Gartner Magic Quadrant for Insight Engines 2019
Attivio was recognized for our completeness in vision and ability to execute.
Attivio Platform 5.6.2 Available Today
We’re pleased to announce the release of 5.6.2 of the Attivio platform. There are several new features we think you’ll be excited about!
Monthly Release Cadence
Before we talk about new features, I’m happy to announce that we’re moving to a monthly release cadence. Starting with 5.6.2, we will release a new version of 5.6 every second Thursday of the month.
There are two reasons why we’re moving to monthly releases –
- Provide a simple and predictable release schedule. You have told us you want to be able to plan for a simple and regular upgrade cadence. Upgrading to the latest Attivio version should be built into the regular IT maintenance schedule, which will allow you to plan the upgrade along with other initiatives in your organization.
- Be agile. We don’t want you to have to wait for a year to get a new feature we developed. We want to give it to you when it’s ready and we can’t wait to hear your feedback!
So, what does it mean?
- Monthly Product Releases – Second Thursday of every month – will include new capabilities and bug fixes
- Upgrade process will be less than 30 minutes and mostly automatic
- Versions are sequentially increased on the third digit (5.6.2, 5.6.3, 5.6.4, …)
To support monthly releases, our R&D team has invested in making our upgrade process super easy. Upgrading from 5.5.1 to 5.6.x is done through an upgrade tool. Moving forward, upgrades will be done through the Attivio installer.
Read more about the updates we made in 5.6.1.
New Features in 5.6.2
And now the part that you’ve all been waiting for – what’s new in 5.6.2.
Improvements to our Machine Learning Relevancy training algorithm
Relevancy is the foundation of the great search Attivio delivers to your employees and customers, and with every release we aim to make it even better. When we introduced Machine Learning to our Search Relevancy capability, we enabled the system to learn from every user interaction and automatically adjust the search results ranking. This was critical to ensuring the accuracy and adoption of search applications, because search applications – and their answers - are not static. We live in a world of ever-changing information, with content being added on a frequent basis and evolving user expectations. Machine Learning Relevancy allows you to automatically optimize search experiences and adapt to changes in your content, your users’ behavior, and your users’ expectations.
Take a Customer Support portal, for example: you want to make sure you have a learning system that can automatically adjust as new products are introduced and to new types of customer questions. Our Machine Learning Relevancy analyzes users’ behavior (click, likes, etc.), understands why they clicked a certain result and didn’t click a different one and generates a relevancy model that will predict the best answers to every question.
5.6.2 includes significant improvements to our Machine Learning Relevancy, making our relevancy models even more accurate and the whole process even easier to operationalize. In 5.6.2 we now calculate features during training, making it easier to add new features to existing relevancy models. We also automatically prune unimportant features making sure only the most essential ones are being used for training.
Our trainRelevancy tool can now automatically generate a Relevancy Golden Set from users’ signals and can compare two models using an informative relevancy model ranking score. These improvements will allow even our non-technical customers to easily generate new models and ensure newly published models will improve users’ search experience.
Search UI – Adding New Signals
We take a pervasive approach with our Machine Learning. For us, everything that the user does is a signal that can be used for search experience optimization. In 5.6.2, our Search UI now tracks what autocomplete suggestion(s) users clicked. Search UI is our generic search application built on top of Attivio’s open source Search UI Toolkit (SUIT) and autocomplete helps users complete their query as well as suggest answers. Because the search journey starts in the search box, even before a query is submitted, we want to track users’ behavior so we can fully understand the journey and evaluate the complete search experience.
Upgrade to Hadoop3
We have upgraded our backend to Hadoop3. This means that we now can go even bigger with our implementations!
More to Come…
As I mentioned, we’ll be releasing monthly (…second Thursday of the month…) so check back soon to hear what we have in store for 5.6.3. And, you can always contact us to learn more in the meantime.