Apr 28, 2021 07:00:00

'Contextualization engine' concept that provides judgment material to counter false information

The founder of the Thoughtful Technology Project, an organization that aims to prevent new technologies from irreversibly harming the information ecosystem, said, 'Search engines are great technology, but they are not enough as they are.' Proposes ' Contextualization engines '.

'Contextualization Engines' can fight misinformation without censorship | by Aviv Ovadya | Apr, 2021 | Medium

https://aviv.medium.com/contextualization-engines-can-fight-misinformation-without-censorship-c5c47222a3b7

Advocate Aviv Ovadya said that the search engine that revolutionized the 2000s and the recommendation engine that revolutionized the 2010s both focus on 'discovery' and rely on relevance. , Preface that the handling of incorrect information is not enough. He claims that what is needed is a tool that can help him quickly 'contextualize' the media he encounters.

'Contextualization' refers to understanding what a sentence or word plays in the context and what it means.

Ovadya's ideal context engine is to target any media object, such as article links, videos, meme images, music files, PDFs, etc., with just one or two taps from the app. It allows users to understand 'how much this makes sense' and 'how it relates to what they know and care about'.

The system built for contextualization not only assists in media literacy, but also provides material for fact checkers to decide what to look for, and emotions to avoid radical reactions to false information. It is also useful for supporting literacy.

This contextualization system is not externally censored or monitored, but is called

'further speech ' to counter false information by providing a place where users can access context and trusted information when needed. It embodies the approach, says Ovadya.

The development of such an engine that performs powerful context analysis has become possible due to the progress of artificial intelligence (AI) technology, but on the other hand, the progress of AI technology has realized a terrifying new technology that spreads false information. It also helps. For this reason, Ovadya describes building contextualization capabilities as a 'game against time.'

Specifically, Mr. Ovadya gives an example of what the context engine will be, 'If the article shared on SNS is something that stirs anger.' Even if you get angry, you should check whether the content is correct, but considering the time and effort you have to verify yourself, it is easy to leave yourself to the flow.

If there is a contextualization engine there, it compares the content to content from a solid source and provides relevant articles and other media results. The methods offered can be something like search engine search results, something like a chatbot, or a hybrid of them. If it is determined that it is not sufficiently relevant, it will warn the user.

What you're doing seems to be similar to search engines like Google, but according to Ovadya, contextualization engines analyze 'media objects' to see if they might be relevant. Focusing on authoritative sources, such as using third-party whitelist authentication, alerting you if you can't find the right information on a topic, important for fact checkers and other organizations to investigate It is different in that it provides various information and supports people who conduct deep investigations.

Ovadya states that the content given here is just the beginning of the possibilities and may also support other parts of the 'SIFT method'. The SIFT method is named after the acronym of four words.

· S top: The context engine encourages the user to pause and notice their emotional reaction to the content.
· I nvestigate the source (to investigate the source): context engine is, if the information it can be determined that the 'source can be trusted' is already, why it indicates the reason that one can trust the user.
· F ind better coverage (better article retrieval): A more complete context engine having a function not only automatically generates a raised voice and video text, the source associated with better understand the context of the content It also automatically interprets images and captions to find them.
· T race claims, quotes, and media to the original context ( assertion, citing media keep track of the original): context engine is, locate the source origin of any content from the web.