|
|
|||
|
|
|
|
|
|
||||
        
|
 |
Chapter 9: Search
Indexing Pages
The next step search engines take is attempting to determine what a page is about. This is usually called indexing. The method each search engine uses varies, but basically an indexer looks at various components of a page, including possibly its <title>, the contents of its <meta> tags, comment text, link titles, text in headings, and body text. From this information it will try to distill the meaning of the page. Each aspect of a page might have different relevance, and within the actual text, the position or frequency of different words will be taken into account as well. However, not all content within a page matters to a search engine. For example, stop words are words that a search engine ignores, normally because they are assumed to be so common as to carry little useful information. Examples of stop words might be "the," "a," "an," and so on. Most search engines have some stop words, but some engines like AltaVista claim to even index common stop words like "the."While the use of stop words may improve a search engine by limiting the size of the index file and focusing it on more content words, it may not match how users think about queries. Novice users may feel "The Best Butler Robot" is a better query than "Best Butler Robot." Sometimes the stop word may be important to the search. Consider searching for a song title like "Rock the Town." "The" is an integral part of the term and without it many other songs may come up. However, if the search were for "Rock the Casbah," it would be easier to throw out the noise word "the," given that "Rock" and "Casbah" rarely occur near each other. Deciding what stop words should be used can be very problematic given the broad topic domain of many Web sites. Once a page has been analyzed for the various keywords, it is ranked in relation to other pages with similar keywords and stored in a database. Ranking is the very secret part of search engine operation. How a particular search engine decides one page should be ranked higher than another is what search engine promotion specialists are always trying to figure out. A very popular way to rank pages today is based upon determined site landmarks. Home pages and major section pages may be given higher weight than other pages in a site. Pages that have numerous incoming links will also be given extremely high ranking. Providing a Search Mechanism
The final aspect of a search engine is the search page itself. A search page is the interface the user makes their query from, and it generally contains a primary query text box as well as other search fields for advanced users who may want to modify a query. The degree of complexity of the search page varies greatly in public search engines. Consider the difference in interfaces between basic and advanced search pages for various public search engines shown in Figure 9-1.
Users can enter queries as simple natural language questions—like, "Why is the sky blue?" (as encouraged by sites like www.ask.com)—or as complex Boolean expressions using advanced filters. Once queried, the search engine will retrieve the pages that meet the criteria and present them on a result page. Figure 9-2 shows a result page for the search engine Google (www.google.com).
From the result page, users can pick some results to explore, further refine the search with a new query, or just give up and try another method to locate what they were hunting for. The general function of search engines is illustrated in Figure 9-3.
Understanding what people expect Web-wide search engines to do is important, because users will bring their past experiences with searching to bear when using your local site search. Labeling, form layout, and result pages should somewhat mimic what users have come to expect from the public search engines. However, be careful not to directly imitate what public engines do. Public search sites aim primarily to get users to starting points for searching, while local search facilities on a site aim to provide a high degree of search accuracy. In fact, public search engines aren't always terribly accurate. They are often geared towards the needs of advertisers and the demands of dealing with the numerous tricks people employ to try to improve their site's ranking. Rule: Utilize past user experience with public search engines by using similar layout and labeling in local search facility design, but avoid imitating aspects of public search engines that deal with the uncontrollable nature of public Web sites. Next: Adding a Search Facility
|
 Overview | Chapters | Examples | Resources | Buy the Book! |
