6. Modifying the Results

Even the best planned searches can produce unexpected results. If the percentage of unwanted material is low, it may be better to print it and weed it out manually later. Or, you may find that you got too few references when you know there should be more. If nothing fits what you thought you searched for, look at:

• What is common about the material that you like? [Combine the common keywords with the set using AND.]
• What is common about the material you don't like? [You could use NOT to take something out, but remember that this can also eliminate some good references. Try to use AND instead.]
• What is missing? [Find more variations on your concepts and use OR to bring them in.]
• Could you be missing something without noticing it? A search for SULPHUR and WINE will turn up a few references, leading you to believe that's all there is. However, if you include the variant spelling (SULPHUR or SULFUR) and WINE you will increase what you get.
• Are there references you are aware of that did not appear? Search for them specifically and see why they were not retrieved. If they are not in the database it is a strong indication that you must look further.
• Review the search terms in light of what was retrieved. Is one of the terms being used in a different sense? It may be necessary to step back from your subject when looking at the words; e.g., a worm means something different to a mechanical engineer, to a computer scientist, or to a biologist. Multidisciplinary databases, such as Science Citation Index can contain many such confusing terms.
• Look for a good article, then look at the subject headings. This is one of the best indicators of how the subject headings (often called descriptors) in the list are being interpreted by the indexer.
• Is anything being published on this topic? Searches for topics with little literature or for a new area can be difficult. Conversely, is this a topic that used to be popular but little work is being carried out now. For older material you might have to use the print indexes.
• Is this the correct database to search?
• Database indexing policies are to assign the most specific subject headings possible. To search for articles about any fruit you must also search for apples, oranges, peaches, pomegranates, limes, etc.
• Try looking for opposite terms (antonyms).

Search for only enough terms to cover many of the possibilities, letting the logic operations drop non-related materials from the answer. By searching for specific terms you are requiring that those terms be present in the record. Omitting some concepts is retrieval by implication. Remember the rule of thumb, given above:

• If everything retrieved is exactly on topic, you are likely to have missed some references.
• The closer you get to retrieving everything on a topic, the more irrelevant material you will retrieve along with it.

Previous Chapter Next Chapter