Skip to main content
All CollectionsMentions
Sentiment Analysis in Mentionlytics
Sentiment Analysis in Mentionlytics
Eva avatar
Written by Eva
Updated over a week ago

Sentiment Analysis is an important metric in online and social media monitoring. It helps you detect the attitudes of the people mentioning your brand.

Sentiment analysis is most often used in reputation and brand management as well as crisis detection and management. However, it is also a valuable tool that gives you an idea about the attitude toward your keywords and the general mood of your audience.

The sentiment is classified as positive, negative, or neutral, and, in Mentionlytics, it is color marked as follows: red for negative, green for positive, and grey for neutral.

Sometimes, you may notice that the sum of the percentages of these 3 classifications do not add up to exactly 100%. This is because sometimes a mention could not be classified to any of the above, so the sentiment is unknown. An example of this could be a photo post on Instagram without any text to classify.

How is the sentiment measured?

Sentiment is measured using different methods among which are dictionaries, taxonomies, and scales. For example, a simple way to determine sentiment would be to use a dictionary with negative, positive, and neutral words that each post is compared with. The algorithm that runs through the mentions determines if the post contains these words and marks it accordingly. This gives you an idea about the overall sentiment of the post.

Does sentiment analysis work in all languages?

Yes, Mentionlytics performs sentiment analysis in almost all languages using state-of-the-art deep learning algorithms. It is extremely rare that a language is not supported by our system. If however, you cannot see sentiment analysis of your language in our mention results, please talk to us, and we will most certainly resolve the issue.

What if the automatically detected sentiment of a mention is not correct?

Sentiment analysis is an exciting technology, allowing you to measure the sentiment of vast amounts of text without needing tens of work hours to go through each mention. That said, no one can claim that automated sentiment is 100% correct. The accuracy can vary, depending on the context of your account. A safe average could be about 80% accuracy, which is great for most use cases to clearly see the trends towards a topic or brand.

The sentiment may not be precise when it comes to cultural jokes or irony. In any case, you can easily change the sentiment of a mention by just clicking on the sentiment icon to rotate through the 3 available options.

We have several functions to make the manual sentiment change easier. For example, when you change the sentiment of a Tweet it will automatically change the sentiment of all Retweets, even the ones discovered before or after your change.

Although the algorithm does not "learn" from these changes automatically, you can create automated rules to be applied to any newly discovered mention, that would do the change automatically. For example, you could create a simple boolean query that would apply a specific sentiment when a specific profile posts something, because you know they are always negative towards a brand. Also, you can specify specific keywords that you know when they exist in the text, they always have a positive sentiment towards your brand. Talk to our team if you are not sure about how to apply these rules, and we will happily advise you and help you with the setup!

Where do I see the sentiment analysis in Mentionlytics?

The sentiment is available under each post and is marked with a face emoji in the respective color. The sentiment is changeable by clicking on the face emoji and selecting the correct color.

You can use the filters on top to filter the mentions according to sentiment. This way you can see, for example, the negative mentions of your keyword for a given period.

The graph in the Overview traces the changes in sentiment along a given period and shows the peaks of positive and negative mentions. If you click on each of the dots in this graph, you will be able to see the conversation that has triggered the peak.

Under the Top keywords menu, you can see different keywords and topics and their overall sentiment regarding your main keywords (brand name, competitors, etc.)

Under Top Mentioners you can also find the leaders of conversation mentioning your main keyword (e.g. brand name) and their attitude towards it.

If you would like to stay on top of the negative mentions, you can set up notifications and Mentionlytics will send you emails when there is an increase in negative mentions.

Also, in the share of voice report, you can see the share of positive, negative, and neutral mentions compared to your competitors

If you have additional questions or need assistance, please contact our support team.

Did this answer your question?