With over 2 billion social media users in 2013 and over 500 million tweets per day, social media is exploding. The importance of social media is also undeniable, from breaking news to consumer’s reaction to new products, understanding and responding to social media chatter is critical. However, social media is, by its very nature, an unruly and noisy platform. It often seems like everybody is shouting at once and each post/tweet is only a fragment in a larger unfolding story. Multiple stories evolve and are intermixed, and people use different language to talk about the same thing.
In brand management and content marketing it is important to find social media conversations about your brand, product or industry, respond to reputation issues and frame marketing messages. These conversations are often hidden in the noise on platforms such as Twitter - you need to find the small set of people (the ‘influencers’) who have the largest impact on how your brand is seen by social media users. Social media also provides a unique advertising platform, but finding the right audience amongst the noise is difficult.
With the speed of social media, platforms such as Twitter are often the source of breaking news, with the most famous example being @ReallyVirtual’s tweets revealing the Bin Laden raid. But again, locating one person’s tweets among 500 million tweets is like finding a needle in a haystack, especially when that haystack is constantly being tossed around.
Traditional social media listening platforms try to deal with finding the relevant information amongst the noise through complex search queries. These queries use boolean operators such as AND, OR, NOT, NEAR to link keywords. With the ever increasing noise and the unique language of Twitter (abbreviations, hashtags, distortions of words), these search queries are ballooning in size, often to 100s of keywords and pages of boolean operators. These queries are difficult to create (and are often created by experts or consultants) and require constant maintenance due to how quickly language on Twitter evolves.
narratif Discover is a new social media platform designed to provide awareness of what you don't know, that has the potential to completely change the way marketing creatives, journalists, bloggers, and PR agencies use Twitter. Rather than users having to focus on specifying long, complex keyword search queries Discover automatically generates, in real-time, a map of keywords (called the semantic schema) from just one or two keywords which specifies the general area of interest.
For example, a search for Google automatically expands the search to include other terms closely associated with the term Google that exist on Twitter now. Today this includes the terms Waze, Lunar X-Prize and WikiLeaks. These relationships between terms are not hardcoded into Discover, but are actually automatically discovered from the real-time data itself. In the ever changing world of Twitter, this is important as today Google may be associated with WikiLeaks, but tomorrow it may be Facebook that is associated with WikiLeaks. A static set of connections just won’t work on social media.
Using the semantic schema has two key advantages over keyword query based listening:
- This large set of keywords casts a very wide net and returns a large number of tweets, many of which do not contain the original Google search term but are still relevant to your topic of interest.
Results from narratif Discover provide complete coverage as well as much needed context.
- As Discover knows how the large set of keywords are related to each other on Twitter, it is also able to automatically group the results into stories (or sub-topics or conversations). For instance, a search for Google gives a number of stories, including one about Sherrif’s wanting Google to turn off Waze’s police tracking feature, and one about WikiLeaks demanding answers after Google handed over emails to the US government.
By separating results into coherent stories, narratif Discover makes it easy for a user to understand all the stories happening in their area of interest at a glance, then zoom in and monitor stories which are of interest while quickly discarding the irrelevant.
The technology behind narratif Discover was developed at The University of Queensland by Dr Andrew Smith. Andrew is a co-founder and Chief Scientist of narratif and has over 14 years of experience in text analytics. The genesis of narratif came from Andrew realising that text contains many different worldviews, and this is especially the case in social media where there are so many different connections and views on important events, you need a system that acts like a prism, which can showcase the different colours of conversation. narratif’s platform is that prism.