We have begun using machine learning to identify human emotions expressed in social media data, a technology known as sentiment analysis.
Twitter sentiment analysis tools enable small businesses to:
- See what people are saying about the business’s brand on Twitter.
- Do market research on how people feel about competitors, market trends, product offerings etc.
- Analyze the impact of marketing campaigns on Twitter users.
We’ll take a look here at a number of free tools for doing sentiment analysis on Twitter data.
Here’s what we’ll cover:
- Enginuity (Web App)
- Revealed Context (API/Excel Add-in)
- Steamcrab (Web App)
- MeaningCloud (API/Excel Add-in)
- socialmention (Web App)
- Data Mining Platforms with Sentiment Analysis Capabilities
NCSU Tweet Sentiment Visualization App (Web App)
Dr. Christopher Healey, Goodnight Distinguished Professor in the Institute of Advanced Analytics at North Carolina State University, has built one of the most robust and highly functional free tools for Twitter sentiment analysis out there: the Tweet Visualizer.
The NCSU Tweet Visualizer is particularly strong in the following areas:
Ease of Use
Simply enter a keyword, and the Tweet Visualizer automatically pulls recent tweets (from the past week, though the time range is shorter for popular subjects).
You can then explore the many visualization options that the tool offers for tweets.
One highly useful feature in the Tweet Visualizer’s visualizations is that you can mouse over bubbles in scatter plots etc. to pull up individual tweets from identified Twitter users and see where they fall in the emotional spectrum.
Sophistication of Sentiment Analysis
Healey explains that his sentiment dictionary “has three emotional dimensions for words:
- Pleasure (how happy you are).
- Activation (how excited you are).
- Dominance (how much does this particular term dominate the overall sentiment of the snippet of text it’s in).”
By measuring pleasure, activation and dominance, the NCSU Tweet Visualizer offers far more dimensions than can be found in many other free sentiment analysis tools. Most of these tools only focus on the “pleasure” dimension and rate sentiment according to a three-value scale: positive, negative and neutral.
By contrast, Healey notes that “our scales run on a nine-point range, so we have a semicontinuous representation of sentiment. Whereas a lot of systems will just say that a text is positive, negative or neutral, we can actually say how positive, how negative etc.”
The “activation” dimension sounds somewhat odd, but Healey observes that it has an important use:
“Suppose I’m very pleased about something and very activated about it: the kinds of words we’d use would be like ‘elated’ or ‘excited.’
Now suppose I’m equally pleased about something but I’m very low on the activation scale. In this case, we’d say that I’m ‘calm’ or ‘relaxed.’
If we only looked at the pleasure scale, we wouldn’t be able to differentiate between ‘excited’ and ‘relaxed.’ The activation scale allows us to do this.”
Topic Clustering Ability
Finally, the Tweet Visualizer doesn’t just handle sentiment classification, it also performs topic clustering.
In other words, it automatically clusters tweets into related topics by leveraging machine-learning algorithms. (See our in-depth explanation of clustering algorithms for more details on how this type of machine learning works.)
The tool combines sentiment analytics with topic clustering to help you understand how people feel about particular topics:
Enginuity (Web App)
Enginuity is a paid solution, but a basic version is available as a free web application. It works differently from many of the free sentiment analytics tools out there. Instead of directly querying tweets related to a certain keyword, Enginuity allows you to search for recent news stories about the keyword.
The tool then queries both Twitter and Facebook to calculate how many times the story has been shared. It also analyzes whether the sentiment of social shares is positive or negative, and gives an aggregate sentiment rating for the news story.
Enginuity is thus a great tool for finding stories to share through your social channels, as well as getting a combined picture of sentiment about recent events trending on social media.
Revealed Context (API/Excel Add-in)
Revealed Context offers a free API for running sentiment analytics on up to 250 documents per day. There’s an Excel add-in as well as a web interface for running analytics independently of the API.
While Revealed Context doesn’t offer an interface for directly scraping Twitter, it’s simple enough to analyze a spreadsheet of tweets without using the API. With the API, you can build a pipeline that feeds recent tweets from the Twitter API into the Revealed Context API for processing.
Steamcrab (Web App)
Steamcrab is a web application for sentiment analytics on Twitter data. It focuses on keyword searches and analyzes tweets according to a two-pole scale (positive and negative). Visualization options are limited to scatter plots and pie charts.
MeaningCloud (API/Excel Add-in)
MeaningCloud is another free API for text analytics, including sentiment analytics. One of the advantages of MeaningCloud is that the API supports a number of text analytics operations in addition to sentiment classification. These operations include topic extraction, text classification, part-of-speech tagging etc. (See our article on text analytics if you’re not familiar with these operations.)
The MeaningCloud API is more flexible for use in topic extraction than the other solutions we’ve considered, since with other tools topic clustering is performed automatically according to the initial keyword you enter.
Additionally, MeaningCloud allows users to upload custom dictionaries for use in topic extraction and sentiment classification.
MeaningCloud offers an Excel add-in, but it doesn’t work with Excel for Mac (a problem with many Excel add-ins).
socialmention (Web App)
Socialmention is a basic, search engine-style web app for topic-level sentiment analysis on Twitter data. You can enter a keyword, and the tool will return aggregate sentiment scores for the keyword as well as related keywords.
One neat feature of social mention is support for basic brand management use cases—the tool returns a “passion” score that measures how likely Twitter users are to discuss your brand, as well as the average reach of the Twitter users discussing your brand.
The caveat is that the tool still returns wonky results for lesser-known brands, but this is an issue with sentiment analysis in general.
Data Mining Platforms With Sentiment Analysis Capabilities
Open-source data mining platforms offer some of the most advanced support for text and sentiment analytics out there essentially for free.
Solutions such as RapidMiner and KNIME have built-in sentiment analysis modules as well as a host of third-party modules. We explain how to use data mining platforms for sentiment analytics in our article on free text mining tools.
Nearly all small businesses understand the value of social media marketing, but few have the tools to analyze the impact of social media marketing efforts. This is because currently, such tools are priced for Fortune 500s.
With free tools for sentiment analysis, however, you can begin understanding how your Twitter marketing efforts are performing without any investment, save for your time. Additionally, you can begin monitoring Twitter for signs of problems (e.g., customer complaints about your brand) as well as wins (e.g., things customers like about your brand).
If you have questions about how to begin using sentiment analytics at your small business, you can email me at firstname.lastname@example.org.