TRACK 2 – 3rd Translation Inference Across Dictionaries (TIAD 2020) shared task

Task Description

The objective of the task is to explore and compare methods and techniques to infer indirect translations between language pairs, based on existing bilingual resources. Such techniques would help in auto-generating new bilingual and multilingual dictionaries based on existing ones.

In particular, the participating systems will be asked to indirectly generate translations among three languages, namely Portuguese, French and English, based on already known translations contained in the Apertium RDF graph (http://linguistic.linkeddata.es/apertium/). The three chosen languages (EN, FR, PT) are not directly connected in the Apertium RDF graph (https://tinyurl.com/apertiumRDF-lang), therefore no direct translations can be obtained among them in Apertium RDF.

Based on the available RDF data, the participants will have to apply their methods and techniques to discover indirect translations (mediated by any other language in the graph) between the pairs: (EN, FR), (FR, PT), and (PT, EN).

In addition, participants are welcome to make use of other freely available sources of background knowledge (e.g., lexical linked open data and parallel corpora) to improve performance, as long as direct translations among the considered language pairs from such extra resources are NOT used.

The inclusion of other language pairs in the evaluation could be considered, in which case this will be conveniently announced to participants.

Evaluation of the results will be carried out by the organisers against manually compiled pairs of K Dictionaries from the Global Series (https://lexicala.com/resources#dictionaries) and other resources.