Among a handful of patents transferred last December 31 from IBM’s portfolio to that of Google, as first discovered by Bill Slawski of SEO By the Sea, is a system for processing text compiled by users of social networks, and ascertaining their common interests. We’ve already seen the rise of tools such as Radian6 for ascertaining social net users’ individual interests; this new technology, which received a U.S. patent only one year ago, would judge what concepts they share with one another.
The goal of this technology, as IBM originally stated, is to literally to filter out irrelevant links to articles that may not pertain to users’ search intentions. What we don’t know yet is whether Google intends to use this technology, or simply keep others from using it first.
“Many Internet users make frequent searches for information, such as product reviews, hotels and travel destinations, and the like, as well as for on-line services such as shopping sites. Such Internet users are typically inundated with meaningless results for each on-line search,” states the Background paragraph for U.S. Patent #7,865,592, entitled, “Using Semantic Networks to Develop a Social Network.” “Search engines have made searching easier, but a user often needs to sort through irrelevant results and irrelevant Web pages before finding a desired piece of information or a desired shopping site. Thus, even with all of the access to information that an Internet user has at his/her disposal today, many users elect to asking for advice from a friend or acquaintance before beginning a search. However, an Internet user might not know who to turn to for information about a specific subject.”
To understand the patent, you have to think of a “social network” not in terms of a site or a service or a company, like Google+ or Facebook, but rather as a mathematical construct that’s on the same order of a “semantic network.” It’s through semantic networks that Google assesses the context to which search terms belong. An article may appear to relate to the search criteria because it contains multiple instances of the search terms. But a semantic network analyzes the common context of terms in the criteria and the documents being searched, to see if there’s a more solid bond of relationship than mere pattern matching.
The “social network” that IBM engineers were working toward is also a mathematical construct – specifically, an arrangement of related people whose posts to social networks include concepts that may belong to the same semantic network. The flowcharts excerpted here from the IBM patent (now Google’s) present a rough order of events in which social links are established between users whose semantic networks have assessed similarities.
Although the patent doesn’t say so as directly as the intentions outlined in its Background paragraph, the idea here is to do a kind of “StackOverflow.com-like” operation in advance, if you will: Find a user within a social scenario who may have already answered the question, rate that answer according to its assessed relevance, and present it as a solution to the search.
Google has already stated that it is leveraging data from Google+ users to present potentially more relevant search results for Google+ users than it would for Google search users not logged into Google+. If Google were to put this former IBM patent to use, it could convert the search process for Google+ users into a more conversational system – into a kind of dialog where people’s existing answers respond to questions that come up in the future.