In this section of 'Making Historical Sources Accessible' I want to talk about databases. They can be a little scary if you have never used one before but the concepts are not too difficult. You may need to find a tame techno-nerd to set it up for you.

The simplest form of database is known as a 'key-value pair' type. This is a direct association between one thing and another. A dictionary is a classic example of this - the key is the word, the value is the definition. It is referred to as a one-to-one relationship.

There is an example on my site at http://www.pascalbonenfant.com/18c/ca... for dictionaries of Georgian Thieves' Cant (the language of rogues - a sort of slang). There are text boxes for 'Cant' and for 'Meaning' and you can search on either of these. (Ignore the other stuff - I'll come to it later).

I use a very simple search mechanism to check for the exact sequence of letters you have entered, regardless of where they appear. In technical terminology this is called a substring search. 'String' is a fancy term for text. 'Substring' is a sequence of letters within that text. To make the search even easier, I ignore capitalisation.

It has its disadvantages - try entering 'cat' in the meaning box and you will get, in addition to cats, catholics, cattle, intoxication and so forth. It is, however, very flexible as you have no need to know the exact word.

The other main type of database is called 'Relational' which means, as you might expect, ways of 'relating' different pieces of information. It is much more powerful than the key-value pair as it allows you relate many items to a single key (known as a one-to-many relationship) or even many items to many other items (a many-to-many relationship).

As a simple example, let's return to the Thieves' Cant database. Below the text boxes you can see four radio buttons, each with an associated date. I used three different sources for this data. Choosing 'Any' - the default - and it will return results from any dictionary. However, if you are looking for material in the early period, the later sources will only get in your way.

When you select the date we now have a second variable on which the result is filtered. We couldn't have done this with just a key-value pair. You can make this sort of search as complex as you like.

I realise this post is a little nerdish and possibly more information than most people want but it helps to be aware of what can be done. If you want to set up a database of your own material and are having difficulty with it, just contact me and I will be happy to help you out.

In my next post, I will be discussing 'metadata' - information which is not part of the original document but which can greatly aid in finding what you want.