Benchmark Study of Desktop Search Tools

To view this page ensure that Adobe Flash Player version 9.0.124 or greater is installed.

Please login or register to make a comment!

Be2(-1&5k St8d< of Desktop Se&5(- Too0s April 20, 2005 Tom Noda Shawn Helwig www.uwebc.org/decisiontools There 9s More to Search than Google & Yahoo! An Evaluation of 12 Leading Desktop Search Tools www.uwebc.org© 2005. UW E-Business Consortium A new generation of desktop search tools is emerging that allows users to quickly Qnd relevant documents in computers across the enterprise the same way search engines help locate information on the Internet.

Companies expect that this technology will boost employee productivity and creativity and allow them to compete successfully in today 9s knowledge- driven economy. Desktop search technology itself is nothing new. In fact, it has been around for years.

However, some well know names (i.e. Google and Yahoo!) have recently entered the space giving this technology a well-deserved boost in visibility. In an effort to help understand the differences between the latest desktop search tools on the market, the UW E-Business Con- sortium recently conducted a benchmark study of 12 popular desktop search tools.

The benchmark criteria that were used for the evaluation included usability, versatility, accuracy, efQciency, security, and enterprise readiness. When all the results were reviewed, it was determined that most of the desktop search tools were still too immature for signiQcant business ... more. less.

use due primarily to a lack of mature security and overall manageability. However, considering the evolution of Instant Messaging from a pure consumer tool to a valuable enterprise application, desktop search may have similar potential. Based on our evaluation, the best overall desktop search tool is Copernic 1.5 Beta with Coveo. Yahoo! Desktop Search 1.1 Beta was rated the second best tool in our evalu - ation. See other notes. Likasoft Archivarius 3000 came in a surprisingly close third in our evaluation. This software is available commercially from Likasoft. Archivarius 9 index efQ - ciency is outstanding and was the clear winner in our tests. The user inter- face and navigation scheme is well designed and easy to use. ExEcuTivE Summary KEy FinDingS E;e(8t.9e S811&5< O9e5&00 r&t.2,s Be2(-1&5k c5.te5.& c5.te5.& r&t.2,s P5od8(t re9.e:s appe2d.; a appe2d.; B 2 3 3 4 4 11 12 TaBlE OF cOnTEnTS The benchmark evaluation testing was performed in March, 2005. This research was not funded or supported by any speciQc companies or institutions. The benchmark evaluations were conducted solely by the UW E-Business Consortium. BEnchmarK nOTES Usability Versatility Accuracy EfQciency Security Enterprise Readiness TOP 3 DESKTOP SEarch 1. cope52.( 1.5 Bet& :.t- co9eo 2. y&-oo! Desktop Se&5(- 1.1 Bet& 3. l.k&soft a5(-.9&5.8s 3000 Source: UW E-Business Consortium 2 De(.s.o2 Too0s | Desktop Search OvErall raTingS Our benchmark evaluation was performed across six main criteria. Each criterion was quantiQed and was given a rating, ranging from 1 (worst) to 5 (best). The rating is based on sub criteria, which align with the main criterion 9s objective. For sub criteria and their rating details, please refer to Appendix A - Comparison Table. 1. us&b.0.t< Good desktop search tools must be easy to use, have a lower learning curve, have professional aesthetics, and require fewer steps to reach desired output. 2. ve5s&t.0.t< Versatility describes how wide and deep the tool al- lows you to search. This includes factors such as sup- ported document types, web/e-mail integration, and multi-language support. 3. a((85&(< cCan you Qnd what you are looking for? d This criterion addresses accuracy of search results as well as other factors that help users Qnd the desired information. 4. Ef 3ciency This criterion assesses the tool 9s technical efQciency including memory usage, indexing time or indexed Qle sizes. The best tool should not jeopardize overall PC performance. 5. Se(85.t< Security and privacy are big concerns, especially in an enterprise environment. This criterion considers how well vendors have incorporated security mechanisms. 