Submit both temporal and text

[polieste]

In the case of QA questions, I want candidates to find the correct video segment and answer the question content related to that video segment.

For example, in this question, I can ask "What is the brand of the accidental truck?". And the answer is "Video name - 924 - 939 - HUYNDAI".

However, I just can find two types of task submission:

• t-KIS/v-KIS Template Submission: Video name - start - end
• LSC QA Template Submission: Video name - text

Is there a template that supports QA Submission (Video name - start - end - text), if so, can you provide it to me?

[lucaro]

You don't need a template. All a template does is pre-select some options in the form for you. You can simply select all the options you actually want from an empty form.

[polieste]

Hi Lucaro, what I mean is I want to submit both "MEDIA SEGMENT" and "TEXT" in the same time, but there is no target type support it.

For example, in these two cases, I just can choose the target is "MEDIA SEGMENT" or "TEXT".

So, do DRES have the type of target support both media segment and text submission at the same time? I saw in a competition did that, but I do not know how to do.

[lucaro]

The options you are referencing are not for the submission, but for the validation. It is possible to have submissions including these fields, there is just currently no mechanism to automatically validate them. If you use Judgement as a validation mechanism, for example, you can manually assess incoming submissions. The semantics of how text and media would need to be assessed is not entirely obvious. Could you provide more details on your use case?

[polieste]

In my case, the QA question requires both evidence (including text [only input number] and video segment), which means that I not only need to find the answer to the question, but also need to submit the video segment corresponding to that answer.

For example, a sample question would be:
"[Question] How many times did the chef flip the cake? [Hint] A chef is making a cake, on the table there are green and red ingredients, ......"
The answer would be: Video 15, 15s-30s (corresponding to the hint), 5 (number of times the cake was flipped).

To automatically validate the answer, just match the video segment and text (number) with the available answer. The video segment matching is already available, the text (number) matching is also available. So I think this combination is possible with the current DRES.

The combination of both video segments and text answers in the form of numbers will increase the fun of the question, and will also avoid players submitting a random answer and getting it right.

[lucaro]

I see. So you would only need exact text matching and exact pairing of exactly one text and exactly one segment.
That should be reasonably straight forward to add.
In case you are willing to give it a go, I'd be happy to support and review a pull request.