Semantic search company TextWise used their Semantic Signatures technology to analyze the content of Twitter messages. They used the streaming API to download a sampling of 8.9 million tweets posted by 2.6 million unique users.
They found that 2.7 million of these tweets, or 31%, were replies to a tweet posted by another user, half a million (6%) were retweets, and almost 2 million (22%) of the messages contained a URL.
They further analyzed the content of 1,000 tweets to see what the content was and found that 30% of tweets relate to a user’s “status” (what they’re currently doing or where they are), and 27% were private conversations. 10% contained links to articles. 4% of tweets analyzed included product recommendations or complaints — a small percentage but a huge amount of data if you consider that Twitter is now logging more than 50 million tweets a day.
See the details at SemanticHacker Blog.
