Optimising Data Accessibility

1
Optimising Data Accessibility via Reference
Metadata Management Principles
Q2006 - April 2006 (Russell Penlington OECD)
2
Search Engine Metadata Indexing
Query Search Engine
Get Results
Find Web Page with Data Link

• Linkage
• Relevancy
• Metatags
• Uniqueness
• Page Structure

Download Data
Understanding Search Engines

• Search engines crawl and index content by
  following hyperlinks found in online reference
  metadata.
• There exists a set of core concepts that serve
  as significant components in major search engine
  algorithms.
• Statistical organisations should provide well
  structured reference metadata reported online in
  a way for search engines to optimally index it.

3
Linkage Structuring Navigation
Linking Strategy
Multidimensional Data Structure
Reference Metadata Attachment Levels

• Common or shared Level
• Dataset Level
• Dimension Level
• Dimension Member Combinations (Coordinate
  Slices) Level
• Series Level
• Observation Level

Fundamentals of Linking

• Metadata content must be linked together so that
  search engines are able to index the entire
  content of the domain.
• Underlying metadata management systems should
  allow reference metadata to be attached to any
  level of the multidimensional data space.
• Online metadata reporting systems should use
  friendly URLs so that each content page can be
  represented independently and linked to.
• Unique metadata content pages should promote
  easy linking methods to facilitate discovery
  links from external domain sources.

4
Relevancy Establish Maintain
Online Content Example
Gross Domestic Product gt Transaction gt Final
Consumption
Content Inherited from higher level pages
Fundamentals of Relevancy

• Each metadata page is attributed a relevancy
  score for each of the keywords and phrases found
  in the page.
• Overall relevancy of a particular phrase for a
  page is derived from the combination relevancy
  scores on the linking pages and page content
  itself.
• Text chosen for linking reference metadata
  together plays an important role in carrying
  through relevancy within the hierarchical
  structure of the content.
• Maintaining relevance when linking reference
  metadata together also provides an intuitive
  navigation system for end users.

5
Metatags Create Dynamically
Metatags
Characteristics
Index Focus
Constructing Optimised Metatags

• Title The first 50 characters are typically
  indexed by the search engine which represents the
  most important content in terms of relevancy.
• Description The first 200-250 characters are
  indexed and should contain grammatically correct
  sentences with no more than 2 commas.
• Keywords The first 5 to 10 keywords are used by
  some minor search engines, however they have
  diminished in importance for large search
  engines.
• Content for each of the important Metatags can
  be constructed dynamically from the underlying
  reference metadata management system.

6
Page Structure Content Code
Uniqueness

• Content deemed to be a copy is inevitably
  discredited and given a low weighting in the
  final search results.
• It is important not to replicate the same
  reference metadata content at different levels of
  the reference metadata reporting structure.
• Metadata should be attached at the highest level
  possible of the datas multidimensional space
  with only the exceptions detailed at the lower
  levels.

Fundamentals of Page Content

• Metadata content found at the beginning of a
  page is given a higher weighting that content
  found at the end.
• Content display methods exist that do not
  involve modification to the actual content such
  as the insertion of introductory headings and
  text drawn from the metadata management system.
• Textual content should be grammatically correct
  and free of spelling errors.

Fundamentals of Coding Techniques

• Externalisation Client-side script (JavaScript
  CSS) should be referenced in separate files to
  avoid weighting dilution of the visible text
  content.
• No Frames Reference metadata displayed in an
  unlinked separate frame from the base access page
  will not be able to be indexed by a search
  engine.
• Valid HTML Search engines favour HTML pages
  that adhere to the HTML standards specified by
  the World Wide Web Consortium (W3C).

7
Summary Online Examples

• Linkage - Online metadata reporting systems
  should utilise a system of inheritance when
  linking together content allowing all content to
  be traversed from a single entry point by using
  friendly URLs.
• Relevance Unique pages within a metadata
  reporting system should retain and pass on
  relevance relative to neighbouring pages linking
  in and out of the particular unique page.
• Metatags Metatag content for each page in a
  metadata reporting system should be relevant to
  the pages content and be dynamically constructed
  from the underlying metadata management system.
• Uniqueness Each page within an online metadata
  reporting system should contain unique content
  and provide links to higher levels of the
  multidimensional data structures metadata to
  inherit parent level metadata.
• Page Structure HTML pages to report metadata
  online should be free from textual content
  errors, not be displayed in frames, and adhere to
  HTML version specifications.

OECD Main Economic Indicators - Sources and
Definitions Pilot project reporting single
dataset metadata online from MetaStore metadata
management system. http//stats.oecd.org/mei/ Goo
gle Keyword Search Examples Consumer Opinion
Surveys Definition, United Kingdom Labour
Compensation, United States Monetary Aggregates
8
Screenshot Backups
9
Google Search Consumer Opinion Surveys Definition
10
Google Search United Kingdom Labour Compensation
11
Google Search United States Monetary Aggregates
12
MEI Sources Definitions United States Monetary
Aggregates
http//stats.oecd.org/mei/default.asp?subject14c
ountryUSA
13
MEI Sources Definitions Monetary Aggregates
http//stats.oecd.org/mei/default.asp?subject14
14
MEI Sources Definitions Home Page
http//stats.oecd.org/mei/