6. E2te5p5.se re&d.2ess While most tools are designed for the consumer/home PC environment, some are ready to be used in an enterprise. This criterion may be especially helpful for IT managers. BEnchmarK criTEria These are the overall benchmark evaluation ratings. Some tools are very good in speciQc areas such as usability, versatil - ity or search accuracy (explained later), but to be the best desktop search tool, a balance of all criteria is critical. cope52.( Desktop Se&5(- y&-oo! Desktop Se&5(- l.k&soft a5(-.9&5.8s 3000 mSn Too0b&5 S8.te goo,0e Desktop ask Jee9es En 3sh Professional iSyS Desktop dtSe&5(- Desktop d.skmETa P5o B0.2k; hotBot Desktop 1.5 Beta 1.1 Beta 3.14 2.0 Beta 1.0 1.0 Beta 6.1 6.0 6.5 1.0.1 3.0 Beta Desktop Search Tool Version Source: UW E-Business Consortium 4.11 3.66 3.62 3.45 3.26 3.16 3.10 3.05 3.02 2.63 2.63 2.34 Score (Min = 1.00, Max = 5.00) Better www.uwebc.org© 2005. UW E-Business Consortium 3 De(.s.o2 Too0s | Desktop Search criTEria raTingS Versatility 5.00 Accuracy 3.00 EfQciency 2.00 Security 3.50 Enterprise Readiness 3.00 Usability 1.50 The following charts summarize the best tools 9 ratings for each criterion. Blinkx and ISYS are versatile tools but struggle to deliver their powerful features in a user-friendly fashion. On the other hand, Ask Jeeves excels in usability, efQciency and security, but lacks versatility. Copernic is excellent in almost all criteria. 1. Usability Copernic Archivarius Google MSN Ask Jeeves 2. Versatility Copernic Yahoo! Blinkx ISYS 3. Accuracy Copernic MSN dtSearch 4. EfQciency Archivarius Copernic Ask Jeeves 5. Security Yahoo! Ask Jeeves Google 6. Enterprise Readiness Copernic ISYS Yahoo! 4.80 4.75 4.40 4.40 4.25 4.14 3.88 3.75 4.50 4.20 3.50 4.40 4.20 3.80 3.29 3.14 3.13 4.00 3.75 4.00 * Copernic with Coveo, and Yahoo! with X1 Source: UW E-Business Consortium PrODucT rEviEwS This section examines the details for each desktop search tool individually. The benchmark performance for each tool is expressed with a Spider Chart (see description), in order to convey the performance in each one of six criteria as well as the overall balance. Sp.de5 c-&5ts Spider Charts have been used to show how each criterion was scored, as well as the overall balance. For instance, the chart at left indicates that this tool is extremely good at Versatility but needs some improvement in Usability and Ef - Qciency. Achieving the maximum scores in all criteria and maintaining a good hexagon shape are ideal, but that is not required by all users. For instance, if Enterprise Readiness is not critical for a speciQc user, an unbalanced shape that lacks Enterprise Readiness features may still be a solid Qt. 4.00 www.uwebc.org© 2005. UW E-Business Consortium 4 De(.s.o2 Too0s | Desktop Search cope52.( Desktop Se&5(- 1.5 Bet& www.copernic.com Copernic is the most well-balanced desktop search tool among those evaluated. The tool is intuitive and easy to use. The new beta version supports FireFox for Web history search, and Thunderbird and Eudora for e-mail (as well as IE, Out- look and Outlook Express) cSearch as you type d and cdynamic indexing, d which detects new and modiQed Qles/e-mails on the Ry, are useful features. Filtering, sorting and grouping search results are well reQned. The application has a small y&-oo! Desktop Se&5(- 1.1 Bet& desktop.yahoo.com Versatility 4.14 Accuracy 4.50 EfQciency 4.20 Security 3.00 Enterprise Readiness 4.00 Usability 4.80 4.11 Source: UW E-Business Consortium Versatility 3.88 Accuracy 3.20 EfQciency 3.60 Security 3.29 Enterprise Readiness 4.00 Usability 4.00 3.66 Source: UW E-Business Consortium technology footprint and pro- vides detailed index controls. One potential improvement, however, is that it should let users choose a default web search engine. It only supports the call- theweb.com d web search within the application, which is not as popular as Google or Yahoo! For business use, Coveo, a spin-off company from Copernic, provides enterprise desktop search products, which enhance security, manage- ability and network capability. The client applications