20

FAST Search

WHAT’S IN THIS CHAPTER?

• What does FAST add to search?
• Tuning the crawler
• Advanced content processing
• Refining relevancy
• Search center enhancements

The previous two chapters have laid the groundwork for your understanding of enterprise search. This chapter introduces you to the advanced search capabilities provided by FAST Search Server for SharePoint. It is a basic overview of the most important components of the FAST product, but it is by no means a complete or deep instruction on this technology. Much of FAST is configured and administered through a complex set of XML files, Web Parts, PowerShell commands, and command line utilities. This gives you an overview of the topic.

FAST SEARCH CAPABILITIES

FAST Search Server 2010 for SharePoint adds new capabilities to enterprise search, as well as enhancing most of the capabilities that are included in the SharePoint Server 2010 enterprise search product. Everything from the crawler to the search experience is substantially improved with the FAST engine. FAST provides the architecture that allows for extreme scaling of the search engine to handle nearly infinite amounts of content while continuing to process and deliver content faster than any other search system.

Tapping into the advanced features of FAST requires more technical depth and a deeper understanding of your content and the search engine than were required to manage SharePoint Server 2010 search. The sheer breadth and depth of configurability that FAST delivers to the search engine is greater than can be effectively delivered through a web user interface. Yet, FAST Search Server begins by leveraging the user interface of the SharePoint Search product. This allows you to gradually move into the FAST world with a common base of understanding. In other words, you do not have to relearn everything about the search engine.

The Capabilities Themselves

Understanding what FAST Search Server 2010 provides above and beyond the SharePoint 2010 enterprise search product is an important starting point for planning and deploying your FAST installation. At the very least, a solid understanding of the capabilities and what they provide to the search experience will provide you with a way of organizing your search configuration efforts.

Advanced Crawl

FAST Search Server provides a level of crawl configuration flexibility that will make your head spin. But not to worry, the FAST search service applications provide content source configuration pages that look and feel identical to the content source pages found in the enterprise search service application. This enables you to begin using the FAST engine for crawling without worrying about the minutiae of crawl configuration. When you are ready to go beyond the Central Administration user interface, FAST allows configuration of every aspect of the crawler.

Contextual Search

FAST Search Server’s contextual search capabilities provide a high level of control over how search results are presented to the end users of your system based on a variety of contexts. The contexts that follow allow you to serve different sets of search results for the same search query based on the user and other factors.

• Relevancy tuning: You can affect the location of items in search results by modifying the weighting value of individual items or all items within a site. By adding to (promotion) or subtracting from (demotion) the weighted value of the item or items, you effectively push the item up or down in the result set.
• Synonyms: SharePoint Server search provides support for one-way synonyms. When the user includes a synonymous term for a keyword in a query, the items for that keyword are returned in the result set. But when a user includes the actual keyword in the query, the synonymous items are not returned. FAST Search adds the capability for two-way synonyms, which means that both result sets will be returned.
• Advanced Managed Metadata: The previous chapter discussed crawled and managed properties in depth because managed properties are a critical concept to search. FAST Search Server expands on the capabilities and importance of managing properties by allowing you to apply stemming, refinement parameters, and priorities with each managed property.
• Property extraction: Property extraction is the capability to map specific content information to a search property. Using the property extraction rules, you can force the inclusion or exclusion of the content when the specified property is used in a query.
• Rank profiles: Search relevance is calculated using a powerful calculation that cannot be affected by SharePoint Server. FAST Search Server, however, provides access to rank profiles that control how content is ranked in search results.

Search Experience

Not only does FAST Search Server add capabilities to the underlying search engine, it also enhances and adds to the search user experience. The installation of FAST includes a new search center template that you can use to deliver a FAST-specific search interface to your end users. The high-visibility items that are included in the enhanced search center revolve around the visual search and conversational search capabilities.

