12.01.2015 Views

zmWmQs

zmWmQs

zmWmQs

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

A Platform that Integrates Quizzes into Videos<br />

Robin Woll, Sven Buschbeck, Tino Steffens, Pascal Berrang, Jörn Loviscach<br />

For reasons of wide access, privacy, and embeddability,<br />

the platform does not require learners to regis-ter and<br />

log in. In addition, by default no personal data of users<br />

are saved on the servers; server-side statistics are only<br />

stored cumulatively so that one can compute the mean<br />

and the standard deviation of parameters over all users.<br />

To gain insight into existing models for the use of quizzes<br />

and the interaction with them, we looked at existing related<br />

solutions. Among the major MOOC providers, Udacity<br />

comes closest to the look of our plat-form, as it often uses<br />

the final image of a video as a backdrop for HTML radio<br />

buttons, checkboxes and text input fields in an ensuing<br />

quiz. Outside the MOOCs sphere, there are several solutions<br />

to create rigidly tem-platized quizzes, such as You-<br />

Tube’s Video Questions Beta, educanon.com, ed.ted.com,<br />

zaption.com, and the video editor that is part of the screen<br />

recording software TechSmith Camtasia. The latter is also<br />

noteworthy be-cause the generated Flash files (which admittedly<br />

represent an outdated format) can communicate<br />

the quiz re-sults to a learning management system using<br />

the SCORM standard. This type of data transfer is of future<br />

inter-est for embeddable learning objects served by<br />

our platform. In a similar vein, the huge range of “multimedia”<br />

quiz types offered by learningapps.org comes close to<br />

our aim of embeddable units. Further inspiration stems<br />

from HTML5 animation editors such as Adobe Edge Animate,<br />

Google Web Designer, Mozilla PopcornMaker, and<br />

Tumult Hype.<br />

Functions of our Platform<br />

Given that the videos we want to equip with quizzes<br />

are hosted on YouTube, the obvious solution for putting<br />

quizzes as layers on to top of the videos was to use the<br />

YouTube API in combination with CSS3 to control which<br />

element is placed on which layer. We wanted to keep the<br />

design and structure of the quizzes as flexible as possible,<br />

so they are encoded in plain HTML. This provides the quiz<br />

author with a rich functionality. For in-stance, CSS3 allows<br />

defining animations, which are frowned upon by many, but<br />

– when used judiciously – can for instance counteract the<br />

change blindness that occurs when a small text field is displayed<br />

for the user to enter the result of computation that<br />

is going on in the video, see Figure 1. Other HTML elements<br />

include lines such as those shown in Figure 2.<br />

Figure 1. The quiz asks the user to complete a computation before<br />

the lecturer does so.<br />

Figure 2. This quiz draws four lines as HTML code over the video.<br />

The user is asked to click on the one with the specified slope.<br />

As we are focusing on mathematics and related subjects,<br />

we found it helpful to also build a domain specific lan-guage<br />

(DSL) to evaluate a string entered by the user in a text field<br />

as a symbolic mathematical expression. This enables the<br />

author to create much deeper tests. Our DSL is basically a<br />

Java port of an interpreter for Standard ML, a functional<br />

programming language. We use this interpreter to parse<br />

strings consisting of mathematical expressions. In addition,<br />

it possesses much syntactic sugar to get closer to the<br />

mathematical syntax that students are used to.<br />

The author enters the expected answer as an expression<br />

in the editor, say (a+b)^2. If the user types in a²+2ab+b²<br />

instead of that (with superscript ² for the square and no<br />

asterisk * for the multiplication), our DSL can assert that<br />

those expressions are semantically equivalent. Furthermore,<br />

the DSL can also check for a lower-level match, that<br />

is, whether or not the user has simplified an expression to<br />

the fullest extent. In this mode, the DSL for instance finds<br />

that a(b+c) is equivalent to (b+c)a, but not to ab+ac, since<br />

this would require a transformation of higher order.<br />

Our focus on mathematics also led to the decision to<br />

support LaTeX. For this purpose we integrated Math-<br />

Jax since it supports a clean rendering of equations in all<br />

modern browsers. Hence, the system can render complex<br />

mathematical expressions on top of any video, see Figure<br />

3. We expect the quiz authors to be familiar with basic La-<br />

TeX syntax.<br />

Research Track |156

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!