Right Relevance (RR) provides curated information and intelligence on over 45 thousand topics with:
- Topic relationships including semantic information like synonyms, acronyms.
- Topical influencers (~2.5M) with score and rank.
- Topical content and information in the form of articles, videos and conversations.
Below is an example from the Right Relevance portal for the topic ‘climate change’.
Right Relevance Insights combines the above Topic and Influencer information with real time conversations to provide actionable intelligence with visualisations that enable decision making. Social media, esp. Twitter, is a rich source of information for events, topics, emergring trends that are of interest amongst the general population and news media. We leverage social media broadly for our case studies.
Communities & Influence
Identifying communities and determining influence of active parties that form the sources of data are critical to the Insights analysis.
Measuring influence is not deterministic. It’s a fairly subjective task with numerous different methodologies and is generally ephemeral in nature. Using graph theory, machine learning and natural language processing, RR discovers how people congregate to form communities that share common interests, within the context of an event (or topic or trend). We also determine influence within those communities, along longer and shorter timelines. At Right Relevance, we measure influence in 2 distinct ways:
‘topical influence’ or Tribes by measuring the quality of network connections within the context of a ‘topic’ and,
‘engagement influence’ or Flocks by measuring quality and quantity of engagements (RTs, mentions, replies), reach of tweets, connections etc. within the context of an event or trend.
For example, the public and real time nature of Twitter enables us to identify people who share long term similar interests (think football team supporters) and those who come together to share an interest about a specific event in time (think a specific game of football) – we call these groups Tribes and Flocks respectively.
Timing is everything. If you want to influence people to vote in a certain way, buy your product or understand your point of view you need to know when they are actively engaged in thinking about the relevant subject and, therefore, highly receptive to your message.
For a specific analysis we start by defining a Subject. A subject is a broader area of interest and can be made up of several Topics. It can reflect some aspects of one or more interrelated Topics. The following are examples of plain English definitions of Subjects.
The public awareness of the importance of environmental issues like Climate Change and Global Warming and the people and organisations who influence the public perception.
The spreading of misinformation about the Ebola Crisis.
The most interesting Business and Gender Rights stories coming from the Davos WEF Conference.
We then identify a specific objective e.g.
Identify people at the time they are actively engaged in a conversation which indicates that an activity is taking place e.g. going to a football match, voting in an election, buying a car, booking a holiday.
Track conversations about a subject to identify Influencers who are negatively impacting public perception of a brand or interest.
RR uses graph theory and algorithmic learning to analyse massive volumes of unstructured data, however, an overall philosophy and process is required to guide the process. For this, we adapted and implement the military intelligence process of OODA Loop which was designed to manage confusing, unpredictable and obscure environments.
OODA – Observe, Orient, Decide, Act
- Observe- Gather large volumes of data, so as not to miss anything.
- Orient- Make sense of the data at a high level. Use information to direct future observation. Repeat.
- Decide- Analyse the information gathered to make decisions about how to act.
- Act- Take action and measure outcome and effect.
Figure 1: OODA Loop
Visualisation of large active networks can provide extremely valuable insights to both guide the Observe/Orient phases and illustrate specific relationships.
For example these two visualisations show very different aspects of the Brexit conversation leading up the UK EU Referendum.
Figure 2: high level polarisation in ‘Brexit’ conversations
Figure 2 above shows the high level of polarisation (2-clear contrasting communities), influencers engaging within each community along with the relative size of each. This network is based on the relationships made just by Retweets. Despite what many people put on their profiles, retweets on the whole, do mean endorsement. People retweet what they agree with which gives us this very striking separation of the opposing factions.
Figure 3 and Figure 4 are maps showing more integrated network based on retweets and mentions. These are Ego Networks showing the connections of a two Twitter users – The Guardian and The Telegraph newspapers.
One can see that when mentions are included the amount of polarisation is reduced in the visualisation because people tend to mention other users in both positive and negative ways so there are more connections between opposing factions. In the Guardian ego network it is clear there is a more connections with both sides of the debate as opposed to the Telegraph which is very tightly connected to the Leave side.
The use cases for reaching these audiences is limitless. Here are some common examples:
- Real-time analysis of breaking events like elections, conferences, product launches, outbreaks like Zika, Ebola etc. Examples include:
- Discovery of trends, influential people and competitors in context of emerging technologies like Blockchain, IoT, Fintech, Docker, SDN etc.
- Identify opinion makers and drivers and understand direction and sentiment of conversations for risk management, PR and communications management. e.g. Climate Change