• Visual Best Bets: SharePoint Search allows you to configure a Best Bet for a specific keyword. When the end user executes a query that uses the specified keyword, the user is presented with a set of links that usually appear at the top of the results page and that point to content associated with that keyword. Visual Best Bets work in the same way but present the user with an image rather than a text link. These are great for product, logo, or other types of searches that are best served with an image.
• Document thumbnails: A picture is worth a thousand words, and when a user is sifting through hundreds of documents to find a specific one, having a thumbnail of the first page of the document can often save a substantial amount of browsing. Having the thumbnail presented as part of the search results eliminates the need to click the link and open the document in Office to see what its contents look like.
• Scrolling PowerPoint: Research conducted by Microsoft has shown that in the case of PowerPoint searches, users most often look for a specific slide or set of slides in presentations. As with documents, opening many PowerPoint presentations to find one or a few slides can be a very time-consuming and frustrating process. FAST offers PowerPoint presentations in an embedded viewer that allows the user to browse through all the slides in the deck without leaving the search results page.
• Sort results on managed properties: SharePoint Server Search allows you to sort results only by relevance or date modified. FAST enhances the sorting of search results by allowing you to define managed properties as sorting fields.
• Deep results refinement: Similar to sorting results, SharePoint Server provides a limited set of facets that the user can select to refine a result set. FAST opens up the taxonomy to the refinement panel allowing you to configure any managed property as a refinement option.
• Similar results and result collapsing: When a user executes a search in the FAST Server Search Center, each result in the result set can include a link to show similar results, which will re-query the search engine with a request that is mapped to the individual search result. It also includes a link that allows the end user to collapse duplicates in the display.

Architecture and Administration

FAST Search Server 2010 integrates closely with SharePoint Server 2010. Architecturally, it is composed of the same basic components described in earlier chapters: crawler, index, and query. In fact, most of the administrative pages and tools work in FAST exactly as they did in enterprise search. This is good news when your enterprise search solution grows to the point that you need to scale up to FAST. You’ll be able to leverage your knowledge of enterprise search to quickly put FAST into place, and you’ll be able to continue to run the enterprise search solution while you scale into the FAST solution. Figure 20-1 shows schematically how it all works.

Although FAST shares an architectural heritage with enterprise search, it expands on its basic components (Figure 20-1). The result is a hybrid of the two search systems that distinguishes between content searches and searches for people. In this hybrid system, People Search continues to be serviced by enterprise search, but all content searches are serviced by the FAST engine.

The hybrid approach to the search solution results in two search service applications for FAST: the FAST Content Search Service Application and the FAST Query Search Service Application.

The FAST Content Search Service Application handles all content crawling functions for the FAST server. FAST uses the same connectors that enterprise search uses to crawl and index content sources, and it includes additional connectors for high volume and Documentum searches. Like enterprise search the FAST Content SSA is able to crawl a variety of content sources.

The FAST Query Search Service Application handles all queries in the search system. It operates as a switchboard between content searches and People Search. If the query is for content, it passes the query to the FAST Search Server for processing. When the query is People Search, it passes the query to SharePoint Search.

PowerShell

If you haven’t already started to learn PowerShell, you will need to learn it to accomplish most meaningful tasks in FAST Search Server. PowerShell is a command line environment that gives Windows many of the shell and scripting capabilities that UNIX users have enjoyed for many years. With it, administrators are able to automate and accomplish more tasks in Windows with a single interface.

PowerShell is built on the .NET Framework, which allows the capabilities of the shell environment to be extended with .NET class-based commands and compiled assemblies. The compiled assemblies are added to PowerShell as snap-ins that provide the user with access to a set of commands defined within the assembly. These commands are known as cmdlets.

Both SharePoint Server and FAST Search Server install a PowerShell snap-in that provides search-specific cmdlets. These snap-ins are Microsoft.SharePoint.PowerShell and Microsoft.FASTSearch.PowerShell. As would be expected, the FAST-specific snap-in provides more cmdlets and a greater ability to configure search than does the SharePoint Server snap-in.

As stated earlier, a cmdlet is a class-based command that performs an action in PowerShell. Cmdlets are organized around a verb-noun command model in which the noun identifies the object and the verb identifies the action to be taken on the object. The most basic example of this is the verb Get and the noun Help which results in a cmdlet of Get-Help, which you can use to access the PowerShell documentation.

Due to the substantial number of cmdlets provided by both snap-ins, it is not practical to show them all here. I encourage you to take the time to read up on both the SharePoint Search cmdlets and the FAST cmdlets in their respective TechNet library entries. You can read about the SharePoint cmdlets at http://technet.microsoft.com/en-us/library/ee890087.aspx and the FAST cmdlets at http://technet.microsoft.com/en-us/library/ff393782.aspx.

