rahul1117kumar
Member
- Joined
- 21 Jun 2013
- Messages
- 10,365
- Reaction score
- 11,058
Facebook introduced DeepText, an advanced system built to understand the text posted or expressed by a user and accordingly tailor the experience delivered by Facebook in terms of the products and services being offered. This goes beyond the conventional system of control and moderation enforced by Facebook currently, where all the problems are initially reported and checked by actual humans.
To do this, Facebook built this system of deep learning to understand the context and overall experiences of people with their products as all this communication happens in textual formats. To do this, the system has to achieve near-human precision during the process of contextualising the textual conversations and at the same time going through approximately 10,000 posts every second in 20 languages.
This system can be used in not only delivering Facebook products according to what you are typing or communication, but at the same time, push down spam and irrelevant data while pushing the positive and relevant data to the front. The most used example of this can be the comments sections of posts where there are hundreds of comments and manually sorting through them would be impossible.
This however does not mean deleting data or comments but rather, a sort of preferential treatment is given to the right kind of data focused around our preferences and interaction. Though it may also be described as a silent form of censorship. We can only hope that the system is capable of and designed in a just way without any biases.
The system can also be used to combat hate-speech while being powered by FBLearner Flow, and it’s also being actively used in the Facebook Bot platform where the NLP services come into play to deliver a seamless experience which is not limited to a particular language and can be scaled up to add more languages. This language flexibility was explained to Business Insider India as a Facebook engineer took Deep Text’s “brains” and trained it to speak Indonesian in a weekend.
Facebook thinks that this project is essential at making robots and machines more human and natural in their responses.
Facebook introduces DeepText, an AI engine that understands text and context Tech2 Mobile
To do this, Facebook built this system of deep learning to understand the context and overall experiences of people with their products as all this communication happens in textual formats. To do this, the system has to achieve near-human precision during the process of contextualising the textual conversations and at the same time going through approximately 10,000 posts every second in 20 languages.
This system can be used in not only delivering Facebook products according to what you are typing or communication, but at the same time, push down spam and irrelevant data while pushing the positive and relevant data to the front. The most used example of this can be the comments sections of posts where there are hundreds of comments and manually sorting through them would be impossible.
This however does not mean deleting data or comments but rather, a sort of preferential treatment is given to the right kind of data focused around our preferences and interaction. Though it may also be described as a silent form of censorship. We can only hope that the system is capable of and designed in a just way without any biases.
The system can also be used to combat hate-speech while being powered by FBLearner Flow, and it’s also being actively used in the Facebook Bot platform where the NLP services come into play to deliver a seamless experience which is not limited to a particular language and can be scaled up to add more languages. This language flexibility was explained to Business Insider India as a Facebook engineer took Deep Text’s “brains” and trained it to speak Indonesian in a weekend.
Facebook thinks that this project is essential at making robots and machines more human and natural in their responses.
Facebook introduces DeepText, an AI engine that understands text and context Tech2 Mobile
