Harvard Law School Library

Bracton Online -- Bracton Searching Tips

HLSL Home  |  Bracton Home  |  Table of Contents  |  English  |  English & Latin  |  Search




Simple word and phrase searches

Searching on a single term is self-explanatory.  Entering multiple terms (eg. cats dogs) causes the terms to be OR'd; the engine would, for example, interpret the terms future interests as an instruction to find all documents which contained either the word future, the word interests, or both. To find only those documents containing the phrase "future interests", enclose the phrase in quotes, eg. "future interests".  Other ways of combining terms are explained below.

Terms are NOT case-sensitive; searching on disseisin, DISSEISIN, and DiSSeIsIN will produce the same results. Word searches with wildcards

Sometimes you want to find a number of variants on a "base word" or stem.  This search engine supports a limited wildcarding scheme in which an asterisk takes the place of all letters.  Thus  base* would match all words beginning with base, eg. baseball,  baseless, baseboard, and Basel.

Combining search terms with Boolean operators


You can use a number of so-called 'boolean' operators to combine your search terms in various ways:

Searching in fields

At the moment we use fielding to support searches within particular volumes and tractates, and searching for text in subheads only.  Volume and tractate restrictions are supported directly by choices on the search form, but if you want to enter them "freehand" you may do so. Volume numbers are those of the Thorne edition. Tractate numbers are sequential from the beginning of volume 2 to the end of volume 4; you must add leading zeroes as illustrated below (basically, it's easier to use the form, unless you're building a "captive search" link).   Restricting the search to subheads is done by adding  bractitle/ in front of your terms.  Some examples: How scoring works

We use the Isearch search engine created at CNIDR.  The scoring method it employs was developed by Gerald Salton for use in his SMART retrieval engine.  It can be summed up simply:  pages which have more occurrences of higher-scoring terms are scored higher.  What do we mean by "higher-scoring terms"?  Each term that you give the engine to search by is itself assigned a score indicating its probable importance. Higher scores go to terms which occur less frequently in the overall collection of documents, on the theory that such words are more likely to be important to the searcher than more common ones.

For any given document the term scores and occurrence scores are combined by multiplying them for each term occurring in each document; these per-term results are then summed to give an overall score for the document.  The scores are then normalized by assigning a value of 100 to the highest and rearranging the others accordingly.

As with any generalized search engine used by specialists on specialized material this approach will, from time to time, yield some results which seem anomalous. To do much better requires the development of engines which are tailored to the material being searched, which is usually impractical and always expensive.  In general, the non-specific approach used by Isearch works well.
 
Caveats


Searching for words with ligatures (eg.æ,Æ)

The advanced engine has trouble indexing and searching on words which begin with the ligatures æ,Æ. The problem does not occur if the ligature occurs in the middle or at the end of the word, only if it is the leading character in a term or phrase.

This is a difficult-to-repair bug in the Isite/Isearch software.  The developers and maintainers of the software are working on it, and we expect that ultimately it will be repaired.  In the meantime we suggest that you use the simple search engine. It doesn't support phrase searches or the fancier boolean operators, but you can use an AND search to approximate a search by phrase.

Searching for long phrases made up of common words

It seemed to us that the best way to "chunk" Bracton's text for this electronic version was to divide it at the page breaks of the Thorne edition.  This is a much finer division than is usually made in online texts, and one which is more or less arbitrary in terms of content; breaks can occur in the middle of paragraphs and sentences, which they would not if we were (say) dividing into tractates.  One side effect, then, is that there is some chance that a phrase search will fail (or return fewer results than it should) because some instances span page breaks.  We suggest searching on the least common word in the phrase as a cross-check; other strategies (such as searching on subphrases combined in different ways) may be more or less useful depending on your particular search.

Searching for phrases containing wildcards

An attempt to search for a phrase which includes a wildcard such as "bees in hiv*" will be interpreted as a search for a phrase containing the literal string hiv*, that is, something which has an actual asterisk in it.  Probably the best approach is to break the search up into components, eg. "bees in" AND hiv*.  While this is not quite as restrictive and likely to return some false positives it will work as an approximation.


Contact: specialc@law.harvard.edu
Page last reviewed April 2003.
© 2003 The President and Fellows of Harvard College