NLP entity sentiment analyses.

Question

I need to perform entity sentiment analyses using Azure sentiment with Opinion Mining and it goes job, but it returns the starts for “discovered” entity not the entity of my chose. Does Azure has a service that would accept an array of my own entities and do the sentiment analyses in a text against my own entities?

Answer 1

@Roman Bessnow (FH TRUE Global Intelligence)

Thanks of reaching out to us. Sure, you can create your own custom text classification model. Below is what is that and how to do it:

What is custom text classification: https://learn.microsoft.com/en-us/azure/cognitive-services/language-service/custom-classification/overview

How to --

Create your project: https://learn.microsoft.com/en-us/azure/cognitive-services/language-service/custom-classification/how-to/create-project?tabs=portal

Define schema: Know your data and identify the classes you want differentiate between, avoid ambiguity.

Tag data: The quality of data tagging is a key factor in determining model performance. Tag all the files you want to include in training. Files that belong to the same class should always have the same class, if you have a file that can fall into two classes use Multiple class classification projects. Avoid class ambiguity, make sure that your classes are clearly separable from each other, especially with Single class classification projects.

Train model: Your model starts learning from your tagged data.

View model evaluation details: View the evaluation details for your model to determine how well it performs when introduced to new data.

Improve model: Work on improving your model performance by examining the incorrect model predictions and examining data distribution.

Deploy model: Deploying a model makes it available for use via the Analyze API.

Classify text: Use your custom modeled for text classification tasks.

By the way, since sentiment analysis is kind of not sensitive to training data set, is there any special scenario you want to do so? I am asking this to check any chance we can improve the product.

Hope this will help. Please let us know if any further queries.

------------------------------

• Please don't forget to click on or upvote button whenever the information provided helps you. Original posters help the community find answers faster by identifying the correct answer. Here is how
• Want a reminder to come back and check responses? Here is how to subscribe to a notification
• If you are interested in joining the VM program and help shape the future of Q&A: Here is how you can be part of Q&A Volunteer Moderators