are identical. However, the enterprise version works with additional server products such as Microsoft SharePoint. Copernic Desktop Search 1.5 Beta Yahoo! Desktop Search is based on X1 Desktop Search, so usability will be familiar to existing X1 users. Yahoo! integrates X1 9s technology into its own portal services such as Yahoo! E-mail and Instant Messaging. It can index Yahoo! IM logs as well as Yahoo! Address Book. Versatility is excellent, as Yahoo! claims their tool supports more than 200 types of documents. It indexes Adobe Pho - toshop and Illustrator Qles in addition to many media Qles. Contents of zip Qles are examined and displayed in a tree structure. The tool 9s preview feature is well reQned, but its search results are somewhat clumsy because too many columns are displayed in a vertical view. As opposed to Copernic, there is no dynamic indexing or web history search. X1 offers an enterprise version of the desktop search tool as a server-based product. IT managers may want to check it out. Yahoo! Desktop Search 1.1 Beta www.uwebc.org© 2005. UW E-Business Consortium 5 De(.s.o2 Too0s | Desktop Search l.k&soft a5(-.9&5.8s 3000 3.14 www.likasoft.com Versatility 3.38 Accuracy 3.20 EfQciency 4.40 Security 3.00 Enterprise Readiness 3.00 Usability 4.75 3.62 Source: UW E-Business Consortium mSn Too0b&5 S8.te 2.0 Bet& toolbar.msn.com Versatility 3.63 Accuracy 4.20 EfQciency 3.60 Security 2.86 Enterprise Readiness 2.00 Usability 4.40 3.45 Source: UW E-Business Consortium In contrast to Copernic or Yahoo!, this is a commercial product, which costs from $25 to $45, depending on your status and purpose of use. Usability and efQciency are astonishing. It is a very simple GUI design, yet organizes a lot of features compactly. It demonstrated the fastest initial indexing time among all products, and memory usage was relatively low in idle time. It does not support any media Qle index - ing (image, audio or video) or web/web history searches. On the other hand, e-mail support is wide, ranging from Outlook and Outlook Express to Eudora, Thunderbird and Lotus Notes/Domino. One unique feature is that it offers remote search functionality. The ap- plication acts as a small Web server, allowing remote users to search the computer through a web browser. Of course, it has user/group account management capability built-in. Likasoft Archivarius 3000 3.14 MSN Toolbar Suite is similar to Google Desktop, and has almost the identical functionality and navigation scheme. Moreover, it includes a Popup Blocker and cAuto Form Fill d features which are already available with the Google Toolbar. MSN performed well in terms of search accuracy. Word accuracy was very good in our test. cShortcut keyword d is a unique feature, which lets users associate a keyword with speciQc Qles. With this association, users can type a keyword in Windows Explorer 9s address bar to fetch a Qle, instead of crawling multiple folders. One challenge is that it does not support PDF Qles by default. To index PDF contents, users must download and install an add-in tool called cIFilter. d MSN Toolbar Suite 2.0 Beta www.uwebc.org© 2005. UW E-Business Consortium 6 De(.s.o2 Too0s | Desktop Search goo,0e Desktop 1.0 desktop.google.com Google seamlessly integrates desktop search into its popular web search en- gine. The browser-based desktop search tool is easy to use and will be familiar to anyone who has used Google. A Roating bar is a unique feature and allows users to type keywords from anywhere on the screen. Google Desktop API is another promising feature, allowing software developers to develop add-ins to ask Jee9es Desktop Se&5(- 1.7 Bet& sp.ask.com/docs/desktop Versatility 3.50 Accuracy 3.20 EfQciency 3.33 Security 3.13 Enterprise Readiness 2.00 Usability 4.40 3.26 Source: UW E-Business Consortium Versatility 2.57 Accuracy 3.20 EfQciency 3.80 Security 3.14 Enterprise Readiness 2.00 Usability 4.25 3.16 Source: UW E-Business Consortium enhance the tool 9s func - tionality. OpenOfQce and ICQ index add-ins are already available. Unfortu - nately, Qltering and sorting functions are quite limited. It appears as if Google is so