CRAWLING

Conceptually, the FAST Search Server crawler is no different than the SharePoint Server crawler. It is the portion of the search system responsible for accessing content at a defined content source and indexing the contents of each source item into the search index. Crawl parameters, crawl rules, and the other configuration options that were covered in the previous chapter apply and are used in the same way.

The key decision point between the FAST crawler and the SharePoint crawler lies in three specific capabilities: large-scale crawling, finding links in JavaScript, and advanced tuning. The FAST crawler is optimized for crawling and indexing large-scale content sources without any degradation in performance. Additionally, the crawler has the ability to follow links located within JavaScript code on a page — something that SharePoint Server cannot accomplish. Finally, you are able to configure nearly every aspect of the crawler.

Often, you’ll scale up to FAST Search Server for one of these three reasons, not because you’ve maxed out the SharePoint Search Server. This tends to be true of FAST’s other aspects as well; the need for advanced capabilities rather than simply growth will cause you to scale up.

Tuning the Crawler

The FAST crawler can be configured with the FAST Content Search Service Application, using the same approaches and configuration pages as SharePoint Server. Within Central Administration, you create and edit content sources (Figure 20-2) that provide the crawler with basic instructions for accessing and indexing content. Similarly, you can create crawl rules, monitor crawl logs, and define file types in exactly the same way you learned in the previous chapter.

Leveraging the same configuration pages in Central Administration that you did for SharePoint Server makes the transition from SharePoint to FAST much easier and requires less of a learning curve. On the other hand, the configuration pages provided by Central Administration are substantially limited when compared to the vast array of configuration options available to the FAST crawler. Therefore, it is often in your best interest to abandon the Central Administration pages and get accustomed to working directly with the configuration files provided by FAST.

Advanced crawl configuration is accomplished in FAST with a set of XML-based configuration files. These files are located within the etc folder of the FAST root folder on the search server (this is usually C:FASTSEARCHetc). You can conceptually think of these files as you would content sources in the search service application, with the exception that the configuration files give you greater control over how the crawler behaves.

As with content source definitions, you should have one configuration file per content source. Because the format of the file is XML, it is valid to include all your crawl configuration in a single file. But once you begin scripting your configuration into the Crawler Admin Utility or PowerShell, you will wish you had separated them into individual configuration files.

FAST Search Server installs three sample configuration files in the etc folder. When configuring your own advanced crawler, it is best to begin by copying one of these files and modifying it to meet your needs. The sample configurations include a simple, an advanced, and an RSS configuration. All three files include copious comments that provide you with more than enough information to understand what each setting does for the crawler.

Crawler Administration Tool

FAST Search Server comes with a command line utility specifically designed to manage crawling in your farm. To use the Crawler Admin Utility you must be a member of the FASTSearchAdministrators group on the search server. The basic syntax for using the Crawler Admin Utility is crawleradmin -<switch>. Table 20-1 shows some of the more useful commands provided by the utility.

Table 20-1: Common Crawler Admin Utility Switches

SWITCH	DESCRIPTION
-f	Adds a content source to the collection
-G	Outputs the configuration of the content source collection
-d	Deletes a content source from the collection
-s	Suspends a crawl
-r	Resumes a crawl
--quarantine	Blocks crawling for specified number of seconds
--unquarantine	Unblocks the crawl
--refetch, --refetchuri, --refetchsite	Recrawls specified set of content
--status	Shows the status of the crawl
--sites	Displays a list of all sites being crawled
--statistics	Displays data about the crawler

CONTENT PROCESSING

At this point, it’s probably a good idea to review how content is indexed by the crawler and processed by the indexer, because this is an important part of search to master. When the crawler indexes the content of a content source, it uses a search connector to understand how to access, read, and extract the content of the file. As it reads the content of an item, it converts the content into crawled properties that it stores in the index. The crawled properties are organized into categories based on the rules defined within the protocol handler or iFilter defined by the search connector.

Crawled properties do not participate in the search engine. It’s important for you to understand this simple concept because many issues of missing content and bad search results begin with a misunderstanding of this basic principle. The crawler is not responsible for processing or presenting search results; it is responsible only for extracting and submitting content according to the configured rules. Therefore, there must be another process by which the results of the crawler’s work is processed, formatted, and made available to the query engine.

