We demonstrated in ‘The AAAPT of it all‘ how the simple classification of ‘entities’ (Actors, Actions, and Artifacts) extracted from source material made available during the discovery stage of Action Intelligence – when properly located in time and space – could accelerate the effective analysis of a given issue or situation.
(There’ll be plenty more examples of the benefits of applying simple AAAPT classification in other parts of this site, and we encourage you to explore them.)
The AAAPT classification scheme, however, is more than just a simple hierarchical taxonomy. It’s an ontology with a core (yet extensible) set of defined relationships that allow one to ‘link’ entities together.
At the beginning of the AAAPT ontology’s development, the relationships we defined weren’t just ‘made-up’ out of thin air but emerged from language analysis of thousands of reports, documents and articles.
At first, the defined relationships were ‘coarse’. They reflected both their repeated usage of natural language and their relative importance in helping answer the sorts of questions our clients were asking us. As time passed, the relationships became more refined as more sophisticated questions were asked of the data.
As always, an example is worth a thousand blog entries, so let’s grab another sample extract from one of our media intelligence feeds for a different domain than the previous Ebola example and ‘add-in’ this extra ability to define relationships.
Beirut: “The Agriculture Ministry Wednesday [Dec 10] morning set the wheels in motion for a reforestation project to plant 40 million trees, sponsored by Prime Minister Tammam Salam.
The U.N. Food and Agriculture Organization representative in Lebanon, Maurice Saade, announced during a news conference at the Grand Serail that the FAO would offer the Agriculture Ministry technical help and funding to the tune of almost $300,000 starting early 2015.
The aid will be dedicated to forming a unit to coordinate the national reforestation program….”
– See more at: http://www.dailystar.com.lb/News/Lebanon-News/2014/Dec-11/280648-ministry-launches-reforestation-project-for-40-million-trees.ashx#sthash.1Xg8knYQ.dpuf
First, let’s parse the AAAPT entities from the extract – Actors: The Agriculture Ministry, The Food and Agriculture Organization (FAO), Tammam Salam, Maurice Saade; Actions: reforestation project (an activity), new conference (an event), national reforestation program (same as existing action); Artifacts (40 million trees, $300,000); Place: Lebanon, Beirut, Grand Serail; and Time: 20141210 (standard date format), 2015.
And now, we define the ‘coarse’ relationships (links) between entities:
- Relationship: Employment, Actor: Tammam Salam; Role: Prime Minister
- Relationship: Sponsorship, Actor: Tammam Salam; Action: National Reforestation Project
- Relationship: Technical Assistance, Actor: FAO (Role: Donor), Actor (Role Recipient): Agriculture Ministry
- Relationship: Financial, Actor: FAO (Role: Donor), Actor (Role Recipient): Agriculture Ministry
- etc.
Let’s imagine that a regional assessment of Middle East environmental projects was being undertaken on behalf of the World Bank Group as part of a 5-year policy planning cycle. Traditional approaches might include an in-depth desk analysis of various media sources and documentation over months to accumulate and extract the data required to inform the analysis.
However, if the full AAAPT process, as indicated above, had been applied to news feeds and reports, and the data stored in an accessible manner, then any number of questions could be asked of the data (and save months of desk research activity). Example queries of the data might include:
- What are the top three recipients of environmental funding for each country over the last 5 years?
- Who are the largest donors for environment projects in the Middle East over the last 5 years?
- Are government or non-governmental organisations the largest recipients of donor funding in the Middle East for environmental projects?
- etc.
Over the last 5 years, we have done exactly this… built the capacity of many country teams to apply the AAAPT classification scheme to thousands of documents, reports and articles to support them in their analysis. Some of the Case Studies elsewhere on this site demonstrate the outputs of such efforts.
In another post, we’ll explore the second step of the Action Intelligence classification stage… the application of interpretive ‘lenses’ to the entities that have been classified according to the AAAPT knowledge representation ontology.