Since the first day we started working in the world of online influencers, we have STRONGLY believed that “influence” simply couldn’t be defined in a vacuum.  We have always said that Influence is ALWAYS contextual.  Just because someone is influential when it comes to web technology does not mean he or she is influential when it comes to organic farming or any other topic.  Danny Brown wrote a good post on this topic last week, but we’ve believed it since Day zero at TRAACKR.

That is why, in our scoring engine, the most important metric we use when assessing someone’s “influence” is his or her RELEVANCE to the topic being searched (the topic being defined by our users).  (see here for more detail on our scoring practices)

RELEVANCE CALCULATION

Up to this point, we have used a bunch of different factors in our RELEVANCE calculation.

The most important factor in our calculation is the FREQUENCY at which the influencer uses the keywords in the search.  This means that a person who mentions any of the keywords 50 times in the previous 3 months will have a higher Relevance score than someone who mentions any of the keywords just 30 times in that same period.

Simple enough.

The second major factor in our calculation is the relative WEIGHT of the keywords.  Current TRAACKR users know that when you set up a search in TRAACKR, you can “prioritize” the keywords based on how important they are to you & your search.  Our scoring engine weighs the keywords based on where the user has “ranked” them — the higher the keyword in the list, the more weight it will carry in the Relevance score.

This is a bit more complicated and adds another level of sophistication to our Relevance scoring.

There are also several other factors we take into account in our current Relevance calculation.  They include:

• Keyword placement (a match in the title of a post counts more than a match in the body of the text)
• Ratio of keyword to text volume (1 keyword hit on a 5-line blog post matters more than the same keyword hit on a 500-line post)
• Rare keyword weight (less common words “Metastatic bone lesions” get more credit than more common words like “cancer“)

DIVERSITY INDEX

And as of next week, we will officially include a third major element to our Relevance calculation – what we call the Diversity Index.  This index will add yet another layer of sophistication to our Relevance calculation.  What the Diversity Index does is give more ‘credit’ to influencers who use a wider variety of unique keywords in their content.

As an example, let’s assume that you have a search with 5 keyword expressions:

• “small business” marketing strategy
• “small business” local marketing
• “small business” “online advertising”
• “small business” “pr”
• “small business” discounting strategy

Before the the Diversity Index, an influencer was given a Relevance score based on the number of keyword matches they had – it didn’t matter if all of those matches came from a single keyword or all different keywords.  But this just didn’t reflect the reality of any person’s relevance to a topic.

The theory behind the Diversity Index is that someone who uses a wider breadth of keywords in any search is more Relevant to that conversation.

By measuring the FREQUENCY of keyword usage, we can measure someone’s DEPTH in a conversation.  By measuring the DIVERSITY of their keyword usage, we can measure their BREADTH in a conversation.  And both matter.

With the Diversity Index, someone who uses more unique keywords will receive more credit (ie – a higher Relevance score) than someone who used fewer unique keywords.

This may sound like a simple, rather obvious refinement, but it is actually a big deal.  It’s a big step forward in the assessment of an individual’s Relevance to any specific topic – which is something that is not happening anywhere else on the web.

It’s just another reason why we think TRAACKR will be one of the hottest social media tools in the market in 2011!  :)

If you have any questions, let us know in the comment section.  Thanks!

DS

Tags: diversity index, relevance score, scoring influence

This entry was posted on Friday, December 3rd, 2010 at 11:33 am and is filed under This We Believe. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.