as an HTML5 Table, since the structure lent itself well to the function of our table (rows, columns, etc.). It was also straightforward to dynamically change the table with floor switches and on loading (since floor statistics depended on the stored data) with JQuery by using class and id selectors with the table being a pure personally-written HTML element.

Lastly, the search was implemented using JQueryUI's Autocomplete. This was done for the ease of the syntax, because it seemed robust, and because it gave all the control we desired in selection from the stored data.

Task 1: Document Completion of Subtask

• Background: Frances Phillips is in the process of moving out of the dorm. He has just returned his key to you.
• Task Detail: Update the system appropriately.

Task 2: Determine Information about a Floor

• Background: The house master is curious about how many vacant rooms there are on the second floor.
• Task Detail: Find a way to look up number of vacant rooms on the 2^nd floor.

Task 3: Start Moving a Student In

• Background: Tina Sullivan has requested to move into Next House on August 30^th, 2013. You want to begin the move in process for her. Additionally, she tells you that she wants to live somewhere on the 2^nd floor. Make sure all the fields are filled in and assume default subtasks.
• Task Detail: Make a new task for Tina Sullivan

Task 4: Start Moving a Student Out

• Background: Andrea Reyes wants to move out of Next House. She says she expect to move out on May 23^rd, 2013. You want to begin the move out process. Make sure all the fields are filled in and assume default subtasks.
• Task Detail: Make a new task for Andrea Reyes

Task 5: Identify Resident

• Task Detail: Who currently lives in room 266?