I have written newsletters, web stories, and feature articles for a wide range of clients, including Pitney Bowes, The University of Rochester Medical Center, The College at Brockport, The Harley School, and more. Included here are articles I wrote for Cornell University and the University of Rochester Medical Center Department of Orthopaedics.
The Art of the Search: from Salton to Singhal, Cornell has revolutionized Web retrieval.
You just heard a song on the popular TV series, “Breaking Bad”, but you can’t remember the name of the band. All you know are three words from the lyrics. You type “my baby blue” into Google Search, and a few seconds later, the song comes up with a link to YouTube and a bio of the band, Badfinger. This kind of “tip of your tongue” search, where you have only vague or partial information, would have been unheard of ten years ago. But thanks to Cornell’s revolutionary work in information retrieval (IR), web searches have become as fast and intuitive as thinking. And, as Cornell alumnus Amit Singhal, ’97, says, “We are just getting started.”
The Director of Google Search since 2000, Dr. Singhal is credited with Google innovations such as Autocomplete (predicting your keyword when you type in the first letters); Universal Search (comprehensive entries from other search engines); Voice Search; Translation; and Google Search for mobile devices. By creating search algorithms that conform to the way humans naturally speak and think, Dr. Singhal is continuing the legacy of his Cornell mentor, the late Professor Gerard Salton.
The “father of modern digital search” and co-founder of Computer Science at Cornell, Professor Salton led the team that developed Cornell’s SMART Information Retrieval System in the early 1960s. Before SMART, information retrieval was based on the Boolean algorithm, where the query had to exactly match the words of the document searched. The Boolean system did not rank documents according to their relevance. And users had to type in search strings using the terms AND, NOT, and OR. So if you were searching for the Badfinger song, you would have to input “baby AND blue”, and then cull through all the documents, listed arbitrarily, whether they were relevant to your search or not. Considering that there are tens of thousands of songs with the words “baby” and “blue” in them, Boolean logic would present the proverbial haystack to anyone searching for their favorite song.
Professor Salton’s SMART system changed the game. Based on a vector-space model, it retrieved documents based on their similarity to the query, and ranked them by best or partial matching. The Cornell team also developed concepts such as term-weighting, which evaluates text relevance, dictionary construction, automatic text processing, and the “relevance feedback” loop which paved the way for site ranking based on search popularity. Internet navigational tools that are second nature to us now-- such as keyword searching and clicking on hypertext links---—are also largely credited to Professor Salton and the Cornell team.
The concepts established at Cornell decades ago paved the way for how we search today—and for even more intuitive searching in the future. At Google, Dr. Singhal and his team are continuing Dr. Salton’s legacy, perhaps in ways he would have never dreamed of. Today, both Google and Cornell are exploring the next generation of Web searching, which focuses not just on matching words, but on replicating the way human minds make connections and associations.
Google’s search architecture of the future will replicate three human activities: answering questions, having conversations, and anticipating needs. Knowledge Graph is a database of the 500-million most-searched-for people and topics in the Google universe. It enables the search engine to make associations, and understand questions using conversational language. Google is also designing algorithms that can connect associations between the search engine and the user’s own world. So for example, when you make an airline reservation, Google can keep you updated on your flight via Gmail. Another feature would be having Google anticipate your questions by making associations between your search queries and interests. Dr. Singhal says, “At Google, we want to help you make the most of your day. We believe that you shouldn’t be spending your time searching, but living your life.”
Dr. Singhal remains an active part of the Cornell Family. His wife is also a Cornelian, and his first child will be graduating Cornell, class of 2018. The concepts he first learned at Cornell, that technology should be collaborative, creative, and designed to serve humanity, have clearly been carried on in his work at Google.
University of Rochester Medical Center
The power of "pixie dust" : an old wartime practice finds a new role in fighting infection.
During the Civil War, doctors sprinkled powdered sulfonamides directly into surgical wounds to assist in sterilization. More than a century later, a young army doctor named Robert Molinari also used this technique while stationed at Kosovo. Today, at the University of Rochester Medical Center, he is again on the frontlines, using this "old trick" in the ongoing battle against surgical infection.
For patients who are particularly vulnerable to infection, Dr. Molinari boosts their standard antibiotic regimen by sprinkling powdered Vancomycin directly into the surgical site before closing the wound. Bombarding the site with "pixie dust", as powdered Vancomycin is nicknamed, dramatically reduces the chances of infection. If this sounds "low-tech" or even a little crude, it is a common practice among surgeons today. However, Dr. Molinari is one of the few to have published results about its efficacy in professional journals.
At the American Academy of Orthopaedic Surgeons annual meeting this year, he presented findings that are promising, though not conclusive. In 1,512 of his cases that used powdered Vancomycin, the infection rate was only 0.99 percent, compared to a national average of approximately 2.1 percent among 108,419 patients in the national database of the Scoliosis Research Society. “Our infection rate is not zero percent, which would be ideal,” Dr. Molinari says. “But no matter how you look at it, our rates are below the national average, and many surgeons are adopting this technique.” Dr. Molinari says that he would like to have his findings confirmed in a large, randomized, multi-site study.
Vancomycin is one of the most powerful antibiotics in treating deadly infections such as Methicillin-resistant Staph aureus, or MRSA. Since strains of staph are notorious for colonizing on foreign objects inside the body, patients with orthopaedic implants are particularly vulnerable. A sprinkle of Vancomycin, in addition to standard antibiotics, can be an effective deterrent against infection. It is also one of the cheapest, costing about $4.00 a gram. Dr. Molinari says, “Our data shows it works, but our findings have to be further validated by a case-controlled study." Clearly, for his patients, a sprinkle of prevention is worth a pound of cure.