Travelers are always looking for the most “bang for your buck” when booking a trip. Are you looking for a hotel by the beach, close to downtown nightlife, with easy access to public transportation, that’s not too far from the highway, provides excellent customer service, and offers a great price? But are you lacking the hours to sift through endless search results on the Internet to find the best place with the best value for money? A new ranking algorithm for product search engines created by NYU Stern Professor Anindya Ghose and his colleagues, Panagiotis Ipeirotis and Beibei Li, now serves up the best value-for-money options with the first click. This novel system mines online user reviews and other social media content across a multitude of websites, and is the first to quantify the economic impact of multiple consumer preferences, such as location and service attributes, on hotel demand.

“There used to be a time when there was not enough information. Today, there is information overload on the Internet. However, by automatically mining the plethora of unstructured data on the web, we can harness the wisdom of the crowds to work to the consumer’s advantage,” says Professor Ghose.

With more than 87 percent of customers relying on online user-generated content to make purchase decisions for hotels, higher than any other product category, Ghose and his colleagues focused on travel search to develop their prototype. Using a dataset of hotel reservations from Travelocity.com of transactions conducted over a three-month period for approximately 1500 hotels in the United States, they:

• Quantified the economic influence of multiple location characteristics and internal and external amenities on hotel demand, including how: ◦ A nearby beach increases demand by 18 percent ◦ Easy access to highways increases demand by 7.9 percent ◦ A one-star increase in hotel class raises demand by 4 percent

• Using 7,800 unique user responses for comparing different rankings, demonstrated that their new ranking system outperforms existing travel search engines such as Expedia, Orbitz and Travelocity, which are based on “number of reviews,” “price,” or “average star ratings”

• Assembled a personalized set of results for users who want to see rankings for “all people like me” based on demographics or trip purpose, reducing search and cognitive costs

“We found that many customers prefer a list of hotels, each specializing in a variety of characteristics, rather than a variety of hotels, each specializing in only one characteristic,” added Ghose. “We now have the capability to crawl, sift, parse and integrate online user-generated information automatically into a ranking system that factors in multiple consumer preferences. This translates into real value for consumers in terms of both time and money saved.”

Prof. Ghose, an expert in the economic impact of user-generated content, unveils their system in a working paper, “Designing Ranking Systems for Hotels on Travel Search Engines by Mining User-Generated and Crowd-Sourced Content,” which received the Best Paper Award at this year’s World Wide Web conference. The paper is available at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1856558, and a demo of their ranking system is available for viewing upon request.

Jessica Neville
416-516-7677