David Monllaó

Data Engineer at BHP

[David Monllaó @DavidMonllao]

Last day at #moodle. Thanks to all the people I worked with during these more than 7 years, it was a pleasure.

[David Monllaó @DavidMonllao]

@jlegarreta @JesusSoriaD @SERConsumidor @JulioBasulto_DN En tetrabrik por favor

[David Monllaó @DavidMonllao]

@jlegarreta a ratos si :) no sempre però heheh

[David Monllaó @DavidMonllao]

Another article published (some months ago) as part of the Master of Philosophy in machine learning at UWA. A supervised learning framework: using assessment to identify students at risk of dropping out of a MOOC https://t.co/eZxCPUuJdC #moodle #learninganalytics

[David Monllaó @DavidMonllao]

@moodlemoot_us @LSUOnline Nice :) This is aligned with the new insights we are adding to core for #moodle 3.7 and 3.8: Reminders about upcoming activities due for students and insights about students who have not logged in recently or since the start of the course. More info in https://t.co/fx4ynouNS0

[David Monllaó @DavidMonllao]

@farhan6318 @sara_arjona @moodle Sure, you can contribute :) E.g. Moodle components need to include training data to help the NLU backend identify users' intent. We need to agree on a format that is easy to translate and that can be reformatted for different NLU backends: https://t.co/6Agkm1zUNM

[David Monllaó @DavidMonllao]

If you would have an AI assistant in @moodle, what would you like it to do? #moodledev #learninganalytics #Rasa_HQ https://t.co/1LYIeufmYe

[David Monllaó @DavidMonllao]

@MatthewPorritt @lmspulse @moodle There it is https://t.co/w05zXlnBd6

[David Monllaó @DavidMonllao]

Today I have presented "Moodle learning analytics from different perspectives" at Moodlemoot Croatia 2019. You can find the slides here: https://t.co/MUgnmvuNA1 #moothr19 #moodlemoot #moodlelearninganalytics

[David Monllaó @DavidMonllao]

@lmspulse @moodle We aim to add an AI assistant to Moodle core although as mentioned above, we first need to identify how an AI assistant adds value for Moodle users. There is no timeline at this stage.

[David Monllaó @DavidMonllao]

@lmspulse @moodle I hacked the messaging UI for the POC in https://t.co/1LYIeufmYe. This is not final, just faster than to integrate a separate UI for this I just wanted to test the NLU backend. The final user interface will be designed by the UX team.

[David Monllaó @DavidMonllao]

@lmspulse @MatthewPorritt @moodle Sorry, I may have added too much information and things get confusing. Both https://t.co/00sFqLODde and https://t.co/1LYIeufmYe are just technical proof of concepts. There is no tracker issue for this yet. I will create one during the EU morning and post the link here.

[David Monllaó @DavidMonllao]

@lmspulse @moodle Everything is modular in #moodle so yes, there would be an API for plugins to add new intents and possibly a JS API for session-related pokes (e.g. https://t.co/00sFqM6e4M). The other important & related API is the analytics one, the assistant would channel its insights to users

[David Monllaó @DavidMonllao]

@icsbcn @moodle It would probably forward you to https://t.co/seoYi6U5qU :)

[David Monllaó @DavidMonllao]

@farhan6318 @sara_arjona @moodle BTW the linked branch is just a quick demo to play, nothing serious. We are looking for an open source NLU solution as a default to avoid sending the data to 3rd parties. However, we would abstract an NLU API so Dialogflow, LUIS or other NLU backends can be developed.

[David Monllaó @DavidMonllao]

@sara_arjona @moodle @farhan6318 lol, I knew someone would ask Chuck to complete homework

[David Monllaó @DavidMonllao]

@rutx @moodle Cool, the assistant should be an active agent and ping us about different aspects that may be interesting for us. These are some good examples. The assistant should also reply to our comments: "how can I xxx?", "where is xxx?", "tell me a joke Chuck"... Any more ideas?

[David Monllaó @DavidMonllao]

We got an article published on @IEEE_TLT https://t.co/IAof8tTmO8. Better to identify students at risk of missing an assignment than to identify them later when they are at risk of dropping out of the course. We used #moodle #learninganalytics API. https://t.co/uNsngfln9j

[Elizabeth Dalton @e_m_dalton]

Those objecting to the way Instructure is monetizing data and analytics from Canvas might want to look at @Moodle. We have robust #LearningAnalytics features, but they stay under your control. We don’t collect your data, you decide what models to use. https://t.co/oZu8w1hyxF