In-Depth

In Search of Desktop Search

Finding a data needle in the haystack that is your hard disk requires more than just a magnifying glass and a huge dose of patience.

• By Peter Varhol
• 05/01/2008

In enterprise search, generally speaking, you have a choice between the Google search appliance and Microsoft Search Server. These products can carry out searches among enterprise servers and storage for documents that may not have been indexed into an intranet or electronic library.

Of course, Windows has always had a search facility integrated into Windows Explorer, but it can only help you locate files by their filenames or text within word processor files. It's also abysmally slow.

Your users have to do more than just search for basic files on their desktops. They have to find e-mails, e-mail attachments, selected text or combinations of text. They have to search non-word processing files like PowerPoint slides and Acrobat documents. They have to search quickly, accurately and without a lot of false positives.

REDMOND RATING Windows Desktop Search Google Desktop Search dtSearch GlobalBrain Personal Edition Documentation 20% 8.0 7.0 8.0 9.0 Installation 20% 7.0 8.0 9.0 9.0 Ease of Use 20% 9.0 9.0 8.0 9.0 Feature Set 20% 7.0 6.0 9.0 8.0 Administration 20% 8.0 8.0 7.0 9.0 Overall Rating: 7.8 7.6 8.2 8.6 Key: 1: Virtually inoperable or nonexistent  5: Average, performs adequately   10: Exceptional

There are many desktop search engines available. Here's a look at two freely available desktop search tools along with two commercial solutions, which include license fees.

Are the freely available desktop search tools sufficient for most uses, or should you invest in a commercial product? The commercial search tools clearly have more functionality than the free ones. They enable more complex searches and can search through more different types of files. You're also paying for that capability, though, so there's a balance between search needs and the ability to fund the commercial product licenses.

If search is just a convenience, it probably doesn't pay to look beyond one of the free solutions. However, if finding the right documents or having a reliable paper trail is a way to make or save money, the commercial products are absolutely worthwhile. Whichever way you go, you must match the product with the business need.

Microsoft Windows Desktop Search 3.01
Either of the Windows XP or Windows Vista versions of Windows Desktop Search (WDS) is a more elegant and functional search facility than the built-in Windows Explorer search. For XP, WDS mounts on the Taskbar and in the System Tray. Just type a search word or phrase in the Taskbar slot, or click on the icon in the System Tray to open the full interface.

[Click on image for larger view.]

Figure 1. Windows Desktop Search: Microsoft's free Windows Desktop Search integrates into the Windows Task Bar and provides fast searching, but it isn't always as accurate as it could be.

WDS is a free addition to XP (Vista has its own integrated search facility). You can download it here.

Based on Windows Explorer, the interface is easy to understand and use. A bar across the top of the page lets you choose the file type to search, or simply search all files. It searches Word documents, e-mail, favorite links, photos and pictures (file names only), appointments and other Outlook-related items, and Windows Messenger instant messages. You can type in a word or phrase (for the exact phrase, use quotation marks) or multiple words separated by commas or spaces, and then hit the magnifying glass search icon.

Windows Desktop Search typically returns its results within a second or two, even when searching a 75GB hard drive. The software uses idle CPU cycles to continually index files and information on those files. In fact, it returned search results faster than it could post previews of the documents.

The previews are definitely a valuable feature. The entire right-side pane displays the contents of many of the files it finds that contain your search terms. This helps you confirm that the document will be useful to you. Oddly, some of my file icons were marked as "Preview Not Available," but in many cases the preview displayed anyway.

One complaint about the WDS results is that there are simply too many. This would be understandable if I had used common search terms, but I searched for several very precise and technical words and phrases. WDS still returned dozens of responses. Those responses were rated between one and three stars, but many of the one-star responses had no relevance to my search terms.

Still, WDS is easy to use. It's fast, free and works across common file types. Don't expect it to work with a non-Microsoft file format, though (except for Acrobat .PDF). You might want to look elsewhere if your enterprise generates a lot of non-Microsoft files. Currently, it doesn't work well with Office 2007 formats, although I expect that will change. If you're looking for a well-integrated way of quickly searching your

Windows system and getting returns of reasonable quality, WDS fits the bill. And did I mention that it's free?

Google Desktop Search
You can download Google Desktop Search for free from Google.com as part of the more comprehensive Google Desktop, a PC dashboard that gives you customized news feeds, Web clips, a scratch pad, stock and local weather feeds, maps and a to-do list. After installation, Google Desktop indexes the e-mail, files and Web history stored on your hard drive. It waits for idle time on your computer to do this in the background, so the initial indexing can take several hours if you're actively using your system.

[Click on image for larger view.]

Figure 2. Google Desktop Search: Google's free Desktop Search serves up results in a plain Web page similar to that found on the Internet.

The desktop search interface is the familiar Google.com search page, customized for PC search. You use it exactly as you would during a regular Google search, by typing a search query into a text box. The software then returns a list of documents matching the search criteria in a Web page.

