Characteristics of the "DatedResource" Topic Types

* DC, Dec 18 - fixed broken link in comments, now at bottom of page,

This is not a PSI-tree, but it is a sketch for one that would try to better define types 1.2.1-7
  1. Defined Subtypes
    1. Their required characteristics (in a 5-D sparse array)
      1. Modality = 1 of enumerated document types (text, graph, sheet,..)
      2. Recurring = 1 of a few enumerated policies on regular  updates
      3. Attribution = 1 of enumerated policies
      4. Tagged = 1 of enumerated policies
      5. Focus = 1 or more of the "entity" types?
        • this is really the "title subject" of the document
        • if none exists, document type may be just  "navigation"
        • this may exist only at instance levels, not at type levels
        • tags will attach other subjects referenced by instance text
    2. Their names (each is a named region in the 5-D array)
      1. Article = text: several sub-types exist, all attributed and tagged
      2. Tearsheet = text; qrtly; attributed, tagged; company or country,
      3. Worth Watching = text,?, not attributed, tagged, ???
      4. Morning Comment = text,daily, not attributed, not tagged, ???
      5. (new product) "Company Spreadsheet" = sheet,qrtly,?,tagged, a company
      6. BondScore Fair Value = graph, qrtly, not attributed, tagged, sector
      7. BondScore CRE = graph, wkly, ?,?, company
  2. Required of all Instances, no matter what the sub-type
    1. a named sub-type via instance of (see 1.2)
    2. URI = a subjectAddress: at your site; multiple as needed
    3. Time Tag = the date of the analysis; ISO 8601
  3. Required for some Instances, as indicated
    1. Authors = first binary association iff author known - between
      1. the authored (an analysis; the page) = 1: the web page instance
      2. the author (the analyzer(s)) = N: who wrote it or else just "staff"
    2. Discusses = second binary association iff focus known between
      1. the discussion (analysis; the page) = 1: the web page instance
      2. the discussed (the analyzed; the focus) = N: an instance of entity type
    3. References, iff tagging required by 2.1 and 1.1.4.  It adds more like 2.2
      1. referencer = 1: the instance
      2. referenced = N: the entities mentioned, not just the central one(s)


* MT, Dec 16 - phone feedback of on stuff below, now incorporated

1.2.1-7 - Fleshed out attributes of each basic type - these are working metrics only
1.2.8 - drop "Top Articles Read" and "Weekend Reader" as avigation types not yet modeled
1.1.5 - Focus is tricky; may need to allow a set of entity instances, not one entity type
3.1-3 - role association may be confusing; better to split into two binary assocations

* DC, Dec 16 - For earlier discussion see typesofreport.htm

The "DatedResource" name is only a placeholder, but it does describe what I'm thinking are the two main characteristics of the "content" that analysts or your UI people want to create or improve :
Some things that seem to fit this model may be just "navigation", like the Entity Homepages

I do not yet REQUIRE attribution and tagging for these types, but Miles' comment below does seems to suggest doing this.  I will instead wonder aloud on these related issues:
Tearsheets, for example seem never to appear in any search, so I wonder if they are actually tagged.

I also wonder if the test I suggest below on focus see 1.1.5 and 3.1.3 seems reasonable?  If there is no clear focus (company, or some sector (industry or subindustry) or region (country, etc), I would argue that it is really just navigation.  

* MS, Dec 12

Looking over the list of types I enumerated [as 1.2.1-7], yes I think we do want attribution, date of publication and tagging associations to core types for every one. Yes, this would apply even to images like Fair Value curves.