The content processor is the part of the search engine responsible for mapping the crawled properties generated by the crawler into a form usable by the query engine. The content processor is part of the indexer and it does its work between the crawler and the index. When the crawler extracts a crawled property, it submits the data to the indexer, which sends the data through the processing engine. The result of content processing is the entry in the index database that is later used by the query engine when processing search requests.

The glue between the crawled property and the index database entry is the managed property. Managed properties map one or more crawled properties into a single managed property term. It is this managed property term that the query engine uses to process search requests.

Managed Properties

Managed properties in FAST Search Server are configured in exactly the way they are in SharePoint Server. The Managed Properties page provides you with a listing of configured managed properties as well as raw and categorized listings of crawled properties that you can analyze to determine mapping schemes.

The first major difference between the managed properties in SharePoint and managed properties in FAST is the location of the configuration page. Simply put, the link is not where you would expect it to be — it’s not even close. Given that the link to managed properties in SharePoint Search is in the Query section of the navigation bar, you may think that the FAST Query Search Service Application (SSA) is the right place to find the link, and you are correct. The problem you’ll face is that there is no link for managed properties in the navigation bar. Instead of a link under the Query section, the link is in the FAST Search Administration page. Therefore, to get to the page, go to FAST Query SSA ⇒ FAST Search Administration ⇒ Managed Properties.

The second difference between FAST managed properties and SharePoint managed properties is that FAST contains some additional options. FAST search provides the ability to apply stemming (Figure 20-3), the function that analyzes the root form of a word (for example, run, running, ran), to any text-based properties you define.

Stemming is a capability that has been a part of the SharePoint Search engine since SharePoint 2007; it just wasn’t available at this low a level until FAST search. In SharePoint Search, stemming is applied to search results by the query engine. This means that the cost of stemming occurs at query time, which can affect performance depending on the search term that is being stemmed and the size of the result set.

As you recall, the mapping of crawled properties to managed properties is part of the content processing pipeline that affects actual entries in the index database. If a particular term is stemmed before it is entered into the index, the cost of stemming is paid up front and will not affect the performance of the query engine at all. To leverage this power and gain the full benefit of stemming at the managed property level, it is important to properly map crawled properties to the managed property in such a way as to maximize the benefit of stemming.

Within and across content sources, the crawler will often identify crawled properties that are simply variations of one another. For example, a crawler may go through a database and find a crawled property of client. Then it goes through a set of documents and defines crawled properties of customer, cust_name, and cust. In each case, the content refers to the same basic business concept: customer. From this content analysis, you would create a managed property called Customer and then map the four crawled properties to it (Figure 20-4).

Managed properties in FAST can be used to sort, query, and refine results in the search user experience. Later in this chapter, I cover how to configure the search Web Parts to allow users to sort, query, and refine using managed properties. Before you are able to configure the Web Parts, however, you must make each managed property available by enabling each taxonomy category (Figure 20-5). Enabling these categories is commonly referred to as taxonomy search refinement.

Finally, FAST allows you to prioritize a managed property by setting a mapping value between 0 and 7 (Figure 20-6). The priority that you configure for a managed property will directly affect the relevance of search results matching this managed property compared with results marching other managed properties. This allows you to make managed properties related to specific business concepts more relevant in search results than properties that may be related to content not important to the business. For example, you may have documents and web pages related to company team sports that result in crawled properties pertaining to team names, ranks, schedules, and so forth, which you’ll map to managed properties that you can push lower in search results.

Property Extraction

Mapping crawled properties to managed properties is an “after the fact” approach to managing content: you must crawl the content first to find the crawled properties, then analyze them, and finally map them to managed properties. Property extraction is a more proactive approach to indexing content. The property extraction process finds and extracts specific information from the content and maps it to one of a set of predefined extractors.

Out of the box, FAST includes extractors for companies, locations, and person names. Using special XML files that are located in the FAST root folder on the server’s filesystem, you can add more extractors to meet your business needs (consult the MSDN documentation for instructions).

When you define a term in the extractor, you either set it as an inclusion or exclusion to the extractor. Inclusions define terms that the extractor should recognize and map to the extractor (Figure 20-7). Exclusions define terms that the extractor should ignore.

