parallel-corpus-mining.md

Parallel corpus mining

Metadata

• Status: Proposed (NB: still has to be discussed with relevant researchers)
• Type: Specific
• Work Package: WP3
• Research Coordinators: Time in Translation group
• Coordinators for CLARIAH: Jesse de Does, Vincent Vandeghinste
• Participating Institutes: INT, UU
• End-users: Time in Translation group
• Developers: (Who is involved in implementing this use-case (if any)? Try to mention name, institute, role/responsibility)
• Interest Groups: (a list of CLARIAH interest groups, such as Text and DevOps, for which this use case may be relevant. See the list of IG's at: https://github.com/clariah/ig/.
• Task IDs: Wp3 search engine extensions: parallel corpora; treebanks

Description

Progress in studying verbal tense and aspect semantics can be made by applying quantitative corpus methods in the field of semantic micro-typology, in particular by exploiting the possibilities of translation corpora.

What is the research about?

Tense-aspect categories found across languages.

What problem is hindering the research?

Absence of a flexible, open source and user-friendly environment to explore the corpus data.

What is needed to do the research?

We propose extensions to blacklab/blacklab-server/autosearch

• to enable parallel concordancing
• extraction of relevant statistics
• upload of parallel data created by researchers into autosearch
• exploitation of existing parallel corpora

Data

Parallel UD-enriched corpora (tagging, lemmatization, dependency syntax)

• created by researchers
• existing corpora (OPUS, etc)

Tools

• extended version of blacklab/autosearch
• Visualization and analysis tools developed by the Time in Translation group

What software and services are involved?

How to evaluate this?

• Is the researched satisfied?

References

References to related resources and publications and especially links to related use-cases:

• CLARIAH