This section deals with the advanced topic of SkXML. You should only tackle this section if you have are completely confident with the DPS System, and have some prior technical knowledge of XML and query languages. Here you will learn:

About SkXML

SkXML is a sophisticated query language that combines powerful text-searching facilities with an awareness of a the data's XML structure. DPS uses SkXML in two ways:

 

This section concentrates on the second use - as 'evidence finder' - where SkXML's power is fully revealed.

What is a 'Corpus'

In terms of language and writing, a 'corpus' refers to a body of work or knowledge relating to a particular subject. The DPS System treats any DPS project as a corpus in that it considers the project to be the whole body of work. Many of these projects will be dictionaries that have been compiled using DPS, but others might be dedicated corpora; DPS does not distinguish between the two types. Using SkXML, you can search any, some, or all of the projects loaded on your DPS server.

Accessing the SkXML Search Screen

Tip: it is often advisable to open SkXML in a new browser tab or window; right-clicking on the SkXML Box button should give you the option to do this.

The SkXML search screen consists of three parts:

Searching with SkXML

To perform an SkXML search:

Default Results View

The SkXML language contains facilities to control how query results are displayed. If you do not use any of these, the default is to display the label of all entries that match your search: that is, all those entries whose contents meet your search criteria at least once.

In addition to viewing search results, the following options are available:

The SkXML Query builder

The SkXML search language is quite complex so there's a query builder to help you:

 

This query builder should be enough for most searches, but SkXML is more capable and to use that you will need to learn (or look up) the SkXML language below. The query builder will always  be a good place to start and then you can add the bits  that are not provided via the interface yourself directly in the search string that has been built up.

The SkXML Language

Having seen how to perform a search, it is high time to tackle the syntax of the SkXML query language - if you don't think you will have to do very sophisticated searches from the start, you can skip this section and return if your needs change.

Note: this is quite a complex topic, and is best tackled when you are feeling confident.

Specifying Words

SkXML searches for words, not for individual characters: searching for 'a' will find the word 'a', but not words that contain the letter 'a'. However, the language contains sophisticated wildcard facilities – usually known as regular expressions – that overcome this apparent limitation; the trick is to make sure that your combination of letters and regular expressions defines a whole word, not just part of a word. (Note that SkXML treats punctuation as words. The text 'me; but' is treated as three words: 'me', ';', and 'but'.)

The following 'building blocks' can be used to specify a word:

In the remainder of this description of the SkXML language, the term word is used to mean a word as specified using letters, regular expressions, or both.

Case- and Accent- Sensitivity

The rules given above aren't quite truthful: by default, letters are matched case- and accent-insensitively, so any letter or regular expression that will match 'a' will also match 'A', 'á', 'Á', etc. The following flags qualify whole words to control how they are matched:

Inflections and Compounds

The following flags indicate that common variants of the specified word should also be matched:

Numbers

If a word is a number, various numerical qualifiers are available:

 

Two qualifiers can be combined: >5<=10 will match '6', '7', '8', '9' and '10'.

Sequences and Proximity

We now have a powerful syntax for matching single words. SkXML can also be instructed as to how two or more words should relate to each other. As you would expect, a simple string of words matches itself: &#timeo &#Danoas &#et &#dona &#ferentes will match 'timeo Danaos et dona ferentes' (Virgil, Aeneid, ii.49). Other options are:

Note the spaces before and after the ;, ;;, etc.

A sequence of words specified with this syntax is called a 'Phrase'. If a phrase is preceded by ~, it will be matched even if some of its component words occur in sub-elements.

Boolean Operations on Phrases

Again, note the spaces before and after the | and ,.

XML Elements

Matches can also be limited to specific XML elements:

Here are some examples to show how these basic operations can be combined:

Count Filters

The number of times a word or XML element must appear can be specified by appending {{<occurrence filter>}}, where <occurrence filter> is an expression consisting of numbers and numerical qualifiers (see above, Numbers).

Important note: precise positioning of parentheses is very important!

The similar [[<occurrence filter>]] filter specifies which instances to match:

Note that the first instance is '0', not number '1'.

These two filters can be combined:

 

Annotations

We give two versions: simple and more elaborate which uses brackets: using brackets is necessary if you want to combine things.

 

Keyword-in-Context

You will recall that, by default, SkXML displays the entry label of all entries in which one or more matches of your query were found but does not show the matches themselves. To show the matches you must specify a so-called "match point" in your query, using the % operator. The position of the % is important: the word or XML element immediately following is colour-coded and used as the focus of a keyword-in-context display. Let's see some examples:

The SkXML query %albatross might produce results like this:

Note in this example that the XML tagging is included in the match point. If you want to include just the content, specify <%def:albatross.

If the match point is an XML element, you can further highlight significant words or phrases within it. The query <%def:%%albatross might produce results like this:

Click to enlarge

You can specify more than one match point in a single query, in which case SkXML will create a separate display column for each match point. The query %.* %albatross might produce results like this:

Click to enlarge

The default column names (xml_0, xml_1) can be replaced by names you specify; for example, %{Preceding}.* %{Albatross}albatross:

Click to enlarge

Controlling the Keyword-in-Context View

The Results layout list-box (top right of the screen) offers two keyword-in-context views:

The amount of context shown before and after the match point can be varied by adjusting the context and fixed context settings.

Click the Change button to apply your changes.

Other Views

The Results layout list-box offers two other options:

Note: that these XML views do not show any context.