Press "Enter" to skip to content

BbWorld16 DEVCON16 Monday 11th July

BbWorld16 DEVCON16

Monday 11th July 2016

These are my notes – I’ll try and clean them up after the conference and please feel free to share with me your comments and corrections.

Getting started here today at 08:45 local time in Las Vegas.

Reminder of product development over the last year:

More coming:

  • Continued development of 9.1, Ultra, collab, Bb Student
  • Bb Planner
  • Bb ID
  • Blackboard Advice
  • Blackboard Predict
  • Security and Encryption
  • Rollout of REST APIs
  • Extensible platform
Bb are adding about 200 more engineers to the organisation
[Should be good to see faster development]
Bb new partnership between IBM and Amazon Web Services
Bb felt that there were better people out there that could run their servers and network but Bb will continue to manage the services running on these servers.
Migration expected over the next 3 years.

Applications that you want to remain on Bb Data centres – these can remain and Bb will maintain these.

Building Blocks misbehaving are probably the biggest cause of outages as they had unrestricted access to the system.
Thus APIs are being provided to make interaction a bit safer.

New API Gateway on Developers Platform.

At Keele we are on Learn 9.1 Self-Hosted.

You can only get Ultra Navigation, courses, bus logic and SaaS backend in SaaS.
Everything else is shared between Learn 9.1 and Learn SaaS.

REST APIs are not just on the SaaS system but also on the 9.1 system too. This was not the original plan but vastly requested as Bb have done it.

Building Blocks will continue. in SaaS they work – but differences need to be considered in e.g. the databases.

Both 9.1 and SaaS now have support for JAVA 8

Still IMS Global Certified.

Caliper Analytics in Learn SaaS – looking at self hosted 9.1 but no timeframe.

Mike Sharkey
VP of Analytics
Previous with Blue Canary until Bb purchased it.

It’s your data Bb are only stewards of it.

Bb want to work with you to find useful ways to interpret your data and make it work for you.


REST 101 RESTful Learn Integrations

What is REpresentational State Transfer (REST)?

Benefits of REST:

  • Preferred HTTP-based model for building internet solutions – supports a greater breadth of integration potential than LTI
  • Doesn’t require JAVA knowledge
  • Lower development cost
  • Near zero impact on Learn performance – reduces time to solution.

Application must be approved by administrator before it can access your data.



not currently used (looking at implementing) but here is an example as it comes up with REST:

Best Practices:


Bb have standardised on GIT repositories and GIT hooks to deploy.

[self note for hosting on windows server and IIS]

Hard limit of 200 items per page in requests.:

Requests are rate limited – this is to prevent DDoS attacks etc. The exact numbers are still being worked on.

Developer portal has all the documentation required.

Dev portal allows you to see who else is using your application.


Current API calls:

Bb Dev Portal does the checking of the key and secret authentication on application calls. Learn issues token, periodic check with Dev Portal to check continued authentication.

Looking into to producing libraries. Not sure yet if they will yet.

When you register your app in the Dev Portal it is available to the world. BUT there is no discoverability for others to look for it.

Big possible security risk – as all the app users use the same key and secret, someone could use your app on your solution. Raised by SV.

Bb’s hope in the long term is that Building Blocks are replaced by REST.


Data Science

What is Data Science? 
data Science happens when a specific, concrete question is answered by the analysis of some set of data.
often conducted at large scale, intended to inform or develop a software application.
Netflix – your recommendations – calculating what you like and suggesting new content you might stay with netflix to watch is data science – predictive models.


How do we identify students who need help and get them back on track? – Data Science question.
Process of data science:
  1. Formulating a question that can be answered with data
  2. Identifying, collecting and cleaning the data
  3. Analyse the data (with machine learning, statistics, neural networks, etc)
  4. And communicating the answer to the question to a relevant audience.
Data Science Team:

  • Data Engineer
    • Build infrastructure
    • Data Set aquisition and cleaning
    • Data stewardship
    • Implement production tools at scale
  • Data Scientist
    • Indentify valuable data sets and perform cleaning
    • Run experiments
    • Visualizing
    • Modelling
  • Data Science Manager
    • Builds the team
    • Sets goals and priorities and removes road blocks
    • Manage the data science process
    • Interact outside team
    • Translates results for broader digestion
What is the data science process?
Develop Expectations -> Collect Data -> Match expectations with data -> [Go to start]
Do you have the right data?
Do you need more data from other sources?
Do you have the right question?

Mining for gold in your institution’s Blackboard Learn data

Staff and students have a high interest in the analytics. Not of every possible analysis (e.g. students weren’t interested in analysis of sessions attended) but overall they have a high interest.

Example HEI:
  Used two goals in strategic plan:
  Identify effective interventions driven by Learning Analytics.

Analytics vary over the duration of the course with effectiveness going up and down. This is different for each course and within each course over the duration.

Started with just a friendly email, then email and video, then email video and sms.
Raising awareness did NOT have an effect in learning outcome.

Only 76% opened the message and only 36% clicked through the message.

Supplemental instruction – teaming students with peers who are doing better – not too much better so as not to put the poorer student off.
Doing this improved the learning outcomes.  Earlier intervention = better.

Students given a realtime measure of how they are doing (use of grade centre) are 1.5-2 times more likely to achieve a C or above.

table in Bb activity_accumulator collects student activity in Bb.

This has a lot of unclean data.  e.g. multiple login clicks when the system is running slow.
Bb stats cleans this data in ODS tables.

In these tables you can start to dig into the data and run analysis on things like “is the student just going in to check for updates and then going to the next course?” / “How many courses does the student access per session?”

Not all courses are created equal

  • Art 101 can not be compared to Law 102
Be careful what comparisons you undertake.

Be First to Comment

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.