Symplur Signals Sentiment Analysis
The sentiment analysis provided in Symplur Signals is powered by a natural language processing (NLP) algorithm optimized for healthcare. This proprietary algorithm extracts subjective information from healthcare conversations on Twitter in order to determine polarity about specific healthcare topics. The method used for determining sentiment employs a scaling system for the three classes of neutral, positive and negative sentiment. This method opens up the opportunity for Symplur Signals users to put their own proprietary customizations on the algorithm for the benefit of their own clients.
Symplur Signals takes sentiment analysis a step further by enabling users to focus on the sentiment specific to the 15 different healthcare stakeholders (doctors, patients, caregivers, etc.) identified in our system. This together with our content filters, allow users to understand concepts the way they related to specific environments and stakeholders.
When choosing the report setting Recent the most recent X tweets will be analyzed. This is the fastest option. You can also analyze the whole dataset and sort the output by most Negative and most Positive tweets. This option can be slow on large datasets since every individual tweet need to be analyzed.
The graph visualizes the overall sentiment of the data set shown in the Data Table.
Our NLP sentiment algorithm is based on two healthcare optimized dictionaries, one for positive words and one for negative words. Each word in the dictionaries has a weight from 1 to 5, where 5 has most weight. The algorithm also interpret text based emoticons like “:)” as a sentiment signal.
- Postitive. The cumulative weight score of all positive words in the tweet.
- Negative. The cumulative weight score of all negative words in the tweet.
- Interactions. Number of retweets and replies the tweet has received.
- Comparative. The overall sentiment score divided by number of words in the tweet (certain stop words and mentions are ignored).
Customization of Sentiment Dictionary
- Do you want change a word from negative to positive? Just select the color-coded word and change the sentiment value to any point in the -5 to +5 value range.
- Do you want to add a completely new word to the dictionary? Simply double-click the word and pick a sentiment value.
These changes are effective for you and all the team members on your account, but not to other customers of Symplur Signals.
Users with the Admin and Analyst roles can add and remove customizations done by team members on the Sentiment page under Account Settings.
- Recent. Most recent tweets (fastest).
- Positive. Most positive tweets.
- Negative. Most negative tweets.