focused on its relevance algorithm that other sorting functions seem to be ignored. Google Desktop 1.0 Ask Jeeves 9 usability is remarkably simple and well reQned. It searches all types of documents simultaneously, and users can look through each type of results via tabbed panes. It has a nice preview pane, which even plays Windows Media Player. The application is very small and efQcient. Users are given index control when they can choose either fast or gradual indexing. However, Ask Jeeves 9 big - gest challenge is to improve versatility. Currently, the sup- ported document types are very lim - ited. Also, the preview pane does not render Microsoft Excel or PowerPoint. Web history search is not supported, either. If it would support more Qle types, Ask Jeeves has the potential to become one of the top desktop search tools. Ask Jeeves Desktop Search 1.7 Beta www.uwebc.org© 2005. UW E-Business Consortium 7 De(.s.o2 Too0s | Desktop Search En 3sh Professional 6.1 www.enQsh.com Versatility 3.38 Accuracy 2.40 EfQciency 3.40 Security 3.00 Enterprise Readiness 3.00 Usability 3.40 3.10 Source: UW E-Business Consortium EnQsh is a commercial software product. We tested the Professional version, which costs $199.95. EnQsh is a lot different from the other 11 tools in terms of usability. It gives users full customization of views. It provides calendar, contacts and weather views in addition to search/preview views. It almost simulates a EnQsh Professional 6.1 Microsoft Outlook environment. Users can create multiple index Qles and associ - ate them with differ- ent views. However, this customization creates a steep learning curve. It takes some time to get used to its operations. EnQsh has cRelevant Search d features, but its purpose appeared unclear. When we searched cOpen OfQce, d for example, the relevant search high - lighted csupport@amazon.com, d which made us wonder why. iSyS Desktop 6.0 www.isys-search.com Versatility 3.75 Accuracy 2.80 EfQciency 3.00 Security 3.00 Enterprise Readiness 4.00 Usability 1.75 3.05 Source: UW E-Business Consortium ISYS is a versatile tool. It supports multi-language indexing, FTP indexing, SQL indexing (requiring XML output), and supports many e-mail clients including Compuserve, Eudora and VIM. It is also good at enhancing search accuracy. It includes a spell checker, synonym rings, fuzzy logic search (which correct misspells in indexed documents) and intelligent date/number format (e.g. Qnd c1/1/05 d from cJan. 1, 2005 d). In contrast to those valuable features, however, us - ability is signiQcantly poor. The application creates multiple Windows menus and ISYS Desktop 6.0 confuses users. Custom query syntaxes create another steep learning curve for users who do not want to memorize them. Toolbar icons don 9t have text descrip - tions, yet their symbols are somewhat vague. When we searched MP3 Qles, we were stuck because it did not provide any links or enable us to play them inside the application. ISYS must improve its usability to be able to capitalize on its powerful versatility and accuracy features. www.uwebc.org© 2005. UW E-Business Consortium 8 De(.s.o2 Too0s | Desktop Search d.skmETa P5o 1.0.1 www.diskmeta.com Versatility 2.43 Accuracy 2.60 EfQciency 3.40 Security 2.86 Enterprise Readiness 2.00 Usability 2.50 2.63 Source: UW E-Business Consortium diskMETA is also a commercial product. We tested the most advanced version, cProfessional, d which costs $97.50. The application is one of the simplest, but comes with very limited functionality. diskMETA does not support any web history or e-mail search. Filtering and sorting are also limited, and there is no preview pane. However, it does have remarkable word accuracy features. It includes a dictionary feature that can identify a word, like ccriterion d from a keyword ccrite - diskMETA Pro 1.0.1 ria. d Surprisingly, most desktop search tools cannot do this. Iterating search results is another nice feature. Most desktop search tools we tested show all search results regardless of its amount (e.g. 2,000 matches), which can often overwhelm users. diskMETA 9s page iteration is intuitive and easy to