Positive? Negative? Neutral?

Discover the tone of documents and individual entities

Overview

Sentiment analysis detects tonality (positive, negative, neutral) and subjectivity (fact, opinion) within documents. It can determine whether a document or even a sentence expresses a subjective negative or positive connotation. It can even show what opinions are expressed in particular clusters of sentences in a given document. Further, it can perform what is known as faceted sentiment analysis, determining what opinion is being expressed about a given entity.

Applications

Sentiment analysis can be used to measure perception about documents, sentences and even entities (products, people, companies, etc.). This permits a deeper understanding of content and facilitates smarter advertising contextualization, research and repurposing of content.

Nsentiment can now enable Public Image/Reputation/Brand Monitoring, Voice of the Customer, etc.. It allows to mine user-generated content and feedback to help better manage brand and reputation.

How it works

Nsentiment assigns tone (positive, negative, neutral) to a document while also identifying its subjectivity (fact, opinion). It uses a statistical model to identify and match subjective patterns within a document. It also filters text according to a set of linguistic rules that determine tone within a given semantic context.

For each sentence of a processed document, the module identifies whether it is a fact or an opinion and assigns a score representing the subjectivity level of the opinion. For sentences containing opinions, Nsentiment then assigns a tone value (positive, negative or neutral) and a score measuring the intensity of the said tone.

The same values are attributed at the document level to express the overall subjectivity and tone intensity levels. Distribution scores are also computed at this level in order to express the density of the opinions and tone throughout the document.

Nstentiment can also identify the tone value of particular extracted entities. It can assign subjectivity and tone scores to any entity extracted by Nfinder, allowing extremely precise visualization and comprehension of content.

Input/Output

Nsentiment ingests documents in any format, and outputs:

• A list of sentences and entities with their subjectivity and tone values
• Subjectivity and tone values for the entire document