Google does a couple of nice things here. First, it shows all the hits on your computer as you type in the search term. Second, by showing the results as a uniform resource identifier in the Web page, it made them nicely RESTful (Representational State Transfer). This way, you can use the locations to manipulate data on the desktop with simple programs. You can also easily click on the link to see a Web representation of those documents, rather than launch the actual applications.

It will also go out and do a separate search on the Web for your search terms. This effectively draws little distinction between local and remote documents. Just as Google does on the Web, it has a lot of flexibility to search and offers an engine that returns highly relevant results. Google returned different and seemingly better organized results than Microsoft.

However, it wasn't all good. Google has an excellent search engine for Web pages and documents. On the desktop, though, it fails to use features that can enhance its ability to work better within those confines. This includes things like document preview and an organized table of information on the documents. The desktop dashboard is nice, but it's such a different way of working that the average user will require a learning curve before they can take full advantage.

Google takes information from other applications and links on your computer and evaluates them. Then it sends you feeds based on that information. An enterprise IT staff may find that a bit disconcerting, although it's an advanced feature you can customize to limit information upload. Most will find the combination of search and feeds to be both enticing and useful.

dtSearch 7.5
dtSearch is a comprehensive search engine that gives you search and text retrieval across a large number of file formats. It has the reputation of being able to scan large volumes of files in a short period of time and returning accurate results. It's indeed a very good search product overall, with high performance and the ability to fine-tune searches and search strategy.

[Click on image for larger view.]

Figure 3. dtSearch: dtSearch creates a very large index, but it returns results noticeably faster than the other alternatives.

dtSearch was easy to install, but building the initial index was a challenge. It took me several attempts to build the index. I didn't change anything to get it to build successfully, so the issue was impossible to diagnose. However, once I built the index -- which took about five hours -- everything went perfectly.

There are a lot of options in creating the index, in terms of the data indexed and directories examined. After adjusting some of the parameters during a subsequent indexing, I was able to reduce the time required to be just more than an hour. You can schedule index updates and merge multiple indices.

The index also has the potential of taking up a fairly large amount of disk space (about 2GB, in my case). However, the resulting search is lightning fast. I used a wide variety of search terms, both simple and complex, and each returned a result immediately after requesting the search. No other product here was as fast. It also seemed accurate. An informal scan didn't reveal any obvious false positives.

dtSearch has numerous options, possibly too many for the average user. You would want to either use the default configuration or offer a standard configuration if you were going to deploy to large numbers of desktops (it does have the ability to create an options package for large deployments). You can also create search reports and search histories, so you can identify the most common search items and organize files by subject.

dtSearch has excellent search capabilities, and its performance was unmatched by any other product here. The time it took to index a modestly sized hard disk and the complexity behind its large number of configuration options is a concern. To a power user, though, these are no problem. The excellent results could well justify these costs. Many users could index just the document portions of their disk, which would save both time and space.

Brainware GlobalBrain Personal Edition 2.0
The GlobalBrain Personal Edition is intended for desktop users operating in a Windows environment to manage, search and classify information on directories to which they have access. It recognizes a large number of different file formats, from the common Microsoft file types to some very specialized formats. In all, it can sort through and identify search terms in more than 400 different file formats.

[Click on image for larger view.]

Figure 4. GlobalBrain: Brainware's GlobalBrain provided a good balance between index size and speed, and was highly accurate in its hits.

The lowest common denominator GlobalBrain uses is the Rich Text Format (RTF) document. It translates all documents it can read into RTF, and does its searching from there, though that matching is done internally. It returns results based on the original file type.

GlobalBrain searches with an interesting algorithm. While indexing the disk in preparation for search, it breaks words and phrases into successive three-letter groups. It then breaks up the search term in the same manner, and performs comparisons of its three-letter groups to those in the index. It then rates the possible hits based on matches to those three-letter groups.

During the indexing process, you can choose to index the entire disk, just the My Documents directory, just Outlook, or both Outlook and Outlook Express. Once the indexing is completed, you can simply type in search terms. GlobalBrain lists the results in order of relevance by default (you can easily select other ordering criteria). It also shows the file type, size, location and date modified. You also get a preview, regardless of the file type.

The Classify feature of GlobalBrain Personal Edition is an interesting and useful addition to the concept and practice of search. This lets you automatically organize documents related to your favorite topics. Classification means grouping unstructured items by common properties. You end up with an entirely new and different view of how data is organized on your computer. For some, this feature alone may be worth the price.

It also lets you scan in documents, perform optical character recognition (OCR) and turn them into Word documents for searching. This makes it easy to take non-electronic documents and make them searchable. While OCR isn't usually completely accurate, using the product's search algorithm makes it easy to find results even in inaccurate documents. Brainware also has an enterprise search version that gives you server-based search across a department or entire organization.

GlobalBrain has a few quirks. Don't install it and expect to see a search function. It's instead called associative access. Also, this is not a simple, type-in-your-search-term-and-get-results tool. You'll need to train your users in how to use it, or at least give them a pass through with the documentation. If you need a comprehensive and exacting search engine that supports a wide variety of different documents, though,

GlobalBrain is one of the few products that will meet your needs.