Combining Topic Aggregation from BlogPulse with Radian6 and SM2/Techrigy – going one step further

Posted by Marshall Sponder on June 26, 2009 | Link It

One of the lacking features of Radian6 and SM2/Techrigy – both leading Buzz Measurement tools, they can’t yet, automatically put information into “buckets” – you, the person monitoring current trends, has to do that – and it takes some skills.   But … there are some shortcuts – and I found them (I guess that’s why you read my blog).

Shortcut one - Use BlogPulse Key Phrases to analyze a contemporary event – and put stories into “buckets” as in the  case of Michael Jackson‘s Death yesterday.   I’ve spoken to Marcel LeBrun (Radian6 founder) and Aaron Newwman (Techrigy Founder) in the past about the need for Topic Aggregation (some would call it “Topic Normalization” such as Twitter Trending Terms – Could Do Better from Matthew Hurst where he points out that “There is no attempt at normalization (RIP MJ == RIP Michael Jackson).

Looking at the top phrase for yesterday (I guess this shortcut doesn’t help you if you don’t have a big story event that got all over the news – too bad Blogpulse doesn’t offer a tool for keying on the key phases for an event – oh well, you can always wish for it – or go, create it, yourself)

A number of the top terms are about Michael Jackson into Radian6 and SM2/Techrigy to compare the results and see the strengths of each platform.

I also noted my searches on Radian6, especially the Influencer Widget, executed much, much faster (about 200 times faster) within a few seconds, when I did this, than, when I just made up keywords of my own – Interesting!

Let’s take a look at what Blogpulse does with the “paramedics arrived phrase

Somehow, and I don’t know the details, but would like to – Blogpulse took all those stories about Michael Jackson and said …. they’re also about “paramedics arrived”.    Now, you might ask … why isn’t a Topic Cloud for June 25th, the day Michael Jackson died, yielding the same thing?   The answer is – the Topic Clouds that Radian6 and SM2/Techrigy provide are for a single word, the Topic cloud Radian t provides is simply an index and ranking to every piece of information about stories around Michael Jackson – but not the “concepts” behind them – Blogpulse provides a multi word phrease and is therefore, close in meaning to the way humans think and describe content.

You can click on any one of the words in Radian6′s topics, but they are not a “bucket” term – you can throw a bunch of stories into “cardiac” or “death” and get much that’s meaningful  – you might get away with it with “Jackson” but what would “Michael” or “heart” or “attack” do for you?  All it does is point back to a bunch of stories – but with out that multi phrase “bucket” Aggreation, as I call it, or Normalization, as Matthew Hurst calls it – your lost – you can’t get to magic land – to the Holy Grail of meaning.

SM2/Techrigy Author Tags might bring you a tad bit closer – by looking at what the authors of content being indexed is categorized as, but it doesn’t have the intelligence to learn new concepts, or bucket information past what the author has already done.

So .. now that I explained what I want to see – how does each platform do, once I use the “bucketed” terms that I came up with from Blogpulse?

New Influencer Widget – Radian6

To be totally honest – I’m leaning towards creating a seporate profile or influencer widget for each “term” that Blogpluse supplied me with – but even here – because I could drill down deeper to the meaning of the content that is behind each phase, I’ve accepted that, for the purposes of this post, I will use all the terms about Michael Jackson that happened on his death yesterday, in one profile.

I can use the Influencer Widget to find out who to contact and how to contact individuals who have a larger say on what opinions get expressed about Michael Jackson’s death – and also sort by Compete data to find out how many visitors each source has (as I covered in a post the other day on Radian6).

Jumping over to SM2/Techrigy, I get the age/sex of content creators (but the content each platform picked up around Michael Jackson might not be identical and probably isn’t).

Brand References – Positive and Negative – is tricky to do, as you might have noticed, if you read my posts aabout SM2/Techrigy and the Iran Election – it works better for some subjects than others

We don’t know what the “positive” or “negative” opinion is about – the tools aren’t that smart yet – I suspect that is why Radian6 stayed away from trying to do this – but it will work better for simpler kinds of subjects like “chocolate” – where it’s easier to pick up that someone likes, or wants something, vs, doesn’t like or want it.

There’s a lot more I want to say about Topic Aggregation – I didn’t really find a good way to do it 100%, though I did speed up my processing of the query data – and I feel, that were Radian6 able to supply Topic Aggregation for any profile, not just top news stories, for example, and then plug that into the influencer widget – a “bucket” at a time (ie: do a influener widget just for “Michael Jackson and paramedics arrivedyou could then go a few steps farther in analysis of current and “pop” events by getting a “handle” as Eugene T. Gendlin wrote about, many years ago, in his groundbreaking book on “Focusing”.

What Gendlin came up with – a human being is a complex set of feelings, some feelings and ideas push others – but in order to understand them, you need to find the “handle” to the idea, so you can “manipulate” and move the feeling.

In analysis of comtemporary events, the Social Media Buzz Monitoring – soon to be morphing into Social CRM and Web Analytics tool platform, should also be able to provide the “handle” that allows a user of Radian6 or SM2/Techrigy, the ability to “summerize” all that news that Radian6 is picking up, for example, about Michael Jackson and say – yes, this information is really all about “Michael Jackson and his Cardaic Arrest at his home, and here’s the influencer and sentiment for that “bucket” of stories.

This is much as the human mind and heart does – when it overlays meaning onto events – that in themselves are meaningless – that’s what Paul Cezanne did with Mount St. Victorie, when he painted the mountain and saw something pure in it, that no one else had been able to quite express – it’s the ability to find meaning out of chaos, and I think Blogpulse, with it’s Key Terms, moved one step closer to what I asked both Marcel LeBrun and Aaron Newwman to provide – Topic Aggregation of information – tell me what it all means – tell me what Radian6 thinks all these stories mean, that’s what I’m looking for.

Reblog this post [with Zemanta]



Post comment as twitter logo facebook logo
Sort: Newest | Oldest

Trackbacks

  1. [...] I don’t trust sentiment analysis that’s automated – it might work OK for some things like “chocolate” (tried that with SM2) but not for more complex analysis such as the “Iran Election” – the emotions are too complex, for machine analysis – SM2 tries, but Radian6 stays away from this kind of analysis because, as is pointed out – it’s too unreliable (see PDF2009 Workshop on “Social Media and Iran” with Katrin Verclas, Davar Iran Ardalan, John Kelly, Olivia Ma, and Nancy Scola and Combining Topic Aggregation from BlogPulse with Radian6 and SM2/Techrigy – going one step furt… [...]





UPCOMING SPEAKING

The inaugural Social Media Analytics Summit is the first ever two-day business conference with a complete focus on social media analytics. Social media analytics enhances customer service, improves brand and reputation management, and measures overall social media success for businesses