Adding a term to the extractor (Figure 20-8) is a very straightforward process. You click the numbered link next to the extractor to either include or exclude a term based on your business need. This brings up a list of all inclusions or exclusions defined for the extractor. You can then use the Add Item button to enter a term into the list. Once added, that term will be mapped to the extractor on all subsequent crawls. This is why it’s a good idea to reset your index after making substantial changes to the configuration of your search engine.

RELEVANCY TUNING

I think that the most important capability offered by FAST Search Server is the ability to tune relevancy at the site-collection level. This is where you are able to materially affect the quality of search results in your organization in ways that are not available in any of the other SharePoint Search offerings. Before we dig into relevancy tuning in FAST, it is important to review what relevancy is and how it relates to search queries.

The main function of the query engine is to strike a balance between recall and precision. Recall is a factor of quantity whereas precision is a factor of quality. To better understand this differentiation, consider a search request that returns a result set. When the query engine makes the request for content it makes that request to the entire index of content. The amount of content that the index returns to the query is the level of recall. In other words, returning all items in the index is 100 percent recall; everything else is a percentage of the whole.

Returning all items in the index for every search request would not yield quality results and it would not enhance performance. As a result, the query engine will return only items that are relevant to the search request. If the returned result set contains only items that are relevant to the search request, you have 100 percent precision. If some items in the result set are not relevant, you have some factor of diminished precision.

The result set that the query engine returns to the search request is an intersection between recall and precision. The configurations for crawling and content processing are those settings that affect the balance between the two. These settings will impact what and how content is delivered to the query. Relevancy tuning, on the other hand, affects the ordering of content in the returned result set.

FAST Search Server uses a complex calculation to apply a ranking score to each item in the search result set. The ranking score determines the item’s location in the search set; higher ranking items appear toward the top of the set and lower ranking toward the bottom. FAST Search Server’s relevancy tuning capabilities allow you to modify aspects of an item’s ranking score in order to move it up or down in the result set.

FAST and SharePoint use six factors to calculate ranking scores (but only FAST Search allows you to tweak the score):

• Quality: This is a static score that is based on the content itself and reflects the item’s level of importance. This score is calculated at the time of indexing.
• Authority: This is a score that reflects the frequency of matches between the search query and the link text of a document. For example, if a document is stored at a URL of <site>/Benefits%20Documents/Dental%20Benefit.docx and the user types in the keyword Benefit, the authority of the item will be boosted by the two exact matches of the string Benefit that occur in the URL. Authority is also affected by search results that have been clicked on in previous queries. If a person makes a query for Benefit and clicks the link to the Dental Benefit document, and then another person executes the same query for Benefit, the Dental Benefit document will be given additional ranking points because of the previous click-through.
• Freshness: This is a calculation that is based on the difference between the last modified date of the document and the date of the search query. The smaller the difference between the two, the higher the ranking value that is added to the document.
• Proximity: When a query consists of more than one word, the proximity score reflects the frequency with which these words in the query occur together in the document. For example, if a person searches for Dental Plan, a document containing the phrase Dental Plan will get a higher score than a document that contains the words Dental and Plan in different places. The number of times that the phrase Dental Plan occurs in the document will also add ranking to the document.
• Context: This score is based on where query terms are found in the document. Managed properties allow you to set Full Text Index Mapping; this is the configuration that impacts the context quality of the document. Therefore, only managed properties that are part of the full-text index are considered in this calculation.
• Managed property boosts: Once again, I can’t emphasize the importance of managed properties enough. FAST allows you to use any managed property to increase or decrease the ranking score of an item. In SharePoint Search many people do not like the fact that PowerPoint documents are rated higher than Word documents. Using a managed property in FAST, you can boost the value of Word documents to be above that of PowerPoint documents.

To discuss all the techniques and tools to tune each of the relevancy categories would fill many pages. It’s a relatively advanced topic and requires deep knowledge of your content and the PowerShell environment. This is something that you should look into after you have mastered all other aspects of search. In the remainder of this section I cover the site settings that are available to you and that also affect the ranking score of documents in the result set. These are modifiers that are applied at query time when the query engine is returning a result set.

