It’s ‘Social Glossary’ scans users’ posts and comments to look for frequently used terms that are not formally part of the English dictionary and stores them into a glossary. Some examples of popular terms include LOL (laugh out loud), OMG (oh my god), totes (totally) and POS (parent over shoulder).
According to the latest version of the patent, which was published overnight, “information related to one or more textual terms in the glossary is provided to enhance auto-correction, provide predictive text input suggestions, or augment social graph data”.
If a term starts to fade in its popularity or its use declines significantly, it will be removed from the glossary, according to the patent.
“Each new textual term may be required to exceed a certain usage threshold before it qualifies as a neologism [slang or informal words] and not a mistake.
“The usage threshold may be determined based on usage by: users of the social networking system, users having a relationship with the [other] user according to the social graph, users using the same language as the user, users located within the same geographic area as the user, users sharing a demographic category with the user, or users sharing similar profile attributes with the user.”
Users would also be able to poll for terms they think should be added to the glossary, the patent stated. These polls could be done within users’ own social groups.
“The distribution group for invitations to participate in such a user poll may be determined on the basis of a defined subgroup, e.g., users within a particular group, users having a particular affiliation with an organisation, users sharing a particular interest, users within a particular locale, users falling into a particular demographic group, users speaking or writing in a particular language, users connected to one or more social sub-networks, or users sharing any other type of characteristic.”
Facebook plans to use its Social Glossary to predict the next word as a user types text into a status update, or when making a post or comment. It could also auto-correct users on their slang as they type or suggest a new term that might help the user better convey what they are trying to communicate.
Read more:Hadoop is not enough for Big Data, says Facebook analytics chief
“If a particular textual term ‘rickrolled’ was added to the Social Glossary as a verb related to an action that one user can perform upon another user, a ‘rickrolling’ edge type may be added to the social graph, so as to capture such actions or activity when occurring between users of the social graph,” stated the patent.
Search for distributors in the Distributor Directory