A quick update from our project around reading and developing plings activity instances from the national Family Information Direct (FID) aggregation platform (we call it pling-o-matic!). All is going well in terms of data development, and we are on course for some good results, building on our initial feasibility study. In doing this, we have spotted a couple of (minor) interesting data issues that we wanted to share – all around the time fields of the records from FID.
Start time only
1 – Records that are missing an end time A few records – of good quality otherwise – state a start time, but no end time. For our testing, we are starting to use a one hour duration on these records, so we can convert to plings instances.
Broad time only
2 – Records that have general times Mornings, afternoons, evenings. We have isolated records that have good quality, but a very general indication of the time. This presents an interesting dilemma for plings – in that our data is always built around being time-specific. We could make a guess with these records, although we would not be 100% confident that it would be correct. Far from it. To us, these issues start to highlight some of the implications of converting directory data into activity events. Of course, this is to be expected. The solution lies in improving the data at source, but we appreciate that people may not always have this available. We hope that through this work, and eventually illustrating the benefits and results, we can then make a stronger case.
