Temporalia: Temporal Query Intent Classification (TQIC)

This project is part of the NTCIR-11 Temporalia challenge, Temporal Query Intent Classification: predicting the temporal orientation of search engine user queries: present, past, future and atemporal. We propose to tackle the task as a machine learning classification problem.

For example, the temporal orientation of a query like weather in Manchester is present, whereas for weather forecast Manchester it's future. Some queries refer to the past (e.g. when did galileo born?), whereas some other don't have a temporal orientation (e.g. sunday times, fairchild dancer lyrics).

Download

Source code

The source code is hosted by GitHub.

Machine learning playroom

The pipeline uses a CRFs model for the identification phase. You can use one of the following pre-trained models, depending on the training set:

Training and test set
(ARFF format)

The first 100 instances are the training set, the rest (300) is the official benchmark test set.
Size: 96.498Kb; MD5: ad0e42312b7b3be6e3c638d1cc45f432, download

Training and test set
(CSV format)

The first 100 instances are the training set, the rest (300) is the official benchmark test set.
Size: 78.127Kb; MD5: 6721a82d8d87f7b86c05569ed881adc9, download

Related publications

Using machine learning to predict temporal orientation of search engines' queries in the Temporalia challenge
M. Filannino, G. Nenadic
Proceedings of the Sixth International Workshop on Evaluating Information Access (EVIA 2014) a Satellite Workshop of the NTCIR-11 Conference

paper, poster, slides, review, demo, source code.