Manual for DeriNet search tool

How to search?

The basic query type is an attribute expression. This describes a single node in a cluster / a single record in a database. Attribute expressions can be nested, creating a tree expression, which describes relations between nodes.

Attribute expressions

An attribute expression describes a single node in a tree. It has the following form: [attribute1="regex1" attribute2="regex2" …], which matches a node, whose attributes match the corresponding regexes. Available attributes are lemma, techlemma, pos (part-of-speech), id a parent. id and parent expect a node ID from database as an argument; [parent=="143581" pos=="N"] is equivalent (almost – they differ in highlighting) to [id=="143581"] [pos=="N"]. A special case is [parent=="-1"] – this matches nodes without a parent.

Example: [lemma="vana" pos="N"] matches all nouns, which contain “vana” as a substring.

Apart from = there are other comparison operators: == for string comparison (strings are equal, when they contain the same characters), != a !== for negation of the previous two.

The form "regex" is available as a shortcut equivalent to [defattr="regex"] and string is equivalent to [defattr=="string"]. You can choose default attribute defattr in a dropdown menu under the search input field.

Tree expressions

A tree expression describes a subtree. Every attribute expression is also a tree expression; if you concatenate several attribute expressions: balit "ík$" [pos=="N"], you'll find a tree, whose root is the first expression, its son is the second one, which has the third one as its son…

the second form is aexpr (texpr1, texpr2, …). This matches a subtree, whose root matches the attribute expression aexpr and every tree expression texprX matches one of its sons.

Examples: (clickable)

[] [pos="V"] [pos="N"] [pos="A"] [pos="D"]

Matches a subtree of height at least 5, which starts with any root, continues with a verb, which has a noun as a descendant, then an adjective and finally an adverb. The adverb doesn't have to be a leaf and the root of the subtree isn't necessarily the root of the whole tree.

[] ([pos="V"], [pos="N"], [pos="A"], [pos="D"])

Matches a subtree of height 2 with any root and four siblings with the corresponding parts of speech.

[][][][][][][]

Matches any tree of height 7.