By cataloging Web sites in advance, search engines can query more than 1.5 billion pages in less than a second. But even the best engines force you to spend time weeding out dead or irrelevant links. Snowtide Informatics Systems strives to eliminate this with Gemini UDS (Unified Datamining System) 1.1.6, a stand-alone application that polls numerous search engines concurrently to hone in on relevant Web pages. While it’s not perfect, the OS X-only Gemini largely delivers on its promise to make Web-based research more productive.
Most search engines treat all your search terms equally, but Gemini’s Query form lets you assign levels of importance to keywords. This form also sports check boxes that let you specify the type of content to look for — JPEG images or MP3 files, for example — and the search services to access.
Gemini begins by sending a query to the services you select. As results come in, the program downloads them and uses a proprietary algorithm to decide whether the pages meet your criteria. It uses a Java class file called an Evaluation Plugin to evaluate, rank, and format the results.
When it’s done evaluating results from the first pass, Gemini examines every link on the highest-ranked pages. This process, called Localized Webcrawling, lets the engine dig beneath the surface to find even more pages. Because this takes more time, Gemini’s advanced Query form lets you specify both the number of passes and the number of pages to “crawl” (or mine) on each pass.
Localized Webcrawling also lets you search individual Web sites. This feature comes in handy when you want to look for information on multipage FAQs and other sites that don’t provide search functions.
As soon as you begin a search, a second tabbed window opens, with progress bars that indicate Gemini’s status. Your list of results appears below the bars and shows each entry’s URL, along with an optional summary, modification date, and other information about the page. You can preview pages in plain text format without leaving Gemini, or you can open the page in your default browser by clicking on the link.
We performed searches on several subjects and used a wide variety of sources, ranging from general-purpose engines, such as Google, to specialized online databases, such as the National Library of Medicine’s PubMed database. Gemini was most beneficial when we weren’t sure where to begin. But the program wasn’t as helpful with specialized databases, which often provide content-specific search functions that Gemini lacks.
Because Gemini actually downloads pages to evaluate them, it’s much slower than conventional search engines, sometimes taking ten minutes or so to complete a search. Although you can tailor Gemini’s network usage, our cable-modem-equipped Power Mac G4/500 bogged down when we jacked up its Webcrawling parameters.
Macworld’s Buying Advice
Gemini UDS 1.1.6’s sophisticated capabilities aren’t for everyone. If you do only casual Web browsing, free services such as Google already do an excellent job. Likewise, people who use one specialized database probably won’t benefit from Gemini. But if your work demands wide-ranging Web-based research, Gemini merits strong consideration.