2. About me
• Julie Boudro
• Consultant
• MCITP, MCTS, MCT
• julieb@cdh.com
• 248.554.3176
3. Why maximize search?
• Increase findability of information
• Reduce frustration in end users
• Increase adoption
• Create a new set of tools for rendering
content in SharePoint
• Way of the future with O365 and Delve
4. Mechanics of Search
• Crawler
– Discovers content
– Least privilege permissions
– Schedule versus continuous crawl (2013)
• Indexing
– Title & Description
– Content
– Meta Data
5. Mechanics of Search
• Ranking
– Relevancy ranking
– Popular items
– “Likes” or Ratings
– Best Bets
• Results
– Refiners
– Crawled and Managed properties
– Document preview
– Content Search web parts
9. Meta data
• Information automatically extracted from
content
– Title
– Description
– Author
– Date
– Contents
• Additional properties captured on content
– Content Types
– Site Columns
10. Content Rankings
• General purpose ranking models.
– General purpose ranking for most types of search
results.
• People search ranking models.
– Calculate how relevant search results are based
on social distance and expertise.
• Special purpose ranking models.
– Special purpose ranking model to calculate the
ranking score for recommendations such as
popularity, recommender based upon related
items
11. How do items rank?
Content
These are the words contained in the items. For items that are
text based, such as documents, this is typically most of the text.
For other types of items, such as videos, there is little or no
content.
Metadata
The metadata associated with items such as title, author, URL and
creation date. Metadata is automatically extracted from most
types of items.
Web graph data
This is information such as authority (from authoritative pages
settings) and anchor text (from the hyperlinks associated with the
item, and items linking to the item).
File type
Some file types can be considered more important for ranking
than others. For example, Word and PowerPoint results are
typically more important than Excel results.
Interaction
Information about the number of times a search result is clicked,
and which queries led to a result being clicked.
13. Advanced Search Principals
• Pay attention to site structure & navigation
• Understand the content being published
• Understand how people find information
• Content types and Meta data
– Global content types
– Leverage Managed Metadata & Term sets
• Search schema
– Crawled Properties
– Managed properties
– Query Rules
16. Plan Content Types
• Base set of required columns
– Department
– Document Type
• Use Managed Metadata Term sets
• Configured on all document libraries
• Create custom library templates
17. Search Schema
• Crawled Properties
– The contents and metadata of the items that you
crawl – Author, File Type
• Managed Properties
– Map crawled properties to managed properties
– Default properties
– Custom Properties
18. Query Rules
• Best bets & keywords (2010)
• Result blocks
• Changing the query
– Ranking changes based upon:
• Keyword match
• URL matching
• Content Types
• Tags
• File types
• Custom ranking models
32. Structuring for Effective Search
• Search Service Application Topology
Small Farm Search Optimized Farm
33. Detroit
1500 Woodward Ave
Suite 400
Detroit, MI 48226
(248) 546-1800
Thank You
Grand Rapids
15 Ionia Ave SW
Suite 270
Grand Rapids, MI 49503
(616) 776-1600
www.cdh.com
Hinweis der Redaktion
Relevancy is based upon learned behavior by the person that conducted the search. In O365 this is becoming critical across all office products with the release of Delve
Relevancy is based upon learned behavior by the person that conducted the search. In O365 this is becoming critical across all office products with the release of Delve
Relevancy is based upon learned behavior by the person that conducted the search. In O365 this is becoming critical across all office products with the release of Delve
Relevancy is based upon learned behavior by the person that conducted the search. In O365 this is becoming critical across all office products with the release of Delve
Relevancy is based upon learned behavior by the person that conducted the search. In O365 this is becoming critical across all office products with the release of Delve
Relevancy is based upon learned behavior by the person that conducted the search. In O365 this is becoming critical across all office products with the release of Delve
Best practice to move the query service onto a web front end and leave the indexing and crawling components on a Application Server. For Search optimized farms you add dedicated search service applications to your farm.