Text/Image/Data classification descriptions are confusing

[beccamc]

I'm a little confused by the various titles, images, and descriptions of classification.

My main issue is...

• Graphic used for Image and Text match (and the scenarios are next to each other)
• Description for Image and Data match more closely.

Maybe this would be fixed by the new graphic we're supposed to be getting? I just know I would be confused as a customer.

[beccamc]

We need to add the new scenario to the doc link here! I would say that's a blocking issue.

[luisquintanilla]

@beccamc can you post the link we need to update docs for.

[beccamc]

Yes! https://aka.ms/vsmbscenario

[luisquintanilla]

@beccamc here's the PR I'm using to track the doc work for these scenarios.

https://github.com/dotnet/docs/pull/32065

[luisquintanilla]

cc @zewditu

[luisquintanilla]

Action Items

• [x] Remove limited scenarios
• [x] Add section groupings to scenarios
  • Tabular
  • Data classification
  • Value prediction
  • Recommendation
  • Forecasting
  • Computer Vision
  • Image classification
  • Object detection
  • Language
  • Text classification
• [x] Add subheadings / descriptions to top-level groupings
  • Tabular: The following scenarios use Automated ML to train and pick the best model for your data (numbers, text, categorical, date).
  • Computer Vision: The following scenarios use deep learning techniques to train models on image-only data.
  • Language: The following scenarios use deep learning techniques to train models on text-only data.
• [ ] Update data classification description to: "Classify data (numeric, categorical, text) into 2+ categories. e.g. categorize flowers based on petal size and width measurements"
• [ ] Update Text classification description to: "Assign a label or category to raw text data. e.g. sentiment analysis on customer review."
• [ ] Docs: Add information about models in link => Luis can do this
• [ ] Update Text classification title from "Text Classification" to "Text classification"
• [x] Move value prediction to 4th position (after Data, Image, Text classification scenarios).

[LittleLittleCloud]

Would it be better to change Language to Nature Language Processing

[luisquintanilla]

Would it be better to change Language to Nature Language Processing

Maybe. I could go with either. That's what most places refer to it as. I went with language to keep it short.

@beccamc @JakeRadMSFT thoughts?

[beccamc]

I like Natural Language Processing but I have no data backed explanation for why.

[luisquintanilla]

I like Natural Language Processing but I have no data backed explanation for why.

I think people are familiar with this concept and there's several examples where that's the term used. So if anything I have no real reason behind "Language" other than it's shorter.

Let's update to Natural Language Processing.

Azure ML

HuggingFace