use. dtSe&5(- Desktop 6.5 www.dtsearch.com Versatility 3.25 Accuracy 3.50 EfQciency 3.00 Security 2.88 Enterprise Readiness 3.00 Usability 2.50 3.02 Source: UW E-Business Consortium dtSearch Desktop is outstanding in terms of word accuracy features. It provides phonic and fuzzy search, boolean and wildcard keywords, multi language and encoding support, noise word list, which excludes common words such as cis d or ca, d and case/accent sensitive indexing. The most regrettable aspect, however, is that it does not deliver those features very well to the end user because of poor usability. There is no search keyword Qeld on the main window, and users have to go through the menu. Setting up the above features one by one is also a tedious task. The indexer treats most of the binary Qles as text and messes up the index Qle with none char - acters. If it enhanced the GUI and reQned the usability, it could become a very interesting desktop search tool. dtSearch Desktop 6.5 www.uwebc.org© 2005. UW E-Business Consortium 9 De(.s.o2 Too0s | Desktop Search B0.2k; 3.0 www.blinkx.com Versatility 3.75 Accuracy 2.60 EfQciency 1.80 Security 2.63 Enterprise Readiness 2.00 Usability 3.00 2.63 Source: UW E-Business Consortium hotBot Desktop Bet& www.hotbot.com/tools/desktop Versatility 3.00 Accuracy 2.20 EfQciency 2.00 Security 2.86 Enterprise Readiness 2.00 Usability 2.00 2.34 Source: UW E-Business Consortium Blinkx 3.0 has a Metal theme, which makes it look somewhat similar to the Ma - cintosh user interface. The tool 9s versatility is somewhat limited, but it does have some unique features. cBlinkx Visualizer d produces a tree view of search results, and users can see the tree growing. cSmart- folder d crawls the web to Qnd relevant informa - tion to the documents in the folder. Unfortu- nately, the application has major problems with efQciency. Indexing is painfully slow. In fact, in our test, the application couldn 9t complete the process. It runs four instances and consumes signiQcant memory. In con - sequence, search outputs are slow, and a window often Rickers. Hopefully, this will be improved in the future release. Blinkx 3.0 HotBot is a toolbar-based desktop search tool and displays output in the brows- er 9s left pane, where Favorite and History links are often displayed. The tool is very compact and has some unique features. It supports RSS indexing and allows users to associate keywords to custom web sites (e.g. ceb <keyword> d for eBay search). Unfortunately, we couldn 9t complete indexing on this tool for unknown reasons. The soft- ware seems to have been rushed for the beta release. HotBot offers a deskbar version, but usability is very inconsistent from the toolbar. ConQguration is text Qle based, and this is simply not user-friendly. It provides a lot of custom search syntax - es but has a steep learning curve. Search results are automatically saved as HTML Qles, and this may cause some security concerns. We expect signiQcant improvements in its Qnal release. HotBot Desktop Beta www.uwebc.org© 2005. UW E-Business Consortium 10 De(.s.o2 Too0s | Desktop Search Appendix A - CompArison TAble 1.75 3.75 2.80 3.00 3.00 4.00 3.05 $570.00 4.00 3.88 3.20 3.60 3.29 4.00 3.66 F5++ 4.75 3.38 3.20 4.40 3.00 2.00 3.62 $25.00 - 45.00 4.40 3.63 4.20 3.60 2.86 2.00 3.45 F5++ 2.00 3.00 2.20 2.00 2.86 2.00 2.34 F5++ 4.40 3.50 3.20 3.33 3.13 2.00 3.26 F5++ 3.40 3.38 2.40 3.40 3.00 3.00 3.10 $199.95 2.50 3.25 3.50 3.00 2.88 3.00 3.02 $199.00 2.50 2.43 2.60 3.40 2.86 2.00 2.63 $97.50 F5++ bW =Browser base, db =Deskbar, Fb =Floating bar, sA =Standalone Application, Tb =Toolbar C5/t+5/a / d+6kt34 s+a5ch T3316A6k J++v+6b1/2k:C34+52/c*/6kmeTA*ts+a5ch En\xcsh G33g1+H3tb3tisYsmsnl/ka63ftYah33! 