Presented by Jennifer McNenly and Matthew Frederick,
Osler, Hoskin & Harcourt LLP
Although this case study describes a legal environment, the learning can be applied to any environment.
Life before the FindIt! Portal:
- thousands of structured and unstructured documents
- no centralized enterprise search
- no metadata
Step 1: Defining the Search Scopes – created four basic scopes: people, precedents and research, firm information, client matters
Step 2: Defining Search Requirements – case sensitivity, spelling, pluralization, word stemming, language
Step 3: Define Metadata Standard, including the controlled vocabulary
Step 4: Conduct Content Inventory – determine anything restricted or sensitive
Step 5: Metadata training for content owners – consistent naming conventions, synonyms, categories (also done for French content). Used Resource Description Framework (RDF) as a tagging framework. Developed a metadata tagging tool to add metadata to documents.
Step 6: Content Sources:
Web pages (ASP files) were indexed too. The website was built in frames, which presented challenges for indexing. Similar issues for Javascript links.
Internal blogs using Moveable Type and WordPress were also crawled and added to the index.
Structured data included InMagic Content Server, people directory (SQL).
Used crawl rules to include or keep out particular content and content types.
Best Bets were created based on frequently requested content, common tasks, stats on content usage, and existing “can’t find it” information.
Step 7: Search Interface and Results Display:
- Customized search display.
- Included link for user feedback and problems.
- Used a lot of out-of-the-box functionality, such as “did you mean” and “still can’t find it?”.
Step 8: Search Usability Testing – compare against actual user behaviour.
Having a project sponsor at a senior level in the firm is crucial.
Current Work: content migration into SharePoint; reviewing metadata fields.
Future Work: Faceted navigation, improved search.