Textual Resources

RALI provides some textual resources to researchers in natural language processing. We would appreciate that a mention of the RALI be made when they are used for experiments.

• Aligned Bilingual French-English corpora
  • Corpus BAF: texts of different types: Hansards, judgements, novels and technical manuals (documentation in French)
  • 1893 TREC questions
  • Meteorological Reports and Warnings
  • Bitexts extracted from Wikipedia
• Dictionaries
  • « Les verbes français » by Jean Dubois and Françoise Dubois-Charlier
  • Dictionnaire Électronique des Mots — Electronic dictionary of words
  • XML entries from the GCIDE (GNU version of the Collaborative International Dictionary of English)
• Coreference corpus
• WikiCoref, an English Coreference-annotated Corpus of Wikipedia Articles
• Coreference in Wikipedia: Main Concept Resolution
• Place Names in Canada
• List of names of places in Canada
• Word Associations
  • XML version of Edinburgh Associative Thesaurus (EAT)
  • XML version of the University of South Florida Free Association Norms
• Text Generation
• SimpleNLG-EnFr is a bilingual French/English adaptation of SimpleNLG v4.2 written by Pierre-Luc Vaudry during the summer of 2011
• Results of the evaluation of a data-to-text system for generating narratives from Activity of Daily Living (ADL) data
• Normalization of SMSs
• An annotated corpus of SMSs in French, along with their normalization is available here, along with a normalization demo system.
• Wordnet Browsing Tool, based on XML
• Search engine for the "Réseau des corpus lexicaux québécois"

Linguistic Tools

Rali has a set of tools for the syntactic, semantic and probabilistic processing of natural language:

• English and French morphologic dictionaries and tools for using them to find all morphological variants (conjugation and declination) of a basic form; we can also find the basic form from an inflected variant;
• Probabilistic taggers which assign part of speech information to each word of a sentence;
• Named entities finder for determining proper nouns, dates, money, etc.
• Complex term extractor;
• Automatic accentuation (Réacc) which can add diacritic information into French texts in which this information was lost or never given in the first place.
• Conversion from SMSs to proper French and vice-versa: on the texto4science project page.
• Yasa, a bilingual sentence aligner, aligns two translations of a text in order to produce a bi-text.

For more information about acquiring these resources, please contact Guy Lapalme.