A student team from the Natural Language Processing Laboratory claimed the first- and third-place prizes in competitions at the ACM SIGIR 2014 Entity Recognition and Disambiguation Challenge in early July. Organized by Google and Microsoft, the competitions challenge teams from around the world to design entity recognition and disambiguation (ERD) systems by measuring the competitors’ abilities to rapidly recognize mentions of entities in a given text, disambiguate them to determine their real meanings, and link them to the entities in a massive knowledge base. Above all, the competition stands as the ultimate test of the speed and accuracy which a team designed ERD system may attain.
Led by Prof. Hsin-Hsi Chen of the Department of Computer Science and Information Engineering, the NTUNLP team won first place in the Short Track competition and took third place in the Long Track challenge. The Short Track competition challenged teams to analyze short texts, such as the search queries presented to Google and Microsoft's search engines, while the Long Track competition focused mainly on web documents, which forced the teams to deal with a wide diversity of writing styles.
The ERD Challenge took place entirely online. During the dry run stage of the competitions, teams were permitted to try out different methods of coding. In the formal run stage, however, the organizers sent information to each team's designated server for analysis. In the case where a team exceeded the time limit for processing a particular entity, that entity was considered unsuccessful, and the team was not given a second chance on that entity. In other words, competing teams were required to design ERD systems that were robust, fast, and accurate.
The NTU team's winning performance demonstrates the thoroughness and effectiveness of the training that students of the Department of Computer Science and Information Engineering and Graduate Institute of Networking and Multimedia receive in both theory and practical applications in massive data analysis, natural language processing, systems integration, and programming and design.
The Annual International ACM SIGIR Conference on Research and Development in Information Retrieval is among the most prestigious international conferences in computer science and information retrieval in the world. ACM SIGIR 2014, the organization's 37th annual conference, was held in Gold Coast, Australia from July 6-11.