Learn how on-site search is similar to (and different from) Web searches and find out about the mechanics of on-site searching. Then you'll need to decide what search options you'll want to offer.

Learn How On-Site Search Compares with Web Searches
When you look for content on the Web using a search tool such as AltaVista or Excite, the search tool looks in its database of indexed Web pages from all over the Internet for the results of your query. When you do an on-site search, the site search engine only searches and indexes pages on that site.

Notice the similarities between Web and on-site searches. Both types of searches are conducted by using keywords, key phrases, and various qualifying terms such as a "+" or "AND" to make a request that only pages that contain both search terms are indexed. In site searches as well as in Web searches, you will see an alphabetical or "ranked" list of pages that contain the term(s) you've typed in to the search box.

Find out how on-site searching works. Typically, a site search engine has a "robot" or "crawler" function that examines the entire site at designated intervals, such as daily or weekly. When it notices a page that has been updated, added, or altered, it adds the content of the page into its database. Then, when you ask it to find pages that contain certain words or phrases, it conducts a search of its database. It will "rank" the relevant pages using such criteria as how many times the requested words appear in the document, how close to the top of the indexed Web page they appear, and how close together each instance of a keyword appears to another. This is called proximity searching. The theory behind this is the more times a site includes a word, and the higher up on a page it indexes it, the more important the word or term is to the page.

Understand Your On-Site Search Alternatives
Learn about the different types of searches, as well as how searchable data fields and indexing work. Choose a search engine that allows as many search options as possible. Make sure that, regardless of what options you offer, you provide clear instructions of how to use your search engine for best results. Find out what makes for a good Web site search by visiting Computerworld's evaluation of Web sites with exceptional search engines.

Understand full-text searches. A full-text search allows for more robust searching, allowing your visitors to find any type of content you've put on your site, no matter where it is. Yet full-text searches run the risk of being overly comprehensive. The inexperienced searcher may find that a full-text search produces no results because the information requested is too precise.

Understand keyword searches. The advantage of a keyword search is that it can identify topical areas and is a common form of easy searching. The disadvantage is this function will only be as accurate as the keywords you use to mark your documents. Keyword searches cannot direct the searcher to specific words or phrases that don't exactly match a keyword they've entered.

Understand searchable data fields. This can be a valuable function, especially if you're selling merchandise on your site and have equipped a commerce server to work with an online catalog of products. All documents must have the included fields to be searched this way. If you have hundreds of products with specific serial numbers and price ranges, you must take extra care to keep the search engine's database current, and build an interface with your inventory management system if at all possible. You must remember to add products as you make them available, however -- and to take them off your search engine index as they are discontinued, or your search results will not be accurate.

Understand indexing. If you're selling a small number of products, you may wish to manually index them alphabetically. If you do this logically and make this page easy to find, you may not need a search engine at all. Storing data in an index reduces the load on the server, allows a large pool of data to be searched by many people at the same time, and lets the search engine list pages in order by the relevance of their text to the search terms. A site search indexer reviews the files on your Web site and creates a searchable index of all the words it finds. These indexers locate your files through either local file indexing or remote spider indexing. Local file indexing resides on the same server and therefore requires space on that server. Remote, or robot, and spider indexers review all pages and then index all currently linked pages. Many large portal sites use robots and Web spiders to index sites. Site search engines, such as Ultraseek Server, can use robots to gather data remotely for indexing. Indexing can allow a site visitor to look for a term in the product specs that describes a product functionality they are seeking. A Web crawler may not be able to access data stored in a password-protected or encrypted data sections of your site. For more information about how indexing works, see the Infoseek article " Robots & Spiders & Crawlers: How Web and Intranet Search Engines Follow Links to Build Indexes."

Plan for sound and video file searches. Audio and video are useful tools for demonstrating your products in use, offering greetings to your customers, or showing your plant or facilities online. If you have audio or video on your site, you should equip your online search engine to index content with audio and video file extensions.

Plan for more complex searching capabilities. You may want to consider whether you'll want to offer multiples queries, date ranges, multiple data formats, and stemming. Generally speaking, the more information and documentation you have on your site, the more valuable these advanced searching methods will be. Multiple queries mean that searchers will be able to look for several terms at once. With data range capability, visitors will be able to look for content that was posted on, before, or after a certain date or within a given range of dates. Stemming means that if your site visitor types in only art of a word, the engine will be able find pages that contain the complete word. Because some people may not know how to spell the names of some of your products, this feature can help them find what they are looking for.

Copyright © 1995-2000 Pinnacle WebWorkz Inc. All rightsreserved. Do not duplicate or redistribute in any form.