GOOGLE Perspective: The tool that tries to put a here until the "TROLLS" of the NEWS - TECHNOTRON
Google: creates a tool in its continuous fight fight against 'trolls' in the news
 |
THE GOOGLE COMPANY
|
It is estimated that about 47% or more have been victims of harassment. 72% have witnessed how these harassment have developed.
So a third of Internet users self-censorship themselves for fear of harassment and its consequences.
These high figures presented by the Research Institute Data Society of the study of Online Harassment and Digital Abuse, are the ones that ended convincing GOOGLE that something had to be done and not be held in arms, in the fight against the TROLLS of the Internet And make the Internet a safer and more pleasant place to use.
The technology giant launched the "Prespective" Thursday a tool that uses Artificial Intelligence to review and identify "Toxic and abusive" comments in the news for the sole purpose of being reviewed and to be able to exclude them from the conversation.
The giant has imposed itself the challenge of creating a good conversation on the Internet, there are many people who try or try to engage in good discussions in the comments section of web pages, blogs, but in the end end up leaving the conversation and retiring Because of other people who are only to insult and disrespect the others.
Jared Cohen founder and president of Jigsaw (formerly known as GOOGLE IDEAS), I explained by telephone conference the purpose of developing the code that can stand up to the trolls, excluding them from the conversations for their aggressive and intolerant attitude to others.
"Perspective is available to any publisher and content developer who wants to make their service can be used for free".
As already said Perspective is free to be used by any publisher and developer of content because it is open source.
Which means it is not a technology that I believe GOOGLE for its exclusive use, since the goal is to empower publishers of publications and platforms so they can determine what can be discussed or can not be on their websites, mentioned Cohen .
Well now publishers can decide what to do with the information they get from Perspective. From Jigsaw have thought many options on the comments: the options range from marking the comments for moderators to review and decide to exclude them from the conversation, allowing users to see the potential toxicity of comments, when they write, another option is To classify the comments based on a criterion, to facilitate the search for great discussions hidden in toxic conversations.
Learn to read insults
The operation so far explained in theory and simple, a group of people will be in charge of reading and reviewing the comments, in order to create a database that will include those that are marked as toxic - for containing insults, obscene words, lack of respect.
After this, Perspective will analyze the thousands of comments that exist on websites.
So that you know what the toxics are, and you can evaluate them by comparing them with those already found in your database, there is similarity with what is already considered harmful, after identifying them, it adds them.
So if someone is telling other comments like ugly, stupid, idiot or some adjective decafilicativo will be identified with some of the toxic comments that the system already has stored in its database, that is already a way to train it, Cohen said.
"The use of artificial intelligence allowing to identify insults and toxic comments, while more everyday use is better results".
As already mentioned, the key to Perspective's success is to use it in machine learning mode. Thus Artificial Intelligence can analyze more data and at the same time create patterns as the hits succeed.
The more you use it, the more you will learn. The data (comments) will be to say their food.
Perspectiva still does not work to perfection because it is in the initial stage, but over time as more data comes to have, the better will be its operation.
Cohen emphasizes that the more comments being analyzed, the more perfect it will be.
While in Beta or training phase, the creators of Perspective know that this will make mistakes, so they are aware of their work of revision.
This phase can be carried out by the moment by the same editors as the users who find themselves commenting. If users or publishers consider that words like "UGLY" is not an insult, Perspective would not consider it as such. This type of dependency raises the likely possibility that Perspective will make erroneous decisions about what can be admitted or can not be admitted, as was previously the case with "TAY" the bot made by Microsoft, which was kept in learning by observing Twitter conversations and As a result of this, I end up proclaiming insults and denying the "NAZI" genocide.
On the hypothetical case that there is a possibility of failure, the founder of Jigsaw maintains confidence in his product, "We offer the possibility of improvements and according to the data that we manage to manage and obtain on the basis that the publishers or users do not like or Which they consider insulting. "
Only available in English
As always and like all the things that the great companies launch, perimero are available solely and exclusively in the English language, it is known that it will soon be in other languages, but there is no specific date yet.
If publishers and platforms from other countries demonstrate their interest in collaborating, they may be starting to collect native data from other languages to build their model or version in another language, as was done earlier with "ENGLISH" and "THE NEW YORK TIMES ".
Jigsaw has begun testing this technology for some months with the American newspaper, which already had its team in charge of examining, moderated the comments of its users before these comments are published.
The newspaper's staff analyzes about 11,000 real-time comments each day. Amount that has allowed only the comments of 10% of the articles are allowed. Perspective in your objective to help you classify faster and at the same time nurturing your database and algorithms with all these examples.
This initiative that has taken is nothing new, companies like Twitter and Mozilla have already taken similar actions.
In the case of Twitter this decision to finish plantandoles face to the "TROLLS" was motivated to see that big companies like "Salesforce" and "Disney" decided to acquire it for the many problems of moderation and added to the dalta of a healthy community.
That is why Twitter made available to usaurios tools to silence conversations and to mark them as intolerable.
This way you can leave the thread of the conversation without receiving notifications and without having a very well-called "blacklist" of words that contain hurtful words to be not related to your profile, nor to be able to citete with them, in spite of these Many users consider that not seeing something when a message is silenced or when someone is blocked is not a definite brake for this.
I would like to know what you think of these actions to stop the "TROLLS" you think they give the result that is expected or do not believe it.
If you liked this article you can share them in your social networks.
Comments
Post a Comment