Reporting a major bug in Azure cognitive-services/text-analytics
I was using the text analysis service API by Azure & found many major flaws : (The API is not mature enough to handle simple real life sentences correctly).
Below are few sample sentences which I used & the result was absolutely wrong.
1. I have many ways to not give you the expected grades
2. There are many ways to not give you the desired grade
3. How could you not understand what I said
4. We have attempted to launch your course in a new window, but a popup blocker is preventing from opening it -> this should be neutral text.
5. I did not always understand what you were saying.
6. I haven't had major luck with finding a job
7. I am indebted by your strong confidence in me
8. In all his life Trump did nothing in a manner worth appreciating
I simply compared it with Sentiment Analysis with Python NLTK Text Classification at below site & found Azure to be highly inaccurate in many cases.
Please fix this asap to improve on the quality & reliability of the API response.
Thanks for your feedback and examples that you’ve provided. I've reached out via email to discuss your issues, thanks!