Sentiment is the attitude, opinion or feeling toward something, such as a person or a topic. On Pulsar we analyse sentiment by default, and are able to measure the positive, negative, or neutral opinions about a particular topic, an organisation, person, product or location. We can analyse the sentiment of an entire News article, a TV or Radio transcript, a Blog post or a Tweet, and each analysed piece of content is assigned a sentiment score on a scale of -1 to +1. However, content with very positive or very negative messages can be scored on a larger scale of +10.0 to -10.0. This post level sentiment score is available in our Feed exports and the Pulsar API. Additionally, we also give users an average sentiment score for their entire dataset, helping them assess how positive or how negative the overall conversation is, as demonstrated below.

Our Sentiment engine looks for words that carry an explicit positive or negative meaning and then figures out which person or place those words are referring to. It also understands negations (i.e. "this car is good" vs. "this car is not good") and modifiers (i.e. "this car is good" vs. "this car is really good"). However the algorithm is not always able to register sarcasm, irony or slang e.g. if a post were to include the following phrase..."That was a sick concert! I really loved it! :)". "Sick" as a keyword in this instance might be originally coded as Negative, when it's in fact used as a positive way for denoting excitement.

In cases where a positive or negative meaning cannot be assigned with certainty, it is normal that the algorithm will return a neutral sentiment. And in instances where the language of the post is not supported the algorithm will return a 'not analysed' label. Below is the list of languages Pulsar supports for sentiment analysis, with many more to come!

Sentiment analysis on Pulsar can be found in a number of sections, including the following:

The Sentiment over time graph is grouped by Volume, Visibility, Media Reach, or Social Impressions.

The Keywords Word Cloud chart shows the most common keywords in a search and the sentiment associated with those keywords. The bigger the size of the keyword and the more central it is in the graph, then the greater the number of posts and articles containing that term.

The Hashtags Word Cloud chart shows the most common hashtags in a search and the sentiment associated with those hashtags. The bigger the size of the hashtag and the more central it is in the graph, then the greater the number of posts and articles containing that hashtag.

The Topics Word Cloud chart shows the most common topics in a search and the sentiment associated with those topics. Topic analysis on Pulsar is based on a top-down taxonomy led classification, which identifies the main themes in a document, even if they aren’t explicitly mentioned. The bigger the size of the topic and the more central it is in the graph, then the greater the number of posts and articles about that topic.

The Entities Word Cloud chart shows the most common entities in a search and the sentiment associated with those entities. Our entity classifier identifies People or Organizations being discussed in a post, as well as People or Organizations mentioned in a TV, Broadcast or Podcast clip. The bigger the size of the entity and the more central it is in the graph, then the greater the number of posts and articles discussing that entity.

We also display Sentiment in the Locations section. Here you can see sentiment breakdown by Country or by City helping you assess how attitudes and opinions differ by geographical region. The City Map further allows you to measure for example how sentiment differs within a country's urban versus rural population.

Lastly, you can also find Sentiment in the Feed Results view. In this view you can see each individual post that we've collected in your search, and the sentiment we have assigned to that post. You'll notice we have incorporated a traffic light palette that makes it easy to discern the sentiment of a piece of content, where a red smiley = negative; a green smiley = positive; and a grey smiley = neutral.