In the past few days, I’ve had several great conversations in different forums that happened to share a common thread. “Is it possible or useful to analyze unstructured social data?”
I argue that it is possible, but it might not be useful. Let’s start with Analysis.
Analysis should be:
- Abstract – detached intellectual process based on a model of understanding, by definition simpler than the messy reality
- Objective – data-centric, fact-based. Does not involve opinion, beliefs, or inference
Analyzing something is gaining understanding of reality by taking it apart based on a simplified model that (hopefully) captures the essence of the subject.
Sentiment Analysis
One of the forums of discussion was about terms like “Socialytics” and “Sentiment Analysis”, in the context of trying to analyze conversations (or “unstructured interactions”) in Social Media.
We are not arguing about the structured meta data (place, time, participants, frequency, etc). We are talking about the real content of conversations.
To analyze, one needs to have a model of understanding. The crudest analytical model of understanding is: “I don’t understand, therefore it doesn’t exist”.
I agree that “sentiment” can be positive or negative or neutral (a valid model) and counting occurrences can give us some measure of the general average “sentiment”. Therefore interpreting content as positive/negative/neutral, in this context, can be called “Analysis”. But it feels to me almost as crude of a model as the crudest.
In my opinion, this model does not capture the essence of human interactions and opinions.
Social Business is not Statistical
Is reducing human interactions to uni-dimensional metrics truly useful? My analytical left-brain say yes, it is useful.
But I am cautious following that reflex. Social Business is about relationships, engaging with each person leveraging Social Media.
The risk is applying old patterns of averages and metrics to customer relationships and falling deeper into the meaningless excessive analysis of the past. Reducing conversations to “positive/negative” is ignoring they exist as individual human interactions.
Social Business models are exactly the attempt to get out of that hole.
The future is about analyzing less (or at least using models that more closely resemble the reality of human relationships). There is no shortcut to engagement. Let’s resist accepting the reductionist view that counting is the same as understanding.
Marcio Saito 