What you often obeserve in implied volatiltiy are higher levels of implied volatility for upcoming events like earnings or presentation of pharma data. For a human being which collects manual the information it is possbile to interpret this "spikes" or higher levels in implied volatility. Higher implied volatility in a single stock could also be observed if the whole market is facing higher implied volatility.
Is there a model or way you suggest, to decompose implied volatility to a market driven component and an idiosyncratic component?