How is sentiment calculated or allocated? Can it be overridden?
Radarr uses natural language processing (NLP) to identify, calculate and allocate sentiments to each and every post that is captured on the platform across platforms online and offline. Machine learning is applied to the allocation of sentiments and based on user input, accuracy of sentiments can be improved as well overtime.
Each post is either marked as positive, negative or neutral and involves the algorithm studying the combination of keywords and phrases within each post to mark it as appropriately. Depending on the language, Radarr assures a sentiment accuracy rate of 85%. Posts which are often categorized tend to be conversations that have sarcasm or irony involved. The accuracy of these, however can be improved overtime by user inputs which Radarr has incorporated by way of provisioning users to override machine/system marked sentiments. By allowing this, the machine algorithm takes user inputs into consideration and gradually improves its marking in tandem to user inputs. Users can override system marked sentiments by clicking on the sentiment indicator icon located at the top right corner of every post and select another sentiment as appropriate.
Posts that are manually remarked for sentiments can also be quickly located by clicking on “Sentiment Type” in the filters. This allows users to quickly identify and locate posts that have had their sentiments manually altered. This sentiment allocation and identification applies to all the 104 languages Radarr covers as well. To view the list of languages Radarr covers, click here.
The sentiment of posts can be easily overridden. Below are a few steps on how you can do the same –
- Head to any of the conversation.
- You should now be able to see a feed of posts
- Each post has a round sentiment indicator icon located at it’s right topmost corner indicating either a positive, negative or neutral sentiment
- You can override the sentiment by clicking on this indicator and choosing the sentiment that best matches the post
Please note that all user inputs will impact the sentiment marking accuracy overtime.
Reach us at email@example.com for any assistance.