Language Pairing Interface – An Interface for Language Pairing in Machine Translation

The project context
This project is a part of a Ph.D. thesis in Artificial Intelligence at Bircham International University (BIU) - Madrid (a Distance Learning Program), with the theme "Agent-based Multimodal translation in Natural Language Processing". BIU is a Top 10 World Ranking Distance Learning University. However, this platform will stay available postdoc for the benefits of potential contributors and the AI community.

What is the interest for such a project?
The problem of automatic induction (inference) of a grammar on the basis of a sample of language sentences is still an open problem in the area of Artificial Intelligence (AI)¹(Please, each time click on the footnote number (1) for details). And Human languages are diverse with about 6000 to 7000 languages spoken worldwide². But only about three (03) hundred languages (approximately 4 percent) have an official wikipedia that could be used as a reliable corpus for Natural Language Processing (NLP) tasks. So, the call for contribution is still required if we want to take into account all spoken languages.
Moreover, through this project we will have the possibility to contribute to creating a richer corpus, ready for machine learning tasks including the Baoule language. This work will be available to any machine translation project. In addition, it is a technique that will be continuously improved thereafter.

What is Language Pairing?
First, let us define Artificial Intelligence as a discipline for creating Machines or Agents or Systems that Think and Act like Humans³. In Artificial Intelligence an immense significance is attached to natural language and an intelligent system’s ability to use it⁴.
In Machine Translation (MT), in order to produce high quality translations, it is important to use reliable Language Pairs produced by an Expert in Language Pair Translation. A Language Pair is a dataset of sentences from two (02) distinct languages (bilingual lexical data), where each sentence is reciprocally translated from one language (the source) towards the other (the reference), the best related prediction, candidate or automatic translation being the target (i.e., the predicted text into the Target Language that we expect to be of high quality or very close to the reference). In case of several references per source, the target still needs to be, not any of the references, but the targeted prediction that is the closest possible to all these references, i.e., that better matches to all of them at once. In theory, a MT system that predicts the so defined target could make similar or even better propositions than the human expert references. The Language Pairing task itself consists in the production of Language Pairs.
But things are not as easy as we can think. As said M. Dion Wiggins, CTO and Founder of Omniscien Technologies: "High-Quality Machine Translation is Complex, any vendor that tells you that MT is easy is either trying to fool you or is incompetent. Time, Effort, Skill and Investment are Mandatory".
Please, you can find the related paper here: Baoulé Related Parallel Corpora for Machine Translation tasks: mtBCI-1.0.

An Overview of the Artificial Intelligence (AI) field

Credit: Konan Jean-Claude, Kouassi. (February 2023) “Understanding the Worldwide Paths towards the Creation of True Intelligence for Machines.” International Journal of Computer Science & Information Technology (IJCSIT), Vol. 15, No. 1, pp. 43-68, Academy and Industry Research Collaboration Center (AIRCC), doi:10.5121/ijcsit.2023.15104.

The above IJCSIT Journal Paper is the Extended version of a 2022 CACIT conference paper, upon selection by the Journal itself, as the conference initiative for outstanding papers with the requirement of at least 30% difference from the conference paper.

ORCID iD: https://orcid.org/0000-0002-4744-1335.

We present below just a sample of the collected data structure.

N°	English	French	Baoulé Latin Script Translation	Baoulé French Script Translation	Category	Sub-Category	Proposed by
30	For contributing, you should first understand the Baoulé language.	Pour participer, tu dois d'abord comprendre la langue Baoulé.	Sɛ a kounndɛ ye wie, ɔ fata kɛ a ti Baoule aniɛn i kwlaa ka.	Sè a kounndè yé wié, ô fata kè a ti Baoulé aniein i klouaa ka.	Crowdsourcing	Creativity	Konan-Jean-Claude-Kouassi
29	Don't insulte people.	N'insulte pas les gens.	Nan kpɛ sran mun be nzoua.	Nan kpê sran moun bé nzoua.	Crowdsourcing	Creativity	Konan-Jean-Claude-Kouassi
28	What shall I write ?	Que dois-je écrire ?		N'klè n'zou ?	Crowdsourcing	Creativity	Konan-Jean-Claude-Kouassi
27	I drink water.	Je bois de l'eau.	N'nɔn nzue.	N'non nzoué.	Crowdsourcing	Creativity	Kjeanclaude
26	How to contribute to this interface ?	Comment contribuer à cette interface ?	Be ye like nga i se ?	Bé yé liké nga i sè ?	Crowdsourcing	Creativity	Konan-Jean-Claude-Kouassi
25	Language or tongue? Both, because you must first understand the language before speaking it, using your tongue.	Le langage ou la langue? Tous les deux, car tu dois d'abord comprendre le langage avant de le parler, en te servant de ta langue.		Aniein annzê tafliman? Bé ngnon, afi'n ô fata kê ê ti aniein i kwlaa ka nan a idjô, ôni wô tafliman.	Crowdsourcing	Creativity	Kjeanclaude
24	Good !	C'est bien !	Ɔ ti kpa !	Ô ti kpa !	Crowdsourcing		Konan-Jean-Claude-Kouassi
23	Please Sir, may I go out ?	Monsieur, permettez-moi de sortir s'il vous plaît. S'il vous plaît Monsieur, pourrais-je sortir ?	Minsie yaci, n'koundε fite. Minsie yaci, n'kwla fite ?	Minsié yatchi, n'koundê fité. Minsié yatchi, n'kloua fité ?	Crowdsourcing	Creativity	Kjeanclaude
22	I am watching TV.	Je suis en train de regarder la télé.	N'su nian television.	N'sou nian télévision.	Crowdsourcing	Creativity	Anonymous
21	I am going to sleep.	Je vais dormir.	N'su ko la.	N'sou ko la.	Crowdsourcing	Creativity	Kjeanclaude

DATA CLASSIFICATION BY CATEGORY
DATA CLASSIFICATION BY SUB-CATEGORY
DATA CLASSIFICATION BY DATA SOURCE
DATA CLASSIFICATION BY DATA QUALITY
STATS ABOUT Latin Script Baoulé Translations
STATS ABOUT French Script Baoulé Translations

CHART PER CATEGORY
CHART PER SUB-CATEGORY
CHART PER DATA SOURCE
CHART PER DATA QUALITY
THE 10 GREATEST CONTRIBUTORS
Google Chart Profile