Natural language processing changes how shoppers search
Natural language processing on search queries goes beyond enabling shoppers to type in a query in much the same language they’d use in asking a question verbally. In fact, that view doesn’t capture all of the things natural language processing can do to improve search, according to Tony Frazier, senior vice president of marketing at iPhrase Technologies Inc., where natural language processing is core to the search and navigation technology it provides.
“Initially some of our customers question whether their site users will use natural language search requests in full sentence queries,” he says. “We help them understand that even if people don’t use full sentences, they’ll use a phrase and doing a lot of the same processes on those queries still brings back better results.”
For example, he adds, web shoppers may not know how to spell a brand, but natural language processing can set the rules for misspellings to “clean up” a user’s input to facilitate a better response. In addition, he adds, shoppers who become accustomed to natural language processing-based search on a site learn with experience to format their queries accordingly so as to get to more relevant results more quickly.
“Over time, people learn with reinforcement on a site that if they ask more expressive queries they will get more direct answers,” Frazier says. “When one of our customers first deployed with us, only a quarter of their queries on the site were more than two words. Six months later, more than 75% of queries were greater than two words.” .
Back...
