Bringing Semantics to the Web

A key factor in the success of the Web today is its simplicity and ease of use. This simplicity allows just about anyone to write and publish an HTML page and immediately see the results of their efforts. The next obvious step in the evolution of the web is to bring in semantics to describe the contents. This would allow not just sharing of information, but also support improved searching, extraction, filtering, monitoring, and, most importantly, integration. This has clearly been a goal of the Semantic Web, and while the Semantic Web is a step in this direction and has achieved some success, it has not been embraced the larger Web community. I believe that the two key reasons that the Semantic Web has not taken the Web by storm are that, unlike the syntactic web, there is a lot more to learn in order to use it and the end users do not immediately see the benefit of their efforts. This is a challenge that needs to be addressed. There are two parts to this challenge. First, how do we make it easy to bring the needed semantics to the information and sources on the Web. Second, how do we provide an immediate benefit to the users such that they see a direct benefit for their efforts. As an example, consider the set of tasks that I undertook to book a hotel in London for the Web Science Workshop. First, I reviewed the recommended hotels and noticed that the cheapest workshop rate was 125 pounds. Given the weak dollar, this converts into over $225, which is more than double what I would typically spend on a hotel in the US. Next, I reviewed a few of the hotel aggregator web sites (e.g., Orbitz) and found that I could get a smaller room in the same hotel for less money, but not a lot less. Finally, I went to the BiddingForTravel.com web site, which lists winning bids for Priceline. (Priceline is a web site that provides steep discounts on hotel rooms and airline tickets, but you have to bid on these products.) After reading through dozens of posts on BiddingForTravel, I found that I could probably bid on a hotel on Priceline for less than $100. In the end I was able to book the Thistle Marble Arch, which is a 4* hotel about 3 miles from the workshop, on Priceline for $75 per night. However, despite knowing where to look, the research required to find this hotel took several hours of work. Let’s consider the time consuming parts of this task. First, I had to read through dozens of posts on BiddingForTravel to find the prices for hotels on similar dates in London. Second, since in Priceline you bid on an area and not on a hotel, I had to take the list of hotels organized by areas, and invoke GoogleMaps on each hotel to see where they were located relative to the workshop location. Third, I had to check the prices on the individual hotels for the required dates to be sure that I would not bid more than the hotels were charging. Finally, I had to combine all of this information together (largely in my head) to make a decision on what hotel to book or what to bid in Priceline. The first task is an information extraction problem, the second and third are tasks that could be automated given better models of the sources, and the last task could be greatly improved with better web integration tools. I will address each of these in turn.