1. U6a(/1/ty 1.1. Application Types 1.2. Features 1.3. Simplicity 1.4. Navigations 1.5. Aesthetic 1.6. Others in Usability 2. V+56at/1/ty 2.1. Supported PC Environment 2.2. Supported Files 2.3. Media Support 2.4. Application Support 2.5. Multi-language Support 2.6. Web Integration 2.7. E-mail Integration 2.8. Others in Versatility 3. Accu5acy 3.1. Word Accuracy 3.2. Additional Word Support 3.3. Index Accuracy 3.4. Output Format 3.5. Filter & Sort 3.6. Others in Accuracy 4. Ef\xcciency 4.1. Download/Installed File Size 4.2. Indexed File Size 4.3. Initial Index Time 4.4. Index Controls 4.5. Memory & CPU Usages 4.6. Others in Ef 0ciency 5. s+cu5/ty 5.1. HTTPS Cache Indexing 5.2. Personal Folder Search 5.3. Possible Intrusion 5.4. Protection Features 5.5. Privacy 5.6. Spyware & Adware 5.7. Product Update 5.8. Others in Security 6. e2t+545/6+ r+a*/2+66 6.1. Enterprise Products 4.25 2.57 3.20 3.80 3.14 2.00 3.16 F5++ 3.00 3.75 2.60 1.80 2.63 2.00 2.63 F5++ 4.80 4.14 4.50 4.20 3.00 4.00 4.11 ov+5a11 sc35+6 C36t6 SASA, TBSA, DBSASASABW, DB, FBBW, TB, DBSABW, TB, DBSASA, TB Source: UW E-Business Consortium ov+5a11 rat/2g6610110975128432 www.uwebc.org © 2005. UW E-Business Consortium 11 d+c/6/32 T3316 | Desktop Search aPPEnDix B We performed benchmark evaluations for all the desktop search tools on the same machine. To prevent any index con - Ricts, we installed/uninstalled one tool at a time. The details of the computer environment information are shown below: Test Computer : CPU : Memory : HDD : OS : Indexed Folder Size : File Types in Indexed Folders : DELL Optiplex GX240 Pentium 4 (1. 7GH) 512 MB 80 GB Windows XP Professional Documents & Folders = 672 MB, Outlook pst Qle = 4.13 MB (the same message con - tents for Outlook Express and Thunderbird), IE web cache = 8 MB, FireFox web cache = 19.2 MB Text (Unicode & ASCII), DLL, Java, Class, HTML, XML, RTF, MS OfQce (doc, xls, ppt, mdb), sql, OpenOfQce Qles, IM logs for Yahoo! & MSN, Adobe PDF, Photoshop, Illustra - tor, InDesign, archives (zip, tar, g-zip, rar), images (bmp, jpg, gif, tif, png, eps), video (asf, wmv, mov, avi, mpeg), audio (mp3, acc), and Asian text Qle and email. Test E29.5o21e2t Be2(-1&5k S8b c5.te5.& Des(5.pt.o2s The followings are the descriptions for the benchmark sub criteria. 1. us&b.0.t< 1.1 Application Types Is the tool standalone, browser based, toolbar or deskbar? (not rated) 1.2 Features How many useful features, preferences and options are available? 1.3 Simplicity How does the tool deal with the following tradeoffs (more features vs. simpler application design)? 1.4 Navigations How simple and easy is it to execute the search and results? How many steps does it take from inserting search keywords to reaching the target Qle? 1.5 Aesthetics How are the user interface components and functions reQned and organized? Does it look professional? How about commands, forms, icons and images? 1.6 Others in Usability Other remarkable usability features if any. This score is only applied if it raises the average. 2. ve5s&t.0.t< 2.1 Supported PC Environment Which operating systems does the tool support? Windows, Mac OS, Linux? 2.2 Supported Files Which Qle formats are supported? OfQce, PDF, IM Qles, Zip, RSS and folder names? 2.3 Supported Media Files Which image/audio/video Qles are supported? www.uwebc.org© 2005. UW E-Business Consortium 12 De(.s.o2 Too0s | Desktop Search 2.5 Multi-language Support Does the tool support multi language searches? Can it search Asian text? Does it support Unicode or other speciQc encoding types? 2.6 Web Search Integration How does the tool seamlessly integrate local machine search, web history, and web site search into one platform? 2.7 E-mail Integration How far does the tool search in the e-mail client? Does it search just e-mail messages, or does it also search attachments, address books, schedules and tasks as well? Does it require the e-mail client be running while indexing? 2.8 Others in Versatility Other remarkable versatility features if any. This score is only applied if it raises the average. 3. a((85&(< 3.1 Word Accuracy How exactly does the tool recognize keywords? If a user types capples, d does it also look for the word capple d? What about ccriterion/criteria d or cit/IT d? Does it support synonyms or a thesaurus? 3.2 Additional Word Support Does the tool have spell checker? What happens if users misspell cMassatusets d? How does the tool handle an ambiguous person 9s name? Does it support wildcard (* character)? What about double equations or boolean keywords? 