BbWorld17 - DevCon - Monday 24th - Tuesday 25th July 2017
These are my notes and thoughts and should not be seen as a fully accurate or complete account of the information provided. Some of the information included here contains features that Blackboard are working on / towards but have no guarantee to bring to market.
If you feel I have missed something / been inaccurate in my note taking please let me know in the comments and I will correct / discuss / debate with you :)
DevCon: Blackboard technology state of the union
Produce Excellence & Quality
Average time to resolution down to 6.5 days for critical issues. New test automation technology and higher coverage is supporting this.
203 clients on SaaS, 68 underway and 52 evaluations / pilots.
- (The backbone)
- Self Hosted
- Managed Hosted
ALLY is available to integration to Bb, Moodle Rooms and Canvas. Later to D2L and others.
Bb Student app is now Blackboard app - still aimed at students.
Bb Instructor is the new app for staff. Bb ask us to remember this is version 1 of the app and grading etc. is coming soon. (no timeframe discussed)
REST APIs are expanding in coverage. Bb's support for open standards is also increasing. Bb is certified with the IMS Global®
TransparencyBb Community allows for more communication and discussion with users. Recent voting allowed for the community to direct the companies development direction etc.
Bb are working on providing a bigger / longer look into the Bb roadmap. With greater visibility comes more chance of change as development progresses.
DevCon: Architectural modularity and adaptability with Blackboard Mobile products
All communications via HTTPS for secure communication.
Modularity- the way for Bb to make their products more usable and enable Bb to be more responsive to development and support.
Base navigation panel for native apps gives the user a consistent way of accessing other functional spaces.
Allows users to view their courses for the past, current and future semesters
This module allows the user to quickly navigate sections within the course.
Bb now have the ability to reuse these widgets into other modules.
Product AdaptabilityModule Repository
Modules in this repository can be shared across all products, they are all based on Bb design and technical standard.
Based on standard modules and added some instructor specialised customisation, can be independently distributed into the market.
Based on standard modules and added some specialised customisation, can be independently distributed into the market.
ReliabilityIndependent Modules > Unit Testing - this guarantees quality of the system - > Continuous Integration Testing - this makes sure all of the modules developed work well with each other - > Manual/Automation Testing - last defence and final testing to make sure nothing has been missed.
Live demonstration of the modules being pulled into the app development and how they work.
DevCon: Going Splunking: Using Splunk for server log analytics in Blackboard Learn
When you get an error emailed to you, you usually try and look in the log files. But there are many log files with different formats and it can be very hard to dig your way through them.
Splunk is Google for Logfiles.
Schema on the fly
- Collect and index data
- search, analyse and visualise
- Monitor, alert and report.
Forwarders can be placed directly on your data sources to send your data to splunk
Indexer goes through your data and transforms the raw data into events and stores the events in an index.
Managed and self-hosted Bb Environments
Not sure about SaaS - Offering ELK (Elasticsearch, LogStash and Kibana)
Alternatives to Splunk
Demonstration of how this can be used to query student claims of submitting but no result etc.:
Simple way to visualise the activity when filtering on the student ID and PK1 value of the quiz.
Recently there is now a log file entry to show when a file is attached to an assignment submission.
This allows you to check if the student did or did not attach a file with their submission.
getSafeFileName is the search term to look for file attachment.
Philips Hue lights can be controlled by splunk analysis. i.e. have a light turn on if there is an issue :)
DevCon: Machine Learning: Past, Present, and FutureHilary Mason, Fast Forward Labs
In 1880 it took 7 years to count everyone who lived in the USA. Someone developed a machine for simply counting the US census.
This machine used punch cards and this lead into the creation of IBM where these cards were used into the 1900's.
These days we have a change in our language where we now refer to software as though it were a person. People refer to an AI as "The AI" and even how they can get it to be their BFF.
In the 1960's they felt that within 10 years we would have a solution that can translate natural language.
In 1952 Bell Labs were able to develop a mechanical 'mouse' that could solve its way through a maze and learn as it went so as to inform future path decisions.
In 2017 we are not doing anything new, but it is now so cheap and fast we can handle a lot more data in a shorter time.
Currently we are able to develop a computer user interface which is the same as the user interface we use between humans - speech.
Data Science is going from "How many people signed up for my course" to "How many are likely to sign up next semester".
- Big Data
- Analytics & BI
- Product data science
We are starting to see real value generated from machine learning.
Data Science is a team sport!
We are augmenting human capabilities not replacing humans.
Intuition and data - it's not one or the other
Fast Forward Labs
Natural Language Generation - going from data to naturally sounding sentences. This has been demonstrated on real estate listing creation.
The real impact will be in making complex data simple.
Deep learning for rich images is something a lot of people have their eye on.
This can go wrong - example: algorithm learnt that anything that looked like fries next to water was a crab so a burger restaurant's images had some listed as "Crabs" when they do not sell crabs and there were none in the image.
Most machine learning is done within black boxes - i.e. you can see what is going on or why but we need to know what is going on and we need to understand them.
DevCon: IMS Caliper: Crossing the event streamsIMS Global learning standards body creating standards such as LTI and Caliper. This allows for Learning ecosystems without vendor lock in.
IMS provide standard implementations of the Caliper Sensor on GitHub and also the Event Store Implementation on GitHub.
Using Caliper with Bb
Self-Hosted Learn support - planned from Fall 2017 - same outbound as MH/SaaS
- Starting or submitting an assignment
- In Ultra, includes Assignment activity
- Saving a question Event
- Starting or submitting an assignment
Question asked about inbound for Self-Hosted - is the roadmap planned? Not yet as it isn't yet on Managed hosted. It is being considered and issues resolved.
DevCon: Accessibility: Common coding mistakes developers make using ARIA
DevCon: Managed hosting: Does it work? Will I get access to my data?
DevCon: Customize teaching and learning with the Learn 2016 theme with CSS
DevCon: Amazon Alexa & building great voice experiences
We've been using voice for 100,000 years to communicate. but for the last 40 + years we've been using nonverbal forms of input / control.
The real trick is the natural language understanding. understanding the context the request comes with / from.
20 second audio buffer looking for wake word "alexa" on device, once found, device sends next 7-8 seconds of audio to server.
Some example code: https://github.com/alexa