How does contextualisation and bundling work?


#1

How does this concept of contextualisation and bundling actually work technically? I can’t visualise it!


#2

Hi @augustinas,

Thanks for your question. You can find technical details and examples in this document. For a bit more general descriptions see Hat Briefing Paper 4.

So once our personal data is collected in our HATs, we can contextualise it in a way that makes sense to us using the 5 categories (Data Tables):

  • People - yourself and the people you interact with
  • Things - what you use or interact with, especially those IoT devices; Things represent both virtual and physical objects
  • Events - what happens to you or what you do in time
  • Locations - where is the stuff around you happening
  • Organisations - what organisations or institutions activities relate to

A connected home, for example, where a sensor measures when a door is opened or closed, could be modelled as a Thing, as well as someone’s Google Calendar with several Events linked to them. The 5 Tables interconnect through a series of cross-reference Tables, where an Event can link to several People, Organisations, Locations, and Things and a Person can link to several Events, Organisations, Locations, and Things, and so on. This is flexible system that allows users to fully model their existing data sets. For example, someone could model a meeting they had in Meeting Space 2 as a connection between an Event with a start time and an end time, the individuals involved as People, and Meeting Space 2 itself as a Location.

The HAT enables us to transform our previously collected data into 5 Entities: Person (Who), Thing (What), Event (When), Location (Where), and Organisation (Where/with Whom), so that we could describe how we live – When and Where an Event took place, What things and Data are involved with Who (and potentially Why) by putting them (Who/When/Where/What) into a Bundle.

Bundles of contextual data are sets of entities (people, things, events, locations and organisations) and their properties that a HAT owner has chosen to combine together for potentially sharing with others through the Direct Data Debit system. The HAT’s Direct Data Debit (D3) system works like a direct debit in a bank: we can decide exactly what Data to share, for how long, to whom such data may be exchanged, and what return may be offered in the exchange: any Data that belongs to the HAT User can be bundled and shared from any Collection or directly from Source Data (e.g. Facebook), either at the level of individual Properties or entire Collections of Data.

Hope this helps!


#3

Hi @snaiste, all,

a couple of questions. I understand that in the world of contextualization, everyone cuts the cake a bit differently; my question would be on what would HATDex recommend?

  1. The Person context

In your examples/diagram, there is a Person called Me. Say if I add another person, what would be the value of the label “Me”? => Do I use the name of the person “John Doe”, or use a nickname like “Funny Guy”

  1. The Organization context

This is basically a “how to use the Organization context”.

a. The context “A Person is a MemberOf Organization” is straightforward
b. Say, I import a Facebook Post, I would have "Person::Me authors a Thing::SocialMediaPost"
Question: Should I link the SocialMediaPost to an Organization called Facobook.
Or, put in a field called Source, with Value=Facebook within the SocialMediaPost

  1. Are there plans to come up with a Recommendation for Standard Practices and Lexicons for Contextualization?

You say Person A is the SisterOf Person B
I would say A “http://purl.org/vocab/relationship/siblingOf” B

Cheers
Terry


#4

Hi @Terry_Lee,

Thank you for your message.

The Person context
In your examples/diagram, there is a Person called Me. Say if I add another person, what would be the value of the label “Me”? => Do I use the name of the person “John Doe”, or use a nickname like “Funny Guy”

It doesn’t really matter - it’s up to the individual to name their entities.

The Organization context
This is basically a “how to use the Organization context”.
a. The context “A Person is a MemberOf Organization” is straightforward
b. Say, I import a Facebook Post, I would have "Person::Me authors a Thing::SocialMediaPost"
Question: Should I link the SocialMediaPost to an Organization called Facobook.
Or, put in a field called Source, with Value=Facebook within the SocialMediaPost

I would suggest to put it in a field called Source, with Value = Fecebook, since Facebook is a source rather than organisation in this case. I would suggest using “Organisation” for things like, for example, the company where you work.

Are there plans to come up with a Recommendation for Standard Practices and Lexicons for Contextualization?
You say Person A is the SisterOf Person B
I would say A “http://purl.org/vocab/relationship/siblingOf” B

A suggested set of Types and Units of Measurement are being compiled for the next release, however best practices and lexicons will need to come from the user base based on their practices.

Hope this helps :slight_smile:


#5

Hi, in this case of inputting into the contextualisation table, if we are collecting data from Facebook, is the correct way to create a record in Things for Facebook? If so, then how does this relate or differs to your reply to Terry? Thanks


#6

Hi @chua_lai_chwang,

You can create any records you like. Contextualisation is intended to be performed by the user, however user interface can guide users in doing so and Rumpel currently in development does it that way. Contextualisation APIs are only available to the owner and hence accessible through HAT management UI, such as Rumpel. They are not available for direct use by data sources.

Best,
Aiste


#7

This has been answered as of now already … thx