Continuing to advance its search engine experience for users, Google has released research on a new algorithm that could significantly cut down on browsing time for users. The research paper is titled “Generating Wikipedia by Summarizing Long Sequences”.
The algorithm described can generate an original article based on existing content on the internet.
By creating this original article, Google can answer a user’s question without them needing to even click on a webpage. This tool could reduce time spent on the user’s part. On the flip slide, it may affect the readership of information-based websites.
So how does this work algorithm work?
The foundation of this algorithm has already been rolled out into the Google search experience in the form of featured snippets. This preliminary algorithm summarises web content by extracting web content, and omitting the parts irrelevant to the search.
The content you are left with is called an “extractive summary” – it reduces the web content to the most important sentences.
After this process, this content is run through another algorithm – a neural abstractive model, which paraphrases the text and ensures the content is in natural looking sentences.
After the content is run through these processes, an article is created – one that Google says can pass human inspection.
If this algorithm were to be incorporated within Google’s search engine, it may drastically reduce the need for online information sources such as Wikipedia. By creating this tool, Google shows it can create a digital encyclopaedia unique to their user’s search objective.
There are downsides to this process – the possibility of this algorithm including false facts in its original article, for instance. Although similar risks are posed with existing online sources, this tool may further detract users from checking the reliability of their information.
While this technology is incredibly innovative, the implications are a bit daunting. We can only expect to see further innovations in technology in the future.