3.3 Index Accuracy What will happen if users move or delete indexed Qles and then try to search them? What about new Qles or modiQed Qles? Does it support dynamic indexing, or does it require reindexing? What about received/sent e-mail? 3.4 Output Format How accurate and user-friendly is the output? Does it pinpoint exact word locations in Qles or just display the Qle name? How easy is it for users to Qnd documents from hundreds of outputs? 3.5 Filter & Sort Can users easily Qlter or sort search output? What kinds of Qltering/sorting options are available? How easy are they to use? 3.6 Others in Accuracy Other remarkable accuracy features if any. This score is only applied if it raises the average. 4. Ef 3ciency 4.1 Download/Indexed File Size How large are downloaded and installed Qle sizes? Are they small or large, considering its features and capabilities? 2.4 Supported Applications Check which applications are supported. This is related to the criteria above, but what about IE or FireFox in terms of web history searches? What about e-mail clients? Does it support Outlook, Express, Thunderbird, Lotus Notes or Eudora? What about IM? www.uwebc.org© 2005. UW E-Business Consortium 4.2 Indexed File Size How large are the indexed Qles? Are they small or large, considering its supported Qle types? 13 De(.s.o2 Too0s | Desktop Search 4.5 Memory & CPU Usages How much memory does the tool require during the idle and indexing time? How much CPU power does the tool require during the usage and indexing time? 4.6 Others in EfQciency Other remarkable efQciency features if any. This score is only applied if it raises the average. 5. Se(85.t< 5.1 HTTPS Cache Indexing Can users search SSL web histories? Do users have an option to prevent those pages from being indexed? 5.2 Personal Folder Search Can the tool allow users to search someone else 9s personal folders? Or does it restrict indexable folders, primarily for privacy/security reasons? 5.3 Possible Intrusion Is there any possible intrusion or security breach? 5.4 Protection Features Can users protect certain folders or documents from desktop search? How about password protected documents? Does the tool index them or ask users for a decision? 5.5 Privacy How does the vendor address privacy and security issues? Is it clearly stated on the web site or during installation? 5.6 Spyware & Adware Does the tool secretly install Spyware or Adware? Is there any unusual network activity occuring when the application is running? 5.7 Product Update Does the tool have auto update features so that users can apply updates as quick and easily as possible? Or does it require uninstall/install? How easy is it to uninstall and reinstall the new one (keep indexed Qles)? 5.8 Others in Security Other remarkable security features if any. This score is only applied if it raises the average. 6. E2te5p5.se re&d.2ess 6.1 Enterprise Products Does the vendor provide enterprise desktop search solutions? 4.4 Index Controls How can users control index performance and frequency? Can users control how much hardware resources the tool can use? How to schedule indexing? Automatic indexing during idle time? 4.3 Initial Index Time How long does the tool initially take to index Qles and e-mail? Considering its indexed Qle size and supported Qle types, is it fast or slow? www.uwebc.org© 2005. UW E-Business Consortium 14 De(.s.o2 Too0s | Desktop Search 15 De(.s.o2 Too0s | Desktop Search www.uwebc.org© 2005. UW E-Business Consortium abo8t uw E-B8s.2ess co2so5t.81 The UW E-Business Consortium (the industry membership base of the UW E-Business Institute) is Wisconsin 9s premier organization that helps companies gain a competitive ad - vantage through e-business. Our members - business executives and senior managers from the Midwest 9s leading companies - tap into world-class university resources and the collective experiences of this B2B and B2C group to address and share strategic e-busi - ness and information technology challenges, best practices and lessons learned. For more information, contact Assistant Director of Member Relations, Christina Paschen (608) 265-0645 or clpaschen@wisc.edu

less