Search Power

Search Power

San Diego County Office of Education has accurate, quick search engine.

Problem: After redesigning its Web page, the San Diego County Office of Education received complaints that the search engine was inaccurate and extremely slow.

Solution: The SDCOE installed Google Mini as its search engine and hasn't received a complaint in over a year.

The Internet has become such a commonplace aspect of everyday life that a poorly functioning Web site is almost worse than no Web site. With all the new Web services available, it is sometimes helpful to see what larger entities are using to determine what will work at the local level. The SDCOE's goal is to provide technology, curriculum, and other support to the smaller school districts in its area so they can provide a quality education to students. The county education office has a wide constituent base composed of internal staff, district faculty and staff, and parents.

Daniel Yee, Web services supervisor, explained that they tried two other products before settling on Google Mini. The first was the Microsoft indexing product that came with their Web hosting system. It generally slowed the system further when it was indexing. Users also complained that it was inaccurate. Since they were on a limited budget, they tried Zoom Search, which was good, but it elicited the same complaints as the Microsoft product.

They were attracted to the Mini because of its features that promised both speed and accuracy. "The accuracy was really what we needed," Yee says, and it also didn't strain their budget. In addition to all hardware and software, the Mini also includes one year of support and hardware replacement coverage. There are four versions of the Mini, with price points that increase with search capabilities: 50,000 documents for $1,995; 100,000 documents for $2,995; 200,000 documents for $5,995; 300,000 documents for $8,995.

Self-taught

"There is a bit of a learning curve with it," Yee cautions. It takes between 10,000 and 20,000 searches for the engine to develop accuracy. But with the SDCOE's site, which receives 4,000 hits a day, it took only a month to reach that level. The Mini uses the same algorithm as Google.com, and now administrators can customize search results by assigning points to Web page titles, body text and meta tags. As people perform searches, the engine learns which results people are looking for when they enter certain key words. Yee says they did not tell users about the new search engine because they didn't want people to "test" it, which could have resulted in false results, delaying their accuracy development.

Another way the IT department is able to enhance speed and accuracy is by creating subcollections for various departments' Web pages. Yee explains that if a user searches for the County's Regional Occupation Program, or ROP, the search results would find all the documents associated with it. He was able to isolate the ROP pages under a separate URL so users can go directly to that Web address, which limits their search to ROP documents. Alternatively, searches performed from the county's home page will search the entire Web site and all subcollections. "It's been a very valuable tool," Yee says.

Fine Tuning

In addition to being able to create subcollections, Yee can also exclude Web pages from the search parameters. So if a page needs maintenance or a project has been discontinued, the general public won't see it. When changes are made they do not take effect until the next time the Mini crawls the site. The unit can be programmed to crawl on a set schedule, or it can be initiated at any time. Yee likes the fact that the Mini is an independent box on the server rack. And it only takes about half an hour for it to crawl 30,000 documents. It will capture most file formats, including Web pages, Word documents, Excel spreadsheets, PowerPoint presentations, and PDFs. Formats can also be excluded.

"I think it's a great tool," Yee concludes. He says outside users can find things faster, and internal staff is better able to assist constituents.

The Google Mini would also benefit a large school district. He suggests that to better utilize its capabilities, before it is launched, administrators need to "sit down and map out how it will be used" by internal and external users. He also advises administrators to build subcollections carefully so search results don't end up in the wrong place.

Ann McClure is associate editor.


Advertisement