How does YouScan determine post sentiment?
YouScan distinguishes between three types of sentiment: positive, neutral, and negative. The sentiment is assigned automatically by default.
YouScan's automatic sentiment detection technology relies on machine learning. The algorithm tries to interpret the meaning of a post similar to how a human would do it, and trains on data marked up by humans.
The sentiment is determined in relation to your monitoring subject in YouScan. The software identifies this subject using your topic search query.
For example, for a query that reads: (Samsung /5 phones) -buy -sell
The tool will evaluate post content as it relates to Samsung as its subject. So YouScan will assign a post that reads "Samsung makes great phones, while other brands make terrible ones" as positive sentiment.
What can I do if I see the wrong sentiment assigned to a post?
Send us examples (mention links), we will analyze them and improve the algorithm's accuracy. Make sure to also change the sentiment manually, so the algorithm can learn from your edits and constantly increase the accuracy of its sentiment analysis in your topic.
How can I manually change the sentiment of a post?
To change post sentiment, click the correct sentiment icon at the bottom of a mention. All changes will be reflected in Analytics and exported data.
How accurate is YouScan's sentiment analysis?
YouScan determines negative and positive sentiment with 90-95% accuracy.
☝️ Topics with several monitoring subjects will have worse overall sentiment accuracy. Try to keep your monitoring to a single monitoring subject (brand, company, person, event, etc.) per topic.
Which languages have automatic sentiment detection capabilities?
Arabic, Armenian, Azerbaijani, Bengali, Bulgarian, Burmese, Catalan, Chinese, Croatian, Czech, Danish, Dutch, English, Finnish, French, Georgian, German, Greek, Hebrew, Hindi, Indonesian, Italian, Japanese, Kazakh, Korean, Kyrgyz, Latvian, Macedonian, Malay, Malayalam, Mongolian, Persian, Polish, Portuguese, Romanian, Russian, Serbian, Slovenian, Spanish, Swedish, Tajik, Tamil, Thai, Turkish, Turkmen, Ukrainian, Urdu, Uzbek, Vietnamese.