BYBLOS—A visitor wouldn’t know it from his humble office, but Joe Tekli, a computer engineer at the Lebanese American University’s campus in Byblos, has lofty aims for his research career. He wants to tidy up the entire online world, making it a more logical and efficient place.
“In recent years there has been a violent expansion of the web without any real thought about how it’s all going to work,” says Tekli.
The World Wide Web is a very messy place, he says, with ever larger databases springing up almost overnight and a practically endless stream of photos, graphics, videos and documents uploaded every day, with many linked to each other. This rapidly expanding number of files and databases need to get better at talking to each other, Tekli says; otherwise the web risks becoming entangled in its own complexity, making it harder for users to navigate and search.
He wants the web’s infrastructure to be better at categorizing and describing its data in more detail, and he’s building software to try and make that happen.
“We might have billions and billions of terabytes of data, but if we can’t actually search for what we want then that data is useless,” Tekli says. “We want to search quickly and effectively.”
Right now, he says, no hard and fast rules govern how to store data online. Few standards categorize what is uploaded to the Internet. As a result, users aren’t getting the most out of the Internet, he says. Search engines built by companies like Google or Microsoft can have difficulty discriminating between potential search results and grasping the different connotations and denotations of words.
These are glitches Tekli believes could be resolved with universal best practices.
Internet users might not immediately notice the problem, because they eventually get the search results they want. But Tekli’s point is that protocols need to change so that tomorrow’s web network can keep pace with the large volume of data headed its way. He hopes researchers like him can at least maintain order on the web, and hopefully improve it.
He has no high-tech or particularly impressive pieces of hardware in his office to help with that ambitious task. His desk houses a laptop computer, a pile of textbooks and some action figures that he likes to assemble in his spare time, a hobby he picked up while doing a postdoctoral fellowship in Japan.
“I don’t require large clusters of computers. I just need a good server and Internet connection,” he says.
After acquiring a master’s degree at Antoine University in his native Lebanon, Tekli went on to get another master’s degree and a Ph.D. at the Université de Bourgogne in France before a series of postdoctoral research projects in Italy, Japan and Brazil. He returned to Lebanon in 2011 with the desire to unclutter the web.
Richard Chbeir, a professor at the department of computer science at the Université de Pau in France, was Tekli’s doctoral supervisor. He says it’s no surprise that Tekli wants to reorganize the web.
“He is like this in everything; he’s a tidy person,” says Chbeir. “He has an ideology; it’s not superficial. He wants to make the digital world better.”
Since Tekli left Chbeir’s lab, the two have kept in touch and collaborate regularly.
“I’ve had excellent students in the past but you don’t find them like Joe very often. He’s one of the best,” says Chbeir.
One of the ways Tekli is trying to make Internet search results less jumbled is with software that would search through all the news services online for their coverage of a breaking story, like an election result. The software would present the results in what he says would be a more meaningful way than current news services do.
“Many news corporations publish similar content in different structures or wordings, but they more or less agree with each other. Then there will be a couple of paragraphs of their own unique analysis,” he says.
The software he is developing will automatically analyze the news, figure out which portions of reporting on a development everyone agrees on, present that in a coherent way and then add the information that is unique to the various outlets.
“His objective is to restructure the web to allow users to get what they want faster and more efficiently,” explains Chbeir.
Tekli is also doing what is called “sentiment-analysis” research, in which he uses software to understand Internet users’ moods based on their online activity and what they are posting on social media. That data would then be used to ferret out search results, online activities and even music relevant to the user’s mood.
For a computer to suggest a sad piece of music, it first needs to learn what it sounds like. For that, the computer needs data, which is why Tekli is trying to make Web databases more sophisticated.
When Tekli has finished developing one of his software projects, he puts it online in a format ready for anyone to download and use. The hope is that tech companies or other researchers might take his work and either implement it or improve it.
While this approach may make it more likely for his work to be adopted, it also means he’s unlikely to profit from it. “I’m a scientist, not a businessman,” he says.
On the issue of turning his coding into reality, Tekli is limited by staying in Lebanon, says Chbeir.
Much of his collaboration will need to be remote and not in person. And “he doesn’t have the funding opportunities that he would at an international university and he could surely work for the biggest and best universities out there,” says Chbeir.
But being in Lebanon is part of what Chbeir calls Tekli’s “ideology.”
“He wants to be close to his origins, and we need people like him to show that the region is capable of research like this.”