The New Hubert is Finally Here!23rd January 2018
Since launching the beta version at the end of July, we received a lot of great feedback from our valued users. After careful consideration of every suggestion, we’ve broken down the points and mixed them with our own internal input. From that, we organized new functionality into a list of prioritization which has been developed and implemented continuously.
Now, we are proud and excited to present the new and significantly improved Hubert to you!
So, what can you expect for the aptly named Beta 2-version? Here’s a couple of highlights:
With the new version, we’re moving away from the linear conversational flow used in the previous Hubert and into a deep-learning model. Our AI-department has put together a brand new management system for dialog handling and with that, we’re able to train Hubert on patterns from real conversations to achieve a much more personalized and engaging conversation.
Hubert will have a better comprehension of what your students are saying and adapt the conversation based on that. And it’s gotten quite good.
We’ve actually been baffled by Huberts’ wits when trying him out around the office.
When confronted with gibberish or unrelated comments, Hubert is now able to understand the context and respond in an appropriate way. When the conversation goes off-topic, Hubert does his best to steer the conversation back into the evaluation by repeating the question or asking the student to rephrase his response.
No more ‘Can you elaborate on that?’ and ‘Could you please narrow it down?’
When Hubert has received a response that makes sense, he moves on to the next question. If a student doesn’t have anything more to share, Hubert is fine with that. But if the response is rude or includes insults, he lets the student know.
Along with that, Hubert is able to cope with longer responses perfectly fine.
The results-view has undergone a massive overhaul and is now divided into a basic tab and an advanced tab where a lot of new functionality has been added.
Presenting the essence of a bunch of open-ended questions in a good way is something that has never been done before. Word clouds are still the go-to solution to show the consensus among a group, but the consensus for word clouds are also quite clear — they’re not ideal as an only measure of extracting information. That’s why we’ve taken the concept of word clouds and made it our own. In the right-hand list, you’ll find the words most mentioned in every separate category which adds detail to the analysis.
We’ve been exploring multiple options before finally settling on a final results-view we feel is informative, easy to navigate and really cool. Here’s how the advanced results tab looks:
Reach this page by clicking the button with three dots in the basic results-view.
The advanced-tab is capable of providing a three-dimensional overview of student comments — Enter the dimensions of activities, entities, and properties.
In addition to exploring comments in the basic view where all comments are sorted by consensus and category, you can now choose to explore strengths and weaknesses in these three overall dimensions, sorted by consensus.
If you’re curious about what your students felt, for example, about a certain activity, all comments related to that are grouped together for an easy examination.
We have a more functions in development right now and will continue to update the results-view as soon as they are ready for release.
To improve Hubert even more, we need more valuable input from you and your students. A few weeks into your lecturing is a perfect time to send out a first formative evaluation so you can make effective change and get to know your students learning style as quickly as possible.
Until next time, peace out.
— The Hubert crew