2nd NLP TOOLS CONTEST on Indian Language Dependency Parsing
The ICON09 tools contest on Indian language (IL) dependency parsing was enthusiastically received among researchers. Many varied parsers were implemented, and baseline performance was set for Hindi, Telugu and Bangla. Crucial parsing issues, many IL specific, came to light and were discussed.
However, efficient Indian Language (IL) parsing still remains a challenging task. This contest aims to bring together researchers working/interested in the area of IL parsing to explore techniques that can further improve the present accuracies, by providing larger data sets for Hindi and more revised data for Telugu and Bangla.
Participants will be provided training, development and testing data to report the efficiency of their dependency parsers. Hindi, Bangla, and Telugu will be explored. Parser efficiency will be measured in terms of standard measures such as Unlabelled Attachment Accuracy and Labelled Attachment Accuracy.
Shortlisted candidates will present their techniques and results in a special session at ICON10. Note that it is not necessary that the participating parser be a statistical one - other types, such as grammar driven parsers or hybrid parsers can also participate.