[This article was published in the days of Windows 3.1, but is left here because some sites still link to it.  Just be aware you are reading a very old article.]

If you have ever worked with Lexis or Westlaw, then you are familiar with the concept of text retrieval. You enter what you hope are the distinguishing key words of the cases you are looking for, and you are presented with every document that contains those words. Text retrieval software such as ZyIndex and dtSearch brings that same capability to your own documents. And as hard disks get larger, and the files on them get older, such software is becoming more essential. Most word processing programs now possess their own text retrieval functions.

WordPerfect, for example, has a function called QuickFinder that indexes your directories, and then allows Boolean searches to find the documents you are seeking. But while these programs are a far cry from previous versions that would search only for file names, they have no where near the breadth of the dedicated text retrieval programs reviewed here.

With ZyIndex and dtSearch, you can index and search documents in any file format you are likely to possess. You can then view the whole document, or go directly to the "hits" to see the searched words in context. You can then print the document, or with a click, launch the program that created the document and make revisions. Aside from the common use of text retrieval programs by lawyers to review deposition transcripts, they are a great adjunct to word processing, especially where an office uses different file formats, such as Word and WordPerfect.

dtSearch 4.0

Both programs installed without incident, although ZyIndex did develop a strange glitch, which is discussed below. DtSearch is available in both a 32 bit version for Windows 95 and NT, as well as a 16 bit version for Windows 3.1. Also available is a program called FindPlus, which is an add-on for Norton Navigator, allowing users of that program to use the more powerful search features contained in dtSearch.

Unlike ZyIndex, dtSearch installs as a single program. With ZyIndex, you must open one program (ZyBuild) to create the index, and then open a second program (ZyFind) to search the index you just created. With dtSearch, all functions are performed from the same program. I found this to be far more efficient, especially since you may wish to add or update indexes on the fly.

For my test index, I selected my pleadings directory, which contains approximately 250 documents, occupying 2.5 MB of space. In most functions, I found dtSearch to be more intuitive in its design, and the indexing function was no exception. You simply type in a name for the index, and then select the directories you want included in that index. If you so elect, you can also indicate any file name extensions you don't want included, such as .exe and .com files. You can also choose to have the index automatically compressed after it is created.

In creating the index, both programs were very fast. But on this test directory, dtSearch was more than twice as fast. DtSearch took just 58 seconds to index the directory, as compared to 2 minutes, 27 seconds for ZyIndex.

One very nice feature of dtSearch is its ability to index "Zipped" files. Without expanding the files, dtSearch goes in and indexes all the words, just as with any other file. You can view and print the zipped files without ever expanding them. I don't like using Window's file compression on my notebook computer because it slow things down too much. I do, however, keep several of my directories zipped in order to save disk space. DtSearch's ability to work with those files is a major plus.

Once the index is created, searches can be performed. Since both programs use indexes, searches are instantaneous, even with 2.5 MB of data. Both programs allow the usual Boolean search terms, but dtSearch also allows easy selection of certain search features. By checking off boxes, you can invoke what dtSearch refers to as natural language searching, phonic searching, stemming, and fuzzy searching.

Natural language searching allows you to enter searches such as, "Get me the letter I wrote to William Johnson regarding the proposed settlement." DtSearch pulls out the key words in such a statement (ignoring "noise" words such as "the" and "to") and searches the index for those words. It then ranks the documents it finds using certain probability factors. For instance, in the above example, if hundreds of documents contained "settlement," but only a few contained "Johnson," then the latter documents would receive a higher relative rating.

Phonic searching finds sound-alike words, and stemming finds other grammatical forms. For example, a search for "apply" would also find "applies." Both programs also permit "fuzzy" searches. If you are concerned that the word you are looking for may be misspelled in some documents, you can structure your search to allow from one to ten variations on the word. With fuzziness set to one, for example, a search for "tomato" would also find "tomatoe" or even "toemato."

The documents located during the search are listed on a "search results" screen. I was surprised to see that ZyIndex lists only the file names. DtSearch lists the file names along with the creation date and the type of document (i.e., WordPerfect, Word, dBase). Obviously, if you are looking for a document you know was created last month, it can assist in finding that document if the date is provided.

Once you have located a document, you can look at it by simply double clicking on the file name. The programs opened all the documents I threw at them with no difficulty, even some old WordStar files. Neither program, however, maintained the formatting of the documents with any consistency. Pleading captions were often jumbled. This does not present much of a problem though, because both programs will launch the document's originating program with the click of a button. You are then free to view or print the file in its original form. The only exception to this is zipped files. Since dtSearch never expands them, zipped files cannot be opened by the parent application. However, you can select "copy file," and dtSearch will extract that single file from the zip file, making it available.

