Semantic and Personalized Search are Fool’s Gold
“Fool’s Gold” is submitted by Adam Jusko of Bessed.
Recently a site called EventSetter.com requested inclusion in Bessed, our human-powered search engine. EventSetter is a site along the lines of Upcoming.com or Eventful, offering listings of upcoming events such as concerts, conferences, community festivals, etc., in cities throughout the United States. While adding EventSetter to our index may sound uneventful (pardon the pun), it actually presented a quandary, which led to me realizing that semantic search, personalized search, or any other search engine technology that attempts to understand a searcher’s intentions will never be totally viable. Here’s why…When people submit their sites to Bessed, we build a results page that we believe most closely corresponds to what someone might search for in order to find that site. For example, if CrocodileBoots.com wanted inclusion in Bessed, we might build a page of search results for the keyword search “boots” and include CrocodileBoots.com as one of the results. Easy, right?
But what search would a person conduct that would lead them to find EventSetter.com? We used various keyword tools to see what people search for around the word “event” r “events.” We found that many people search for “upcoming events”, “calendar of events” and “event calendar.” We created a page for the search phrase “calendar of events” and included EventSetter as well as other sites that aggregate event listings.
But that didn’t satisfy us. Because we really have no idea why so many people search for “upcoming events” or “calendar of events.” When they do those searches, are they looking for a general site that lists events in many cities? Are they looking for local events and imagining that the search engine will magically know where they are located? Or are they searching for software that would help them add an event calendar to their Web sites? We have no way of knowing, unless that searcher qualifies the search further by typing in something such as “upcoming events new york.” So, if we as humans have no way of knowing what people want when they do a search for “calendar of events,” how is a search engine algorithm, no matter how sophisticated, ever going to be able to figure it out? It can’t. Which shows that, even if they are possible, next-generation semantic search engines that use artificial intelligence aren’t going to get us much further along the curve. Likewise, personalized search results (using your past searches to help understand the meaning of your future searches) might be helpful, but unless you habitually search for similar things, it’s not going to significantly improve the search experience, and isn’t worth the trouble for most searchers.
In short, no search engine is ever going to be perfect at understanding searchers’ every query. There’s no oncoming Utopia in which a search engine knows your every thought (and maybe that wouldn’t be Utopia anyway).
However, search satisfaction will continue to grow—from incremental improvements by the engines themselves, and, just as important, from searchers who will get better and better at phrasing their queries in ways that make it clear what they want.
