A resource for bridging words and meaning

Linking raw text to structured meaning representations — and vice versa — is a subtask that comes up again and again when dealing with natural language. Valentin Spitkovsky and Angel Chang, who interned for Research at Google last summer, are releasing a resource that they built for this task, based on more than three billion anchor texts associated with web links, into Wikipedia's organization of knowledge. They will present their data set next week at the 8th Biennual International Conference on Language Resources and Evaluation (LREC 2012), in Istanbul, Turkey (http://www.lrec-conf.org/lrec2012/).
Shared publiclyView activity