Laboratoire ICAR » Plateforme TXM

Currently "text" is the only structure used to cut concordance contexts. In some contexts, in can be useful (necessary) to cut or to show limits of concordance contexts.

Usecase 1: alternate contexts

In survey corpora (CSV import), all the answers of one respondant can be organized in a same 'text' separated by a separator tag. When we display the concordance results of a search, contents of different answers must not be mixed in the same concordance context.

To do so, we should be able to:

• cut concordance contexts to answer limits
• or display answer limits within the same 'mixing' concordance context

Usecase 2: speech turns

In record transcriptions or drama (TRS import), speaker speeches are separated by a <sp> tag. When we display the concordance results of a search, contents of different speech turns must not be mixed in the same concordance context.

To do so, we should be able to:

• cut concordance contexts to turn limits
• or display turn limits within the same 'mixing' concordance context

Solution

• choose which structures are used to cut concordance contexts: for example "text" or "plu" & "deplu" or "p"
• choose which structures are used to mark concordance contexts limits: for example, the <sp> structure can generate a "...|..." marker (meaning 'the concordance context part on the left of the '|' is from another speech turn - for contexts displayed on the left of the pivot)
  • an evolution could be to add to the limit display a pattern of structure attributes: example "...@Roméo ..." or "...|Roméo ..." or "...|@Roméo ..." (context after '@Roméo' mark is said by Roméo)

The structures to use can be defined :

• in each corpus preferences (may be declared at import)
• in TXM general preferences