Shuji Yamaguchi: Enhancing Reuters AlertNet
Shuji Yamaguchi came to the Digital Vision Program with an innovative plan to lower the language barriers dividing humanitarian workers around the globe. He soon changed focus, however, as he discovered that a lack of access to news reports was a far more pressing need for his constituents. But rather than derail his vision, this realization actually served to expand his notion of how he might help aid workers overcome linguistic hurdles in the future.
Shuji’s initial proposal called for the creation of an online dictionary in support of AlertNet, a Reuters Foundation news service that tracks evolving crises around the world. The dictionary would be maintained by its users, ensuring that it would keep pace with the ever-changing lexicon of the humanitarian community. Another technology called Cross-Language Information Retrieval would allow users to search foreign-language sources by way of keywords in their native tongues. With this tool, he believed, AlertNet would be vastly more informative, enhancing its users’ ability to identify and assist communities in need.
However, after consulting with an AlertNet editor early in his fellowship, Shuji turned his attention to designing a program that would automatically categorize news stories by subject and identify the place names within them. With this system in place, reports from different sources could be grouped together easily, leading to a more reliable and complete picture of developing situations.
Shuji understood that pursuing this project would mean straying from his original vision, but he maintained that doing so was beneficial both to AlertNet and to himself. “Overall,” he said, “it was very effective for my project to have a concrete target like AlertNet. With it I was able to have delivered something visible within nine months.”
In addition, Shuji said, by working closely with AlertNet representatives, he was able to continue working on the development of his primary idea. “My work on cross-lingual search has attracted attention within the Reuters news group as a potential tool to extend its automatic news categorization beyond the current target language of English. I am working on this now.”
Based on his work as a fellow, Shuji looks forward to addressing the problems of linguistic intelligibility throughout the digital realm. “With ‘keyboard and display’ becoming ‘pen and paper’ in the modern world, native language support on computers and software has the potential for a far-reaching impact to develop domestic economies from the bottom up. I would like to contribute in some way or another to promote a collaborative approach to address the language support issue of hardware and software, beyond my original scope of information search by native language.”
—Mike Nowak