Models
Aspect and sentiment Identification (medical device provider)
This model does the following:Identifies the aspects in survey feedback responseIdentifies the sentiment based on the aspects
Sentiment Classification
The model detects the sentiment of a given text. This is a binary classification model that classifies the text as either positive or negative. The model also provides a confidence score of the predicted class. The model has been trained on a dataset of Yelp restaurant reviews. The Yelp reviews dataset is constructed by considering stars 1 and 2 as negative, and 3 and 4 as positive. The model showed a 96% accuracy on a validation set of 1000 records.
Question Answering in Text
In Question Answering tasks, the model receives a question regarding text content and is required to mark the beginning and end of the answer in the text.
Sequence Tagging
This model demonstrates sequence tagging of a given text. It recognizes varies entities in a given piece of text. This can be used to tag an incoming email or extract knowledge from a text document.
Text Extraction (Product Search)
Extract text from product labels or any other text based images to identify products.
Content Generation V2
This is a newer version of our content generation model. It uses and intermediate step of extracting product specific features and specifications from the description. The model then generates or recreates a product's description using information from its title, brand, and either a set of product features or an old description. Optionally, choose what audience the description should be marketed to (e.g. "parents" or "professional musicians", and a tone for the new description (e.g. "fun" or "technical"). To specifically avoid terms, add them to the Avoid field (e.g. "Warning", "warranty") or to specifically include them, add them to the Keywords field.