F5.5G Leap-forward Development of Broadband in Africa The Africa Broadband Forum 2024 (BBAF 2024) was successfully held in Cape Town, South Africa recently, under…
Text mining reveals best and worst words to use in social media
A recent study conducted by SmartData Collective pinpoints the most frequently used and popular words in social media — and those that are most likely to get you retweeted.
The company used a system of predictive analytics that employed text mining to analyse more than 3 000 posts and identify the most important subjects. Text mining refers to the process of gathering high-quality information from written text.
SmartData Collective’s first task was to identify the most frequently occurring words in popular posts. The first round of research lead the company to identify words like “social”, “media”, “Twitter” and “Facebook” as the highest ranked words in popular posts.
Unfortunately this initial list of single words did not offer much insight, so a list of words that appear most frequently adjacent to each other was created, called “2-grams”. This list offers more insight into the minds of social media users and possibly illustrates how a user can increase the popularity of their social network and blog postings.
Top ranked 2-gram words include: “how to”, “ways to”, “media strategy” and “social networking”.
SmartData Collective concluded that “How to’s” were the most likely way to improve the popularity of a user’s posts and achieve higher click-through rates.
According to the research, one could dig even further to identify the topics that were the most successful such as: “How to monetise”, “How to be successful” and “How to avoid mistakes”. These topics have the absolute highest popularity rates and will most likely secure an article or post the most users, but so long as the information remains fresh. Just re-using a stale old “How to” will end up doing more harm than good, so topics have to be well thought through.
Finally, SmartData Collective set out to find words and phrases that were most commonly found in posts with a high number of retweets. Text mining was used once again and the following factors were taken into account:
- Author of the post
- Title of the post
- The number of retweets achieved on Twitter by the post.
The graph shows words with a negative impact on the number of retweets weighted to the left and include words like: “write”, “talk”, “trust” and “sentiment”. Words like “launch”, “America”, “personal” and “link”, shown on the right, led to the highest number of retweets.
The company analysed the data and drew the following conclusions.
- Personal Branding appears to be one of the hottest retweeting topics on Twitter. People tend to be primarily interested in the various ways they can increase their appearance of “personal worth” in their social media circle.
- IWOM: Internet word-of-mouth is another concept that frequently plays a role in social media posts with many retweets.
- Positive & Possible: It appears that posts which discuss various possibilities in a positive way and use the word could were found to be re-tweeted more than negative, depressing posts.
In a nutshell? Talk about yourself as a positive, successful American and offer advice on how others could become positive, successful Americans and you could be sitting on a retweeting goldmine.