Domain & Aspect Specific sentiments
BigTapp’s DASA is a SaaS solution that takes in a text input like a review or feedback and returns the individual sentiment on each of the key entities mentioned in the text. The key entities could differ for different domains and are trained as part of deployment using domain-specific ontologies.
The inherent ambiguity of natural languages (NLP) causes aspect-based Sentiment identification to be biased with false positives. To overcome this challenge, we have created a hybrid approa based on the InFoActiV platform using a - DL (RNN) (Deep Learning with Recurrent Neural Networks) method which helped us to attain an accuracy upwards of 90%. Most of the DASA would be based on Keywords. In contrast, BigTapp’s engine is based on NLP / Concept based - Handles Negation, Domain dependency, Double adjectives, POS Tagged, entity/features / critical words .. etc.)
How DASA works is represented in the below process flow:
Key differentiators of Bigtapp's DASA
BigTapp’s DASA solution comprehensively analyses sentiment on individual key entities based on domain-specific ontologies, making it a valuable tool for businesses. The following key differentiators with examples would provide insight into how BigTapp’s DASA would help you overcome the challenges.
For more details and walkthrough, please fill out the form below. We will get back to you soon
Aspect-based sentiment
E.g. “The food was awesome, but it was pricey” – DASA identifies the Food aspect & positive sentiment associated with it and the Value aspect with negative sentiment.
Explicit Vs Implicit Feature
Chicken Fajitas (Food) was very tasty, but It was costly (Price is a feature and expensive is an indicator) Vs I had a great meal, and the food was worth it-> is implicit, which gives the sentiment for Value(Price) without directly talking about it.
Handling Negation in an effective manner
E.g. The meal wasn’t bad -> Indirect way of telling that it was a decent meal which is positive
Negating the Negation
E.g. Don’t waste your money on this restaurant -> Here, waste is negative; even though there is a negation, it still gives a negative sentiment to Value.
Handling Special phrases / North American Informal adjectives / Words
E.g. “ The waiter was kick-ass, and the price was bomb-ass -> Indicates that the waiter was aggressive and the price was great.
The same word can have different sentiments based on the context
E.g. Price is high – Negative sentiment on Value / Price
E.g. food quality is high – Positive sentiment on food
Handling Sarcasm (to an extent)
E.g. The most remarkable thing about this restaurant is that for over five years, they‘ve served us nothing but leftovers. The original meal has never been found!!
Customer Case Studies
Identify the sentiments for different aspects and overall sentiments of customers for a restaurant chain
A full-service dining company was enabled with sentiment analytics to understand the overlying cause of positive and negative sentiment with increased accuracy. BigTapp's Recommendation engine used...