The documentation for dtSearch was very thin, but adequate given the straightforward nature of the program. The on-line help files were good, and included the phone numbers for technical support -- a seemingly obvious need that is missing from most applications.

ZyIndex 5.21

As soon as you open the box containing ZyIndex, it becomes apparent, from the bulk of the documentation alone, that this is a sophisticated program. Not surprisingly, it contains more features than does dtSearch. However, this added sophistication does not necessarily translate to a superior text retrieval program.

The program does not yet come in a Windows 95 version, and while it will run under that operating system, I did encounter one problem. I installed the program with no difficulties, and had no problems using it the first day. However, when I returned the next day and attempted to open the program, it crashed. Even after completely reinstalling it twice, the program refused to load, yielding only a fatal error message. Nor did installing the program on a different computer solve the problem. In one final attempt to get the program to run, I decided to duplicate all the conditions from the prior day, which included loading several other programs before running ZyIndex. That solved the problem on both computers. For whatever reason, at least on the two computers I tested, ZyIndex would only run under Win95 if other programs are also running. No doubt it has something to do with memory allocation. I was informed by ZyLab's technical support department that a patch is available to correct this problem.

With that problem solved, I set about creating an index. You start the process with ZyIndex by opening the program ZyBuild. You are presented with a dialog box which says "index directory." This is not a request for the directory you want to index, but rather is a request for the directory where you want the index to appear. Each index requires its own directory. Once you have created the directory for the index, you select the directory or directories you want included, and the type of files. ZyIndex will automatically exclude .com, .exe, and .sys files, and can be set to exclude any other file types you select. You can also select a file type directly, such as Word files. ZyIndex would then index only the Word files, and disregard all others. This latter method is not dependant on the file extension. Like dtSearch, ZyIndex can determine the file format regardless of the extension.

After an index is created, you must open a separate program, ZyFind, to search for documents. The search is performed in the same manner as dtSearch. You simply select the index you wish to search, and type in the key words you want to locate. ZyIndex does not have dtSearch's ability to perform natural language searches.

Once you have located a document, ZyIndex gives you far more ways to work with that document. Especially impressive are its hypertext capabilities. To link a document to another document, you just highlight some text in the first document, and then link it to some text in the second document. The text then appears in a colored block in the first document, and a click on the text takes you to the designated spot in the second document. Additionally, links can be made to graphic files, or even audio and video files. Clicking on such a link would automatically play the linked audio or video. With this capability, you could verbally annotate documents.

DtSearch also has hypertext capabilities, but they are far more convoluted. Rather than the simple highlighting method employed by ZyIndex, dtSearch requires coded text to be inserted in the document. Even then, the reference only opens another document -- it will not take you to a specified point in that document. These links can also call up image files. In the alternative, if you name the image files with a truncated version of the document's name, dtSearch will find them when you click on "image." For example, if the file is named johnson.wpd, the image files could be named john001.tif, john002.tif, etc.

ZyIndex also permits easy annotation of documents. Again, you simply highlight the text you want to annotate, and click on "note." A drop down box appears, and you type in your notes. You can select globally whether you want the notes to become part of the index. For easy recognition, note links are highlighted in a different color than hypertext links. If you are using ZyIndex on a network, you can decide whether the links you create are available to everyone on the system, or can be viewed only by you.

So, what if you need to gather together all of the documents you have located and have them available for court? In that case, you could buy the add on module ZyPublish, and download the documents to a CD-ROM. ZyPublish also adds a copy of the search engine to the CD-ROM, so you end up with all the documents and a means to search them.

Which Is Better?

To decide which is the better text retrieval program, you must have in mind exactly what it is you want to accomplish. As a means to locate documents and text within those documents, dtSearch is the clear winner. It is far more intuitive, and provides more information about the files located by your searches. DtSearch also makes it easier to get the documents into the system to begin with, since it has the ability to directly link with scanners (so long as you are using Xerox Text Bridge). It also makes better use of the Windows environment -- ZyIndex feels like a DOS program. As an added bonus, dtSearch is also faster than ZyIndex. So, dtSearch would be the right choice for a litigator who wants to be able to search deposition transcripts and discovery responses, or someone who wants to beef up the indexing capabilities of Microsoft Word.

If you require the ability to annotate the documents revealed by your searches, or link words in one document to a specific passage in another document, then you must forego the ease of dtSearch and go with the more powerful ZyIndex.
 

AARON P. MORRIS is a trial attorney with the law firm of Morris & Stone, LLP. His practice areas include business litigation and employment law. Mr. Morris has published numerous articles on business and computer related topics. He can be contacted at (714) 954-0700.