A devoted user in telegram recently asked us some great questions about our latest updates to our telegram translator and our additional features for validating user suggestions.
His questions are as follows:
Question -1. Can the users use the app in any Telegram group and be rewarded with TMN for contributions, or is there are only some selected groups where users get rewarded?
Answer: 1. The telegram translator can be used as a normal telegram application with the added features TranslateMe has implemented. Translation in any group or personal chat. The features to contribute are also possible in any group, each time the user converts conversation with the translation feature there is an opportunity for the user to spot mistakes and submit a correction.
Question -2. As far as I understand, TranslateMe now has a group of sort of validators for each language.
So, are all the contributions/corrections really checked manually (by validators/real people)?
What if one-day TranslateMe will have, say, 100,000 contributors for Russian (and there will be something like millions of contributions
for just 1 language per day). Will TranslateMe still be validating the corrections manually or with the help of AI?
Answer: 2. Validation is currently under development, sentences are manually checked currently to review theoretic approaches to solving validation autonomously. This is important for our next phase which is blockchain migration of suggestions. Recently we updated the application to feature a new section that randomly serves sentences that are translated by machine and can be corrected by users in their native language. This new feature also issues tokens per sentence but this won't be a permanent reward method. As we now move from simply using suggestions to including voting phases. Our algorithm is essentially validating automatically now with this update but rewards of the token TMN is still manual.
The groups we have created are for us to manage some of the tests we need to be done by these users for a specific language, they contribute daily corrections. We review and adjust our algorithm to ensure we have above 90% success rate for suggestions. The groups will be launch pads for different languages as we release updates and request specific testing. All manual validation is handled by our team internally.
Question- 3A. When we go to “Earn TMN” tab/page on the Android app latest version, we see “Suggest and Vote” page.
Are the functionalities of this page still under development/not ready, right?
Answer: 3A. This is the early testing format of our validation process, our goal remember is to validate suggestions automatically before we can move this to our decentralized phase. This section is under construction, we are using it to establish rules and guidelines as we move towards autonomous validation.
Question- 3B. What is the purpose of it (“Suggest and Vote” page)?
Will the users be able to vote for suggestions done by other users, right?
Answer: 3B. The team is working on a controlled list of sentences, these sentences are optimized for conversation. The base list is server-side stored and displayed on request of the user from the suggestions screen. The data is also specially created to focus on conversation based translation. Our short term goal for our commercial API is a service that handles conversation-based translation, this would apply to most tourism platforms, ratings of products and services, support chat platforms, social platforms and dating services. The collected data is more valuable if focused on a specific domain or topic. Suggestions and voting enable us to validate sentences by using community members. The approach is similar to Steemit we have a lot of inspiration from platforms like Steemit where content is valued by using voting too ultimately determine the best sentence suggested automatically.
Question- 4. What if a person can speak 2 languages fluently?
Can he/she make corrections for 2 languages? Or TranslateMe requires making them from 1 person for just 1 language?
Answer: 5. A user can theoretically suggest in any language, we may, however, change this later. As we improve our validation we will be looking at the reputation of a contributor, link this to a language which builds credibility based on that language and factor in a reward bonus for contributors that build consistent positive suggestions.
The update of the telegram has new features which help our team test theory and put it in practice. This loop fast tracks the product and goals.
Moving this phase to an autonomous validation with reward issued on claims for suggestions, votes and staking.
Suggestions reward the winning user who suggested the best correction.
Voting rewards all users who formed part of the winning suggestion.
Staking users will earn a set ratio of all payments made to voters and suggestion winners for each sentence, this will be paid in exchange for securing a distributed private ledger of all transactions recorded for claims of votes and suggestions.
This would be our first phase of a truly decentralized Network
TranslateMe Network will essentially accomplish a decentralized collection protocol that will issue rewards in blocks, based on sentences successfully validated and useful for our project in long term improvement of each machine translation model. We are very close to completing this phase of our goal in creating a truly decentralised translation solution.
TranslateMe Network is the future of machine translation, with a focus on privacy. It has a unique approach to machine translation and the only algorithm that learns from human corrections. Its goal is to create a permissionless translation service using AI and blockchain. The industry is currently valued at 1.5 Billion USD a year. Growing at 19% per year and estimated to double in value by 2025. As we enter the 4th industrial revolution and we witness a shift to Web 3.0 we believe TranslateMe Network will serve as a seamless layer for translation online.