Contents
Automated sentiment analysis inside Mediaportal is here!
What is automated sentiment analysis?
Sentiment analysis engines inspect any given text to identify emotional opinion and determine the writer’s attitude, whether that be positive, negative, or neutral. Sentiment analysis is performed through NLP (Natural Language Processing). The technology attempts to understand the overall attitude (positive or negative) expressed.
Why is sentiment analysis important?
Sentiment analysis is used to quickly monitor the tone of content being monitored and can direct users to where they need to focus and help prioritise their day.
Ongoing sentiment analysis is useful to help shape company and product strategy by understanding public opinion. Sentiment analysis can be used to:
• Understand company reputation
• Understand product perception/expectation
• Detect negative conversations
• Recognise opportunities
• Provide insight on competitor’s reputation and brand perception
• Assist in market research
How does it work?
The technology we use includes powerful machine learning models to analyse the text inside media items.
For English language we use a model created by Stanford University. This model takes advantage of both grammatical structure and word selection. As well as recognising words that are positive, negative and neutral, the model will also look at the structure of sentences. For example, the below phrases would be categorised as follows:
‘The movie wasn’t bad at all’ Positive
‘The movie was bad’ Negative
Sentiment at this stage is at a document (media item) level and does not account for the users’ brands/keywords individually.
No matter how science progresses, assigning sentiment to a document always involves some element of value judgement. When a client has a view that is different to the sentiment badge on a media item, the user can use the override button to change the view.
Why is sentiment analysis inside Mediaportal extra special?
In addition to providing automated sentiment analysis for social media items, we will be assigning automated sentiment across ALL media types.
By providing sentiment analysis across all content, we’re able to provide a holistic view. Users will be able to dive into what matters quickly, put out fires, celebrate and share positive feedback and be even more organised!
What other features will include sentiment?
In addition to showing sentiment badges on media item cards, sentiment is also included in the following features:
- Sentiment badge on media item cards for English content
- Sentiment badge on Korean, Simplified Chinese, and Traditional Chinese language content
- Filter on sentiment in Search and Personal Folders
- Group by sentiment option in briefs, folders and report builder
- Option to include sentiment in Mediaportal Dashboard and coverage reports
- Option to manually override the automated sentiment result on each item card
- Sort by sentiment
