Extracting sentiment, entities, and key phrases from text with Amazon Comprehend | NLP | Python

[@saiprasannakumar4848]

thank you

[@cCcs6]

Hi Pradip, I just tried out quickly the comprehend-service for a quick sentiment analysis and wondered if I did sth wrong - the ability for aspect-based sentiment analysis seems poor and not state-of-the-art at all: In my test phrase "The lectures were great, however, the professors could be more competent." the service rated both entities lectures and professors as positive. However, professors should be clearly rated as negative. Would you confirm this? Thanks & Greetings

[@FutureSmartAI]

Try this model: huggingface.co/yangheng/deberta-v3-base-absa-v1.1

[@cCcs6]

Thanks, I know pyabsa has a multilingual model but as far as I know it's not specifically well trained on German language. Do you know by any chance a model working well in ABSA for German language? :) I haven't found any good though!

[@cCcs6]

For sentence level perhaps, but not aspect level

[@moussimasangojoachim2790]

bon tuto

[@muhammadtalmeez3276]

But this is a paid service we can't use this for every client. Am I right?

[@FutureSmartAI]

If clients gets good quality output they are ready to pay. Now days we can use GPT 3.5

[@74Gee]

There's no sentiment analysis here.

[@paologazzola]

your audio is horrible. But video is useful.

[@FutureSmartAI]

In my recent videos audio is improved