Developing Your Search Strategy

Think of a PILOTS search as an exercise in pattern matching. You tell the computer what pattern of letters, words, or phrases you are looking for, and it attempts to match that pattern with those it finds in the database. You can tell the computer where in the database to look for a pattern, and you can tell it to search for a combination of patterns. The success of your search depends on

• the clarity with which you form the pattern you try to match
• the accuracy with which you type it into the computer
• the skill and completeness of the database producer.

Two out of the three are up to you.

Basic Approaches

There are two basic approaches to searching the PILOTS database:

• controlled vocabulary
• natural language.

In controlled vocabulary searching, you are instructing the computer to match terms from a prescribed list against fields within the bibliographical citations in the database in which those terms are used to describe the papers indexed.

In natural language searching (sometimes called "free text" searching), you are telling the computer to match words or phrases that you think might occur in the bibliographical records, regardless of whether they appear on a prescribed list of terms.

Each offers advantages and disadvantages. Many users will find that a combination of both types of searching will produce the best results.

Controlled Vocabulary

Controlled vocabulary searching takes advantage of the work done by National Center staff to standardize the terminology used by the thousands of authors and editors who produce the traumatic stress literature. This standardization is especially important in an interdisciplinary field, as there is no assurance that the terms used by psychiatrists will necessarily match those used by criminologists, or art therapists, or social workers. Even within a discipline, changes in terminology occur over time, or across geographic or ideological boundaries.

We use two vehicles for standardizing terminology in the PILOTS database.

• Term Lists ensure consistency in the way that names (of authors, journals, incidents, etc.) are entered in PILOTS. These are simply continually-updated alphabetical lists that we maintain at the NCPTSD. When adding new records to the database, we check all names gainst the appropriate Term Lists.
• The PILOTS Thesaurus is a listing of descriptors -- terms used to describe the subject content of a document -- in the PILOTS database. It consists of two parts: a hierarchically-arranged listing of descriptors and an alphabetical index. The Systematic Table specifies the relationship between broader and narrower terms. The Alphabetical Index lists every descriptor used in the PILOTS database, and for each one gives a list of broader, narrower, and related terms, as well as unapproved terms for which the descriptor is used. It also includes several hundred "entry terms" (non-descriptor terms that a database user might have in mind) with a reference in each case to the appropriate descriptor.

The PILOTS database employs post-coordinate indexing. This means that descriptors usually are short and deal with individual concepts. Combinations of these concepts are employed by users when the database is searched, rather than by the database producers when each document is indexed. Thus the concept of "Rape Survivors" is expressed in PILOTS by using the two descriptors "Rape" and "Survivors" rather than a single phrase. This form of indexing offers users great flexibility in formulating and executing searches. Although it occasionally retrieves papers that do not embody the precise combination of concepts sought by the searcher, it does not restrict searching to those combinations foreseen by indexers.

Natural Language

Natural language searching allows you to use the terms that you are most comfortable with; it does not require you to use the PILOTS Thesaurus. And it provides a way to locate material on subjects that are too new to be included in the Thesaurus, or that the Thesaurus does not cover well enough for your particular need. However, it is neither as precise nor as complete a way of searching as using a controlled vocabulary.

If you simply want to find a few publications relevant to your area of interest, natural language searching is an easy way to go about it. But if you need to make a thorough study of the literature, and you wish to be sure that you do not miss important papers, you should not rely upon natural language searching alone.

Some Things to Note...

Some things to note in designing your search strategy:

• PILOTS bends one of the rules. Theoretically, a document should never be indexed with two descriptors, one of which is hierarchically a narrower term of the other. Thus an article dealing with both cocaine abuse and drug abuse should be indexed only under "Drug Abuse." Because many papers in the traumatic stress literature deal with a broader topic in a general way and then with a specific aspect of it in more detail, it seemed more useful to index such papers under both terms, and this is what we do.
  
  
• Even if you feel certain that the term you have in mind will retrieve the publications you are looking for, we strongly recommend that you consult the alphabetical index to the PILOTS Thesaurus. You might find that the entry for your chosen search term suggests additional broader, narrower, or related terms. You may need to include some of these terms in your search in order to retrieve all the papers relevant to your enquiry. (For example, a search for "Animal Attacks" will not retrieve publications specifically on attacks by dogs, for which the narrower term "Dog Bites" is used.) In addition, many terms listed in the Thesaurus are provided with scope notes that explain precisely how they are used in the PILOTS database.

When you have completed your search, you should have identified several documents that contain material relevant to your needs. Do not expect an exact match: there will no doubt be some among them that turn out to be useless to you. The PILOTS database is constructed with the expectation that your own examination of the search output will be an important part -- perhaps the most important part -- of the whole search process. The goal of PILOTS is to reduce from several thousand to at most a few dozen the number of papers you need to examine in order to find the information you need.

Modifying Your Search Strategy

It often happens that a search of PILOTS (or any other database) doesn't produce the results that you expect. Database searching works best as an iterative process. Don't expect to get definitive results with your first try; plan on doing an exploratory search, and then modify your search strategy according to the results you get. Here are some suggestions:

Too Many Citations

If your search produces an impossibly large number of citations, examine at least a few of them to see whether you defined your topic too broadly, or used too broad a search strategy.

• If almost all of them are indeed relevant, ask yourself how you can redefine your objective. (Perhaps you should choose a narrower topic: for example, natural disasters rather than stressors in general.)
• If many of the citations your search has retrieved are irrelevant, you need to refine your search strategy. Look at some of the irrelevant citations, and see what they have in common. Does the same descriptor appear in all of them? If you repeated your search without using that descriptor, would you be eliminating valuable citations as well as irrelevant ones? If not, you've found one way of bringing your search results down to a more manageable size. (Other methods might include restricting your search by language, or by date, or by format.)

Too Few Citations

What if your search has retrieved fewer citations than you think it should have?

• Perhaps there really are very few papers in your area. (Or at least very few that have found their way into PILOTS.)
• Or perhaps your search strategy was too narrow. Again, look at your results. Find a citation that is directly relevant, and see what descriptors were applied to it. Perhaps you might want to add one or more of them to your search strategy.
• And don't forget to double check to be sure that you weren't done in by a simple typing error.

No Relevant Citations

And what if you could find no relevant citations? Is there a paper that you know to be relevant? Then search for that paper (by author and title), retrieve the citation, and see how it was indexed in PILOTS. That might suggest one or more descriptors to use in searching.

Don't be discouraged if your first search strategy doesn't work perfectly. Experts at database searching often have to modify their search techniques, especially when working with a database that is new to them. And don't be surprised if you come across a citation whose indexing seems strange to you. This is a complex literature, and the indexer is, after all, perforce a generalist. You may well know more about the topic than the indexer does. (If you find a paper that you believe has been incorrectly indexed, please let us know. We don't mind correcting our mistakes.)