Use CasesUse Cases
Find the current sentiment and add a positive spin to new Blog Posts.
- Social Media Sentiment
Track how your company is perceived and take action.
- Product Reviews
Correct or expand the reaction to your product.
Automated Sentiment Analysis
We have an extensive compiled entity dictionary as part of our knowledge base, and we update it daily. Our knowledge base automates the NLP (Natural Language Processing) sentiment analysis for overall text, entities, and keywords. By selecting a domain area such as news, product reviews, or social media, the service tailors to the domain and refines the sentiment accuracy. We have been refining our algorithms to reflect a wide spectrum of polarities. Natural Language Understanding (NLU) allows us to continually train our service to recognize the similarities within the text in the domain and become more accurate with the results.
Several customers have special entities and keywords particular to their business. We have modified our automated text analysis to allow for user-defined customer dictionaries for additional entities. When selected, the custom entity dictionary is used to identify and extract entities that are not in the current dictionary. You can update the dictionary as these new entities become part of your business.
Smart Sentiment services assigns an overall sentiment: positive, negative, neutral, or no sentiment to the original text. The sentiment is based on a weighted comparison of all sentiment words in the initial text document. Sentiments are based on a weighted comparison of all natural language words denoting or evoking a sentiment, such as “good”, “bad”, “lovely”, “splendid”.
The “neutral” and “no sentiment” are not the same. “No sentiment” means there is not enough sentiment words to determine a sentiment. A “neutral” sentiment results from the presence of both “positive” and “negative” sentiments that neutralize each other.
Smart Sentiment services extracts entities (persons, places, or organizations) from the text, and assigns a sentiment per entity. The sentiment uses the local context or sentiment words in the vicinity of the entity to help determine the sentiment score. We identify at least one category for every entity and multiple categories if they exist. The additional categories are possibilities for new audiences or business opportunities.
The keyword score shows the importance of the keyword to the original text. We assign a sentiment to each keyword or keyword phrase listed in the text. The sentiment score reflects the only the sentiment of this keyword in this text. The same keyword could have a different sentiment in another text because the sentiment is determined by the context. Having an individual keyword sentiment score, and an entity sentiment score, and overall score gives a more complete picture of the sentiment around your topics and categories.
Smart Sentiment Pricing
- No daily limit on requests
- Easy to scale
- Pay only for what you use