My name is Anina and I am working on my Data Science master thesis. The topic is:
"Targeted Aspect Based Sentiment Analysis of Social Media Data"
I have retrieved tweets from the internet and I need to annotate them with the aspects that appear in them and the sentiments that are expressed about those aspects. So for example if the tweet is about a plant based burger product:
"I enjoyed the Beyond Meat burger but I am not sure I can afford to eat it every day"
… Here we have:
Target entity: Beyond Burger
Aspects: Taste & Price
Sentiments: Positive and Negative
And so the labels would be the following:
(Beyond Meat, taste, positive)
(Beyond Meat, price, Negative)
*Note that the target entity/entities for each tweet will already be given. It is necessary to annotate the aspects and the sentiments.
I think that to train a deep neural network I would need between 5000 to 6000 labeled observations to get good accuracy results. I was also wondering if you could advice me on this number. Perhaps less would also be enough.
Please let me know if you are able to help me with the work. I would appreciate it very much!
Posted On: January 14, 2021 13:21 UTC
Category: Data Entry
Skills:Data Entry, Accuracy Verification, English Grammar
click to apply
All jobs | upwork.com