[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
|
Re: [mylyn-dev] Towards the improvement of task context?
|
Dear Mik,
Thanks for your quick reply.
Interesting work!
Thank you for your interest in our work!
The original premise of the edit events was that time spent in the editor is more productive than time spent looking around the structure views that eventually get you to the editor. The edit events don’t have to be textual edits, which is why we decided not to base this event on something like keystrokes in the editor. For example, selecting a bunch of text, then copying and pasting it, could produce a number of edit events, and be more meaningful work than 30 keystrokes.
I completely agree that some of these edits that are not textual edits
show certain level of relevance.
That said, the argument that there is noise in the edit events makes sense to me. The thing that’s less clear to me is a crips definition of what the signal really is. In our initial research we assumed only this split between interactions in editor being “good”, and those in the structure views being wasteful in formatting the “edit ratio”. I do think that, given a more precise definition of what edit signal is, it would be great to identify how to remove move of the noise.
We looked at these edits and we identified what some of them are ... But
we cannot assume that it covers all the possibilities that can happen.
Actually we are looking for automatic identification of them.
I would like to know your opinion on when to possibly clean these events
... when creating interaction events or when using them to build task
context?
Regards,
Zéphyrin
Mik
--
Dr. @mik_kersten, CEO, http://tasktop.com
Assistant: jill.charlton@xxxxxxxxxxx, +1-778-588-6896
On Oct 6, 2015, at 1:16 PM, Zéphyrin Soh <zephyrin.soh@xxxxxxxxx> wrote:
Dear Mylyn developers,
With the main goal of improving how Mylyn build the task context (using the degree of interest), we study the Mylyn interaction traces and observe that they contain more than 80% of edit events that are created by text selections in the editor [1]. So they are not actually the changes performed on source code.
We think that because Mylyn uses the kind of event (e.g., edit event) as an indication of the relevance, we wondering if we could process to the filtering of these edit events prior to forming the context or defining other levels of relevance (and their interest weight) under edit events.
For more information about our observations, please take a look at [2].
What do you think or do you have any advice/comments on that?
[1] http://wiki.eclipse.org/index.php/Mylyn/Integrator_Reference#Event_Kinds
[2] https://zsoh.files.wordpress.com/2015/08/esem_camera-ready-v3.pdf
Regards,
Zéphyrin
_______________________________________________
mylyn-dev mailing list
mylyn-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/mylyn-dev
_______________________________________________
mylyn-dev mailing list
mylyn-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/mylyn-dev
