This API extract locations as well as the 8 main sentiments from text (trust, fear, anticipation, sadness, disgust, anger, surprise, joy) from which it derives the overall sentiment (positive/negative)
Input: "I love Boston for its great universities and standard of living. Some areas however, can be a bit rough"
Output: Region of search: US
Match Location
Boston,US Confidence Score: 0.9
Sentiment Analysis:
This text is: positive (+0.666666666666667)
great love rough
The full range of sentiments in this text is:
positive 0.4, anticipation 0.2, joy 0.2, fear 0, anger 0, trust 0, surprise 0, negative 0, sadness 0, disgust 0,
'I don't love Boston as some areas can be a bit rough'
shows as a positive sentiment
positive 0.333333333333333, joy 0.333333333333333, fear 0, anticipation 0, anger 0, trust 0, surprise 0, negative 0, sadness 0, disgust 0
Another potential issue for some people might be well known acronyms ..
example for input :
'I don't love UK as some areas can be a bit rough'
UK is not appearing in top 10 results also.
You are correct, it does not work so well with very short text. That's an area I'm improving on.
It could be spot on with longer text, especially with book excerpts, which is what I designed the system for originally, for eg: https://goo.gl/Vq4U2p
"Well, a relatively minor operation changed me into a black-skinned Terran, though the surgeon/replacers could do nothing, ironically enough in view of my new color, to increase my resistance to heat. I remember those stirring days of combat sometimes, usually when I am making my semi-annual flight between Churchill, Manitoba, and Tierra Del Fuego. In fact, during those flights when I am practically alone is the only time I have to reflect or remember, because on both of my estates there is nothing but noise, children, and wives."
Match Location
Manitoba,CA Confidence Score: 0.1
Churchill,CA Confidence Score: 0.1
Sentiment Analysis:
This text is: negative (-0.166666666666667)
ironicallyenoughwellresistancenoise
The full range of sentiments in this text is:
fear 0.1875, negative 0.1875, anticipation 0.125, anger 0.125, trust 0.125, positive 0.125, joy 0.0625, surprise 0, sadness 0, disgust 0,
For example: https://geocode.xyz/?scantext=I%20love%20Boston%20for%20its%...
Input: "I love Boston for its great universities and standard of living. Some areas however, can be a bit rough"
Output: Region of search: US Match Location Boston,US Confidence Score: 0.9 Sentiment Analysis: This text is: positive (+0.666666666666667)
great love rough
The full range of sentiments in this text is: positive 0.4, anticipation 0.2, joy 0.2, fear 0, anger 0, trust 0, surprise 0, negative 0, sadness 0, disgust 0,
positive 0.333333333333333, joy 0.333333333333333, fear 0, anticipation 0, anger 0, trust 0, surprise 0, negative 0, sadness 0, disgust 0
Another potential issue for some people might be well known acronyms .. example for input : 'I don't love UK as some areas can be a bit rough' UK is not appearing in top 10 results also.
It could be spot on with longer text, especially with book excerpts, which is what I designed the system for originally, for eg: https://goo.gl/Vq4U2p
"Well, a relatively minor operation changed me into a black-skinned Terran, though the surgeon/replacers could do nothing, ironically enough in view of my new color, to increase my resistance to heat. I remember those stirring days of combat sometimes, usually when I am making my semi-annual flight between Churchill, Manitoba, and Tierra Del Fuego. In fact, during those flights when I am practically alone is the only time I have to reflect or remember, because on both of my estates there is nothing but noise, children, and wives."
Match Location Manitoba,CA Confidence Score: 0.1 Churchill,CA Confidence Score: 0.1
Sentiment Analysis: This text is: negative (-0.166666666666667)
ironicallyenoughwellresistancenoise
The full range of sentiments in this text is: fear 0.1875, negative 0.1875, anticipation 0.125, anger 0.125, trust 0.125, positive 0.125, joy 0.0625, surprise 0, sadness 0, disgust 0,