Just before Christmas, Pete Nickless of 1up design combed through the data available in the Parent Know How Database (PKHD) and compared it to the data available in Plings. Pete was searching for positive activities contained within the broader PKHD data. And there were some to be found. As Plings.net moves to provide a national service, being able to draw on the data within PKHD and to turn descriptions of provisions into real activity records, could turn out to extent the information we can offer to young people.
You can read Pete’s full feasibility study here (PDF). As you will see, whilst there is positive activity data to be found in PKHD, it’s not always in standard formats, nor is there a uniform amount of information available across the country. However, being able to find perhaps 50 key positive activities for a local authority are is better than nothing, and provides a start to a broader Plings service.
So, Pete and his team are hard at work now building a prototype tool to integrate PKHD data with Plings in some of the Plings pilot local authority areas, to evaluate in practice the possibility to linking the two systems.
Making the links, however, does raise a number of key issues:
- PKHD describes providers and their services, Plings describes activity instances. The prototype import system uses a complex mix of regular expressions to try and locate activity, date and time information on the free-text activity descriptions of PKHD.
Because different system suppliers offer different interfaces for uploading information to PKHD, and because local authorities collect information in different ways, accomodating all the different ways information is expressed can be a challenge.
- When we turn a description like ‘Badmington every Friday at 5pm’ into instance data (i.e. a set of entries on the calendar), how far ahead in the future should we project the activity? We’ll be checking PKHD every day for updates – but if we only project forward one week, people browsing for things to do in a couple of weeks time will miss out on discovering an activity that is very probably taking place.
And what should we do about holidays? If a record doesn’t specify – how do we work out whether to display it during school holidays or not?
- How do we handle assertions about the safety and security of an activity. Right now, Plings activities come from trusted local authority partners, and we link activities back to those partners. However, with activities coming from a wide range of sources, and with PKHD containing basic information about the extent to which the authorities responsible for services have checked the information is correct, we need to think about how to carry over these indicators or ‘trust’ and ‘assurance’ into Plings.
If you’ve been exploring any of these issues in your own work with PKHD or other systems – then we’d love to hear from you. Get in touch via the comments below…
Hiya Tim,
Just thought I’d say Hi. I’m responsible for the development of the open source metadata aggregation platform that sits underneath the PKHD system. Although it feels like I’m a million miles away from PKHD propper, I’m about as familiar is it’s possible to be with the PKHD system in a soup-to-nuts way (Don’t blame me for the rubbish bits, although in honesty it’s probably my fault
)
Many of the issues you raise here have been banded around the various technical working groups for a long while no. They often play off the need for accurate information against the ability of LA’s and system vendors to provide that information. There are ways to modify, improve and extend the schemas, but no ways unfortunately to ensure that data is provided, or that the spirit of the schemas will be adhered to.
your info about converting free text descriptions into instance data is really interesting. Is that out in the public domain? Certainly, once thing we could do is automatically apply metadata enrichment routines to the records uploaded into PKHD… Is that something that would interest you? Also really interested in the question “How far ahead should we project”. We’ve (At times) asked ourselves how long a record should last in PKHD. If it’s not touched for 18 months, should it be deleted etc? Seems to me that a shared understanding of this question would be really helpful.
Also, with my other hat on, I’m one of a small team that organises grass-roots events in sheffield and south yorkshire (monthly GeekUp events, barcamps, Social media surgeries etc) and we’ve been working with some people in similar roles over the country to create a diary of (Apologies for the name) Geek events. One of the ideas I’ve been playing with there is a data submission API and some example html scrapers that will allow hyperlocal data providers to scrape local sites and contribute the data to a central portal. Are there ways to restrict event information to a specific sub-class of activity in Plings? We might be able to send you some events data. Now I think about it, those screen-scraping kits could be easily adapted to generic events and activities.
Anyways, you’ve gained a reader, cheers for the interesting blog!
best,
Ian.
@Ian Ibbotson – Many thanks for the comment Ian – really interesting.
Our feasibility found that the metadata underneath PKHD *is* adopted, and so utilisation of it would be a good next step. If, for example, the “positive activities” category was in wider use, it could also help us. Although, this then relies on a widespread agreement on what positive activities are!
Pete and colleagues are now in the process of writing and testing various tools to extract the data out – using some of the methods described in the study. We will try and get as much as we can out for inspection and discussion. I know some early data from Halton has come through this route – http://plings.net/la/00ET – not masses, but a start..
As for how long to post data for, I agree that is an interesting question. We found the same issue also applies when thinking about building an application/interface with the data. How long / how much data do you draw down/cache, and how much do you query live? As you suggest, some more focus and discussion on this thorny issue is needed
The scraping kits and API sound very interesting too (as does the central calendar of geek events – now that would be useful!). From the other end, we have often thought that if a youth club – for example – could format and prepare their data on a webpage in a fairly uniform way, then we could authenticate (eg: the URL associated with that Youth Club) and scrape it. I suppose the trick is to put this in language that the Youth club would follow, presuming that Linked Data and APIs are not top of their long agenda of things to keep on top on!
The report is excellent and I agree with the findings about data quality. I work with data that goes to the PKHD and I have had a lot of problems with regards getting data to parse correctly, for example some fields that are mandatory clearly do not apply to our services or events.
Hope it gets sorted soon because in theory there is lots of useful things that can be done with the PKHD data feed.