Hugo Aujourd'hui
Prototype >> IA Assistant pour Multilingue Traduction Publication
|
IA Assistant pour Multilingue
Traduction Publication
|
| v1.00.2021-02-10 v1.01.2021-02-11 v1.02.2021-11-11 |
| #nlp #multilingual #webeditor #hci #machinelearning #ai
#concurrent . |
| Bonjour,
NLP enthusiastes, 《IA Assistant pour Multilingue Traduction Publication》, a challenging task, lots of the work are repetitive、 automatable、AI’able. 1 lengthy article I am composing, ie. 1 testbed named “Yan Runtao vs Elizabeth Helen Blackburn” for Labo experiment. https://hugoaujourdhui.org/blogs/histoire/yan-runtao-vs-elizabeth-helen-blackburn.html 3 questions for us to trigger brainstorming, all of your input, comments, critics are welcome. Stay safe, enjoy every creative day:) |
Convention: CO: Contexts Original CX, CY, CZ: Context translation result in 3 diff languages CF: Context in French, symbolic language CI: Context in Italian, symbolic language CE: Context in English, symbolic language CC: Context in 漢語 | Mandarin, ideographic language DT: DeepL Translator
GT: Google Translator
HT: Human Translation
BT: Better Translation
TC: Translation Criteria
|
Translation & Publishing Process 1) Compose CO, 2) Translate CO to CX 2.1) use DT to translate CO, get CX1, 2.2) use GT to translate CO, get CX2, 2.3) compare among {DT result CX1、GT result CX2、HT initiative}，according certain TC, pick up 1 relatively better one or mixed one as BT 2.4) machine/human translate BT back to CO1, compare CO1 vs CO, verify translation quality 2.5) if there are some syntax mismatch and/or morphology mismatch and/or rhetoric mismatch, revise BT, until human satisfied BT as CX ~= CO 2.6) based upon promotion and/or ISO standardization for embedded-symbol universalism to polish CX 2.7) semantic touch-up on CO & CX, ie. add URL links to some components inside both CO & CX 3) Translate CO to CY, go through same steps as what 2.1 ~ 2.7 do, until CY ~= CO, polish, touch-up 4) Translate CO to CZ, go through same steps as what 2.1 ~ 2.7 do, until CZ ~= CO, polish, touch-up Note: The steps 2)、3)、4) are concurrently executed on HPC facility. 5) Export the {CO、CX、CY、CZ} results as 1 synchronized、united HTML file for presentation & publishing |
|
3 Questions
Q1: HCI Tool On the market, is there any tool(s) already available for automation for above repetitive Human-Computer-Interaction process? Q2: Editor Here, we have 4 languages involved in the whole process. We edit those 4-languages Context in a Table, then produce HTML file for presentation & publishing. Is Table the best vehicle to "data feed" translation results, then present & publish? Q3: AI Assistant There should be possible AI assistant be used for this whole process, does AI industry already have some practical application(s)? These 3 questions for us to trigger brainstorming, all of your input, comments, critics are welcome. |
|
If you are interested to join
adventure in this challenging R & D domain, contact VULABO
in hugodemain@yahoo.fr
Stay safe, enjoy every creative day :) Cordialement.
|
Natoue ne unatibus
Mauris urna purus rutrum vitae iaculis
Nestibulum laoreet risuis convallis consectetur in nuege. Naim placerat ulamce sollicitue duipretium licituin necerais feugiamt congue puru volutpat sucipit convallis anmegte alsqiu sevitae nulla libero tincidunt laoret convallis.
MORBI SEM
Sern soituin
Nestibulum lareo risu conis nuege naim placerat ulamce sm duipretiu licituin necerais feugiat a voluitpat suscipit convallis anmegte sevitae nulla libero tincidunt laoir placerat ulamce sollicitue duipretiu.
Fusce mauris
Nestibulum lareo risu conis nuege naim placerat ulamce sm duipretiu licituin necerais feugiat a voluitpat suscipit convallis anmegte sevitae nulla libero tincidunt laoir placerat ulamce sollicitue duipretiu.