The site collection-based ranking techniques are all located in the Site Collection Administration section of Site Settings. It is important to remember that some items in Site Collection Administration are specific to SharePoint Search and some items are specific to FAST Search (Figure 20-9). The techniques discussed in this section rely on the FAST settings.

Keywords

Before digging into the specific ranking techniques for site collections, we must review keywords because many of the refinement settings are tied to keywords. A keyword (Figure 20-10) is a word or phrase that matches the word or phrase entered by a user in a search request. By capturing a match for specific words or phrases, you can configure search to respond to common business terms, acronyms, or important business concepts. The most common two responses that you will configure for keywords are synonyms and Best Bets (Best Bets are covered later in this section).

When the user includes a synonymous term for a keyword in a query, the items for that keyword are returned in the result set. But when a user includes the actual keyword in the query, the synonymous items are not returned. For example, suppose you create a keyword for the phrase Annual Enrollment and create a one-way synonym for Benefits. What this means is that if a user types the word Benefits in a search query, he will get results that include matches for Benefits and matches for Annual Enrollment. If the user executes a search for Annual Enrollment, he will get matches for Annual Enrollment only. FAST allows you to set up two-way synonyms so that a search for Annual Enrollment will also return matches for Benefits.

After creating a keyword, you will be able to configure synonyms by editing the keyword from the Keywords page. The drop-down item menu (Figure 20-11) on the items page will also give you access to configure the other FAST relevancy refinements that are linked to the keyword.

Document Promotions and Demotions

FAST Search Server allows you to promote and demote documents (Figure 20-12) that match a keyword. What a promotion does to the specified document is add or subtract 1,000 rating points to the document within the result set returned by the keyword. For example, suppose that you are analyzing the result set for Annual Enrollment documents and you see that the document called “Summary of Annual Enrollment Changes” is at the bottom of the first page, but you want it at the top. By setting a document promotion for this document, it will likely push the document to the top of the results page. Similarly, you may find that last year’s document is also on the listing; by applying a demotion you can push it down in the set.

Many of the relevancy refinements in FAST can have start and end dates assigned to them. If the date values are left blank, the refinement will begin immediately and run indefinitely. Yet at times having a start and end may be a desired factor of the refinement. For example, perhaps you want to boost Annual Enrollment specific documents only during the annual enrollment period, after which you want documents to be listed according to their natural ranking.

Site Promotions and Demotions

Site promotions and demotions (Figure 20-13) work exactly the same as document promotions and demotions, by adding or subtracting 1,000 document ranking points. Unlike document promotions/demotions, the site promotions/demotions are not tied to a keyword, and they apply to all documents contained within the site.

Each promotion/demotion that you create can be mapped to more than one site. This is helpful when certain documents are located in a variety of content sources. For example, suppose you are a major food manufacturer and you want to promote all documents related to a new line of cookies. You may have documents about the research and development of the cookies, marketing brochures and materials about the cookies, logos and product images, and other documents. All these may be stored in various team sites, file shares, or other sources. By creating a single site promotion for the cookie line and then mapping each of the site URLs to this promotion, you can apply the promotion to all documents for this cookie rather than creating a keyword and adding a document promotion for each and every document that you have.

User Context

FAST Search Server allows you to define promotions, demotions, and Best Bets according to user context. The user context is a specific group that is configured (Figure 20-14) in site settings to which the promotion/demotion or Best Bet will apply.

User context is a power refinement-tuning mechanism because it means that different users will be able to see different result sets for the same search term based on the user’s context. For example, a human resources employee may need to see a different set of search results for the term Annual Enrollment than another employee of the company. In this case, creating a user context for the human resources employee and applying that context to promotions, demotions, or Best Bets will provide the human resources employee with a better search experience.

FAST SEARCH EXPERIENCE

FAST Search Server adds a number of enhancements to the search experience by expanding on the existing search Web Parts and by adding new Web Parts to the system. The enhancements that FAST brings to the search experience are ones that allow you to provide richer and deeper search results to the end user. FAST adds a new search center template to SharePoint, appropriately called the FAST Search Center. This template contains a preconfigured search experience that mirrors the one that comes with SharePoint Server but includes the new FAST-specific Web Parts.

