Sentiment Analysis Of Comments on Social Media

Social media comments analysis is the process of understanding the sentiments expressed in social media feeds to see how they affect a brand. Through sentiment analysis of comments, companies can use social media-based marketing and advertising to creatively build their brand spotlight. They can confidently leverage insights from a sentiment analysis API for data-backed holistic growth strategies that are customer-centric and compounding.

Why Do We Need Social Media Comments Analysis?

Any company that launches a new product, advertising campaign, brand ambassador, store location, or simply wants to know where it stands in the current market space compared to its competitors, needs to know how sentiment analysis of comments on social media is done. This gives it a heads up as to what to expect from a good social media PR standpoint, and also to see what it can do to improve itself.

In the current scheme of things where people have access to social media readily, not only are business-to-customer (B2C) companies leveraging the channel but also B2B companies that were once averse to it. Take, for example, Intel, with their highly successful Sponsors of Tomorrow campaign. The company realized that even though they were not directly affected by consumers on social media, social media-driven brand awareness is an advantage as an overall business strategy.

Overall, social media comments analysis can help businesses with:

• Brand intelligence
• Improving customer experience
• Better sales conversions
• Brand awareness
• Brand loyalty
• Competitor awareness
• Product enhancements; and more

How Is Sentiment Analysis Of Comments Done?

A machine learning-based voice of customer analysis platform performs sentiment analysis of comments in 4 stages.

Stage 1: Gathering comments from social media

Once you have decided the social media channels you want to analyze, it is time for the first step in the social media comments analysis process. In this stage, data is gathered in one of two ways, or both, if necessary. It is then annotated. You can gather data either through APIs or upload it manually.

• Data gathering through APIs — Social media channels usually have APIs through which you can download publically available data as with the Facebook API or Twitter
• Data gathering manually — The sentiment analysis platform also allows you to manually upload data as a .csv file

Stage 2: Processing the comments

Now that all the data is collected, cleaned, and labelled, it will be processed with several machine learning tasks so that all formats of data present in the comments are prepped for sentiment analysis. This is important because people often respond in comments with memes, gifs, images, and emojis. The ML algorithm recognizes the text in all these formats and includes them with the text from comments. It is now ready to analyze this refined data to extract entities, aspects, and topics from it.

Stage 3: Analysing the data

Now that all the data has been processed, the model needs to be trained so that AI tasks like natural language processing, named entity recognition, semantic analysis, topic classifications, and sentiment analysis of comments can advance.

• ML model training — A dataset pertaining to the industry the brand is from, is used to train the model. Another data set is used to validate the results and compared with the correctly classified data. The model is trained multiple times to ensure that the results are as accurate as can be.
• NLP for multilingual data — Natural language processing (NLP) tasks analyze data in each language separately. In the case of Repustate, the social media sentiment analysis platform analyzes data in 23 languages natively without translations. It uses a speech tagger for each language individually.
• Tags — Tags can be auto-generated or created manually for aspects, themes, and topics identified in the data. Thus, when insights are needed, for example, TikTok insights, custom tags will identify brand mentions, celebrity names, product names, etc.
• Topic classification — The topic classifier attaches a theme to a text, like price, healthcare, education, food, etc.
• Sentiment analysis — All the topics, aspects, and features are analyzed for sentiments that may have been expressed for them and sentiment scores between -1 and +1 are assigned. Eventually, all these sentiment scores are aggregated and the subject matter is assigned a positive or negative score anywhere between 1 and 100. 1 being the most negative and 100 being the most positive.

Which Platforms Are Accessible To Social Media Sentiment Analysis?

As long as a social media platform allows you access to publicly available data, you can gather that information for social media listening. Popular platforms from which you can gain sentiment analysis of comments are:

• YouTube

One platform that is flooded with product reviews, as well as marketing viral videos that garner hundreds of views and comments, is YouTube. A social listening platform enabled with video content analysis can access YouTube through its API and download the comments pertaining to a particular video seamlessly. In the case of Repustate IQ, all you need to do is paste the URL of the source video, and voila! You can now instantly see the sentiment analysis of all the comments under that video.

• TikTok

As digital agencies explore TikTok to reach a wider audience on the internet, it is the right time to invest in a social media comments analysis platform that allows you to scrape comments from these short-format videos. From political parties turning to TikTok, to non-profit organizations, to football leagues, organizations across industries are using the platform for growth. You can harness social listening data from TikTok to make sure that you too are ahead of the curve.

• Douyin

Unlike TikTok, which is still in the nascent stages for marketing, although incredibly popular for user-generated content, Douyin is actually used by brands to reach a more affluent customer base that is Chinese. Brands like MiuMiu, Guess, Prada and other premium companies are already using Douyin for campaigns. Getting a platform that can analyse Douyin can be a challenge because of China’s strict privacy laws, but you can leverage the channel through Repustate IQ, which has the capability to access Douyin video comments.

• Youku

Aimed at the Chinese customer demographic, Youku is a video platform exactly like YouTube. With China’s stringent policy of not giving permission to a number of non-Chinese digital apps and websites, it’s no surprise that the country has developed its own version of YouTube. With millions of users, Youku is a social media platform that allows an API to download information with regards to public comments.

• Twitch

Catering to gamers, Twitch has live-streaming videos, games, and other content that allow you to live chat with fellow gamers. Many companies use Twitch to advertise their product to this niche market — products like digital monitors, apparel, digital headsets, gaming consoles, ergonomic chairs, and others. Sentiment analysis of comments and chats to gain twitch insights on the platform can give you interesting insights that you can use for your own advertising strategies.