Our work for the NTCIR-11 Temporalia challenge, Temporal Query Intent Classification sub-task: predicting the temporal orientation of search engine user queries.
We tackled the task as a machine learning classification problem, by proposing th use of temporal-oriented attributes specifically designed to minimise the sparsity of the models.
The best submitted run achieved 66.33% of accuracy, by correctly predicting the temporal orientation of 199 test instances out of 300.
##Requirements
Python libraries:
- NLTK (web page)
Non-Python resources:
please update the feature_extractor.py file with the right paths.
##Contact