1. Knowledge Management Solution Line
Focus on Enterprise Search approach
Do not disclose without prior consent from the Project Management Practice Contact: pmp@altran.be Version 1.5
2. What will the enterprise search imply?
Search is for action or production
• Poor or no result > no action, no production
• Bad result > Bad action, bad production
Search is THE access to information : simple & quick
• User centric tool : accessible, straightforward
• Reliable tool : permanent, durable
Search is THE window to the information repository
• All relevant sources must be accessible
• Incredible developer, marker of the quality of the information management
2
3. The risks
Why the users would stop using your search ?
Bad results (too much, poor, not relevant, obsolete)
= bad index
= poor index
= bad representation of the results
Interface not user-friendly
Response time
Not exhaustive
The lost users are extremely hard to get back !
3
4. The key success factors
Strong management of quality & quantity of information
Interface matching with expectations of all users
From Google like to Expert search
A search that is contextual to the company, the location and the business
A search that is versatile (synonymous, cleansing, navigation…)
Response time accepted and ensured
Functional upgradeability (favourites, related links, quotes …) & technical scalability
A quality roadmap (reporting & monitoring leading to actions)
4
5. The implementation of a search strategy
Analysis
What is the status of my information repository? Quality Audit, Sizing, volumes, Dispersion, Number of
sources, Variation, evolution, Not “indexable” as is Content
What is the expectation, expressed or not, of my users ?
Architecture
What are the key functions ? What is the evolution path ? What is the search engine? What will be the
interface tool(s)? What is the sizing ?
Phased implementation
A: Scalability
B: Function Evolution
Quality process
Usage Monitoring > Tuning of the search service + actions on information mgt
5
6. The different levels of the search service
Search criteria
• Google like : full text
• Synonyms, business dictionaries
• Source filtering
• Metadata (classification, filters, facet's) > depend on the produced information
• Taxonomy, Ontology, …
Integration/accessibility
Local data Enterprise data Personal mails Shared mails
WEB X X
Client X--- X X--- X
Full client (incl. local index) X++ X X++ X
Results
• list ordered by relevancy
• list with quick view
• contextual filters
• faceted navigation
These criteria will imply the complexity of the indexation customization
6
7. Conclusion
Do not consider search as an isolated technical tool
Search MUST be user centric
Prefer a modular architecture (interface independent from engine)
Use of standards
Must be considered as a service (quality control, support, SLA …)
Don’t rush = reduce the risk of rejection at first use (difficult to recover)
Trap to avoid : the meta-search
Loose relevancy, loose context differentiation > loose quality of results
Loose index improvement > loose upgradeability
7