Testing out ToneAPI from Lymbix

Earlier today I created a simple app to test out the ToneAPI from Lymbix. I’m very interested in sentiment analysis  and think tools such as ToneAPI are very powerful.

Its easy enough to use tools such as Seesmic to watch out for mentions of your brand name on Twitter but I think its even more useful if you monitor the tone of the messages too. If the tone of a message is negative, perhaps one member of staff might be better at responding to it than another. Likewise, you could monitor the tone of messages about your brand over time to see if you’re doing any better or worse.

The app I built allows you to search Twitter showing the most recent 100 mentions of your search term. Using the ToneAPI, any Tweets that are positive are marked in Green and any Tweets that are negative are marked in Red. Try it out, search for something on twitter and see peoples moods.


I’m only using a single piece of information that is returned by the API. Some further refinement could be made by taking into account some of the other variables that are available. More than showing just positive or negative its possible to determine if the user is expressing gratitude, grief or even loathing.

Sample Response from ToneAPI

[ignored_terms] => “”
[affection_friendliness] => 0.47
[enjoyment_elation] => 2.2
[amusement_excitement] => 2.05
[contentment_gratitude] => 2.54
[sadness_grief] => -0.25
[anger_loathing] => -0.42
[fear_uneasiness] => -0.34
[humiliation_shame] => -0.26
[dominant_emotion] => contentment_gratitude
[average_intensity] => 3.26
[article_sentiment] =>
[sentiment] => Positive
[score] => 0.85
[coverage] => 36
[intense_sentence] =>
[sentence] => “”
[dominant_emotion] => contentment_gratitude
[intensity] => 0.32
[clarity] => 61.52

After building this, I got to thinking that maybe social media APIs could be used to help prevent spam.. (another blog post)

Leave a Reply

Your email address will not be published. Required fields are marked *