Becoming Sensei - Training the Trainer Along With Training the System

[#1]

Becoming Sensei - Training the Trainer Along With Training the System

 

by gdWct

Overview

Digital asset management and delivery is one of the most useful features of AEM. While it's tempting to release Adobe Sensei and Smart Content Service in the wild and hope for the best, if not set up properly, it can become a daunting task for content managers, digital marketers and content authors to understand how to train the system before taking advantage of its full potential. Thankfully, there are great resources online and AEM best practice guidance to take advantage of machine learning and automatic tagging of assets.

Smart Content Tagging that Fits Your Enterprise Taxonomy

Going through smart content training currently, I can recommend these great resources to use as guides:

• Efficient Smart Tagging
• AEM Smart Tag Training Experience

 

Tag Modeling and Training the System

Once there's sufficient level of understanding on what a training model is, it's time to use your tag library and train the system to recognize your enterprise taxonomy:

• Enhanced Smart Tags Guidelines 
• Image Training Guidelines

On-Going Periodic Training, Add, Remove, Promote Tags

Once you set it, don't forget it! Smart content improves with continuous grooming of tags.

Content managers and authors can add tags that describe visual elements and visual aspects that machine learning might not be able to detect on its own. Also, deleting tags is an important ingredient of fine-tuning your smart content service, as well as promoting tags that should be elevated based on your business goals and enterprise taxonomy.

Remember to be very matter-of-fact with your model training, so you can just focus on periodic adjustments and tagging quality control. Before setting up your model, review tagging-related limitations and best practices. Content authors and digital marketers will really appreciate the accurate, time-saving and easy to use digital asset management and delivery that AEM has to offer. And your 'go to market' speed with impactful, eye-catching content will be Sensei-worthy.

Q&A

Please use this thread to ask questions

[#2]

@gdWebct Great content!! Thank you for sharing this with the community. 

[#3]

@kautuk_sahni appreciate it, it's great to have this blogging feature to share with the community. We'll try to keep sharing useful content here at regular intervals.

[#4]

What happens to the Smart Tags on existing assets after I do some training? Do they get removed/added?

And if I have already deleted an incorrect tag, is there a chance that training might add it back again?

[#5]

Good question! If you perform existing asset smart tag training periodically, any tags you add are added to your existing tags. Adobe recommends to not add more than 50 tags per tagging model (and to use at least ten different images for each tag). You can then delete any tags you do not need, via Manage Tags menu (manually, for now).

Keep in mind that just like machine learning, asset smart tag training should be a on-going hybrid process of adding tags manually (enhanced or custom smart tags), removing unwanted tags and letting the tag reporting system and tagging workflow run and update your tags.