Search Core Results

The FAST Search Core Results Web Part (Figure 20-15) adds the ability to display similar results, document previews for PowerPoint, and document thumbnails for Word, as well as specific settings pertaining to these FAST-specific capabilities. Similar Results is a link that shows with each item in the search results page. When the user clicks the link, FAST will re-query the index using the search terms associated with the Similar Results link. This capability is disabled by default but is easily enabled by checking the box in the Display Properties of the Search Core Results Web Part.

Document previews and thumbnails are enabled by default in FAST search, but you may find that you want to tweak things like the size of the preview or the number of previews and thumbnails that the result set displays. Most users do not progress in search results beyond the first few pages of the result set. Considering this, it is inefficient to return document previews and thumbnails for documents that will appear further down in the result set. Therefore, you can limit the number of previews and help the performance of the results display.

Search Action Links

The Search Action Links Web Part displays the sorting drop-down box and links to RSS and Search from Windows. By default it is located at the top of the results page above the results display. FAST Search allows you to hide or display the links individually by selecting or deselecting the checkboxes at the top of the Search Action Links properties pane (Figure 20-16).

More importantly, the Search Action Links Web Part is where you enable the sorting of search results based on managed properties. As you recall from the earlier discussion of managed properties, FAST includes a selection to make the managed property available for sorting. The managed properties that have sorting enabled will appear in this Web Part, but are not enabled by default. To allow users to sort on the managed property, you select the enabled checkbox in the Web Part properties pane and then select a sorting direction for the property.

Visual Best Bets

Best Bets are a functionality available in SharePoint Search. When you configure a keyword, you can associate a set of links that you want to appear whenever a user executes a search using the specified keyword. The SharePoint Search Best Bets are simply a set of links that appear within the Search Best Bets Web Part (Figure 20-17), but sometimes a Best Bet is better served with an image. For example, if you set up a keyword for Company Logo, you may want to have a Best Bet that points to the logo usage guidelines document, but you may also want to have the logo itself show up at the top of the results. FAST adds the Visual Best Bet Web Part to your search center so that you can accomplish this requirement.

Visual Best Bets, like Best Bets, are contained within a keyword. In the keyword drop-down menu (shown earlier in Figure 20-10) you were given the option to add Best Bets and Visual Best Bets directly to the keyword definition. You may also select the keyword details option to see a summary view (Figure 20-18) of the keyword, which includes a listing of all Best Bets and Visual Best Bets assigned to the keyword. The summary screen also provides you with links to add, edit, or remove Best Bets and Visual Best Bets.

To add a Visual Best Bet to a keyword, you must provide a unique title and a complete URL that points to the image you want to display in the Visual Best Bet (Figure 20-19). As with promotions and demotions, you can specify a user context for the Visual Best Bet and set the start and end dates.

The one key difference between Best Bets and Visual Best Bets (besides the visual part) that you need to be aware of is that the Visual Best Bet is not a link, as opposed to the Best Bet, which is a link to specific content. This means that in most cases, you will want to add a matching Best Bet that links to the content relevant to the keyword.

Refinement Panel

In the search world, refining results based on content metadata is a concept referred to as faceted search. In SharePoint, facets are displayed in the Refinement Panel Web Part and are referred to as refiners. Out of the box, SharePoint Search provides refiners for document properties that are extracted during content processing and stored in the index. The default set of refiners that SharePoint handles are Result Type, Site, Author, Modified Date, and Companies.

FAST Search Server allows you to add managed metadata and tags as refiners in the Refinement panel (Figure 20-20); this is known as taxonomy refinement. The Filter Category Definition property of the Web Part contains an XML document that defines the facets available to the Refinement Web Part. This chapter does not provide an in-depth discussion of the XML contained in this property, but when you get to the point where you are ready to begin adding properties to the panel, it will be much easier to copy the text into an IDE that has color coding, such as Visual Studio, for easier editing of the configuration.

SUMMARY

Unfortunately, a short chapter such as this one cannot sufficiently explain all the different aspects and configurations available in the FAST Search Server product. It is a very capable search platform that gives you a great deal of control over the search engine and search experience. The downside is that it requires a deep and technical knowledge to work with all the different options.

This chapter wraps up the section on enterprise search. At this point, you should have a solid understanding of the components of a search engine and how those components are exposed in the SharePoint 2010 stack of search products. You should also be able to configure a SharePoint Search engine that will provide high quality and high performance search results to your end users.