The Zen Principle Behind All Online Legal Research
November 18, 2009
Have you ever wondered why your Lexis/Westlaw searches yield too many hits or too few? Even if you have mastered the L/WL help screens on boolean (i.e., terms and connectors) searching or asked for help many times from L/WL Reference attorneys, you will eventually hit a brick wall that is not your own fault. I call it the Zen Principle. Computer scientists may call it something else, but it is an inherent principle in all fulltext, online searching. If you understand it, you will start to feel better right away because you will stop blaming yourself and start to appreciate the limits of computers. Let’s start with the principle itself. I will conclude with some suggestions for minimizing its effect.
The following statement of the Zen Principle has been adapted from Christopher G. Wren and Jill Robinson Wren, Using Computers in Legal Research: a Guide to Lexis and Westlaw (Madison, Wis., Adams & Ambrose, 1994), Appendix M, p. 767, (Biddle Call No.: KF 242.A1 W74 1994).
Recall is the percentage of all relevant documents that are retrieved:
Recall = ————
Precision is the percentage of all retrieved documents that are relevant:
Precision = ————
Recall and precision are inversely related to each other. As recall goes up, precision goes down. As precision goes up, recall goes down. This is the Zen Principle.
For example, say there are 100 relevant documents in a database of 1000 documents. You and opposing counsel are trying to find all 100 documents. If you retrieve 60 out of 100, your rate of recall is 60%, using the first formula below:
60% = —–
If your search also finds 180 irrelevant documents, your total number of retrieved documents is 240 (60 + 180). Your rate of precision is 60 out of 240, or 25%, using the second formula below:
25% = —-
As you try to improve recall above 60%, you will retrieve so many additional irrelevant documents that your precision will drop below 25%. You will not be able to read through all the retrieved documents to achieve 100% recall, that is, unless you are omniscient and can read (and remember!) all 1000 documents in the database. Your pain in striving for perfection will become so great that you will collapse. Your managing partner may conclude, in spite of all your work, that you are an inefficient researcher and too expensive for the client and the firm.
Expansive boolean commands like or improve recall and diminish precision because they increase the number of alternative words being searched: any document which has any search word will be found. Restrictive boolean commands like and, w/n, /p and /s (i.e., search terms within ”n” terms of each other; search terms in the same paragraph; search terms in the same sentence) improve precision and diminish recall because they require all the search words to be present in each document. W/n, /p and /s are even more precise than and because they bring the search words closer together.
To minimize the effect of the Zen Principle, you can try a boolean search which combines both expansive and restrictive commands organized by one or two sets of parentheses. Parentheses help not only to visualize your search but also to tell the computer which commands to perform before other commands. Stop by the Reference Desk anytime for advice on improving your boolean searches. Another tip is to step back and consult some secondary sources in the area of law you are researching. These may be online or in paper. Law reviews, books, looseleaf services, encyclopedias and ALR annotations, to name just a few examples, break out of the Zen Principle in a dramatic way by selecting relevant documents by thinking human beings. Computers merely count embedded words in a mechanical, albeit powerful, way. They have no intuition or creativity. They cannot see the forest for the trees.