Sentiment Analysis

Elegant mask

Detecting the sentiment of a utterance related to a concept, brand or groups is useful to determine the opinions of people.

simMachines recently predicted the results of an election based on Tweets and sentiment analysis. For each tweet we identify the sentiment of it.

For example, given that we focus our attention on Tweets that mention the country Costa Rica:

  • Costa Rica is a beautiful: Sentiment is POSITIVE
  • Costa Rica is too crowded: Sentiment is NEGATIVE

Aggregating the sentiments of many Twitter users allowed us to predict the results of the 2014 Costa Rican elections.

Sentiments can be plotted over time to obtain a high level overview of the perception of a concept:

Sentiment of tweets over time.

The image shows the positive, negative and neutral perception of a concept in time.

And also a plot of concepts can be extracted based on the aggregated sentiment:

Word cloud

The image displays positive, negative and neutral words related to a concept.



  • Auditable
  • Cloud based
  • Compliance
  • Comprehensive
  • Real-time Predictions
  • State-of-the-art accuracy

Related Work

Credit Worthiness

Credit Worthiness

simMachines created an accurate predictive engine for credit worthiness that gives small businesses loans in a very efficient way.

Logistics Optimization

Logistics Optimization

We predict the duty a product will have to pay when it enters a foreign country.

Human Resources Optimization

Human Resources Optimization

We predict if a resume is going to be a good fit for your organization. Machine learning at its best!