QPX is designed to help airlines and travel distributors offer travelers exactly what they want – user-friendly shopping options, increased transparency, improved self-service offerings, and a broader and more accurate array of results.

QPX loads industry information about fares, rules, schedules, and seat availability and presents an XML interface to the outside world. It can evaluate literally billions of flight and fare combinations in order to arrive at a comprehensive set of recommendations for a given travel query. Along the way, QPX employs novel graph algorithms, search techniques, optimization strategies, heuristics, and a tremendous amount of domain-specific knowledge to determine which combinations are valid and available.

We work hard to make QPX flexible and extensible, by moving as much as possible out of source code and into configuration languages. And we always try to do things the right way, even when the domain and the data are messy.

If you want to learn more about the computer science behind ITA and QPX:

This map plots the arcs in the 10,000 quickest paths from San Francisco to Boston, all still arriving on the same day. For each one of these paths there are approximately 10,000 ways to price the itinerary, depending on the query. Now imagine performing this same search, combining pricing, availability, and user preferences, with the goal of returning the top 10 results that best match each customer’s unique travel objectives (e.g., lowest fare, flexible dates, one passenger vs. multiple, etc…). This merely hints at the complexity of returning the “best solution” or “lowest fare option” for any given search.