Temponym Resolution

In this project we address the problem of detecting temponyms such as "Clinton's term as secretary of state, the late-2000s financial crisis", inferring their temporal scopes, and mapping them to events in a knowledge base if present there.

We present methods for this kind of temponym resolution, using an entity- and tempex-oriented document model and the Yago knowledge base for distant supervision. We develop a family of Integer  linear Programs for jointly inferring temponym mappings to the timeline and knowledge base. This  enriches the document representation and also extends the knowledge base by obtaining new alias names for events.

For scientific works about temponyms cite this paper.

Temponym Tagger

In this section we provide HeidelTime temporal tagger with temponym tagging functionality.

HeidelTime is on GitHub:

github.com/HeidelTime/heideltime/

Downloads can be found here:

github.com/HeidelTime/heideltime/releases

 

If you use HeidelTime for temponym tagging, please cite this paper.

Extracting Named Events from News

EVIN (EVents In News) is a system that can extract named events from a news corpus, organizes them into ontological classes, and supports interactive exploration. EVIN exploits different kinds of similarities between news items referring to textual contents, entity occurrences, and temporal ordering, and captures these similarities in a multi-view attributed graph. To distill canonicalized events that can serve to populate a clean ontology, EVIN coarsens the graph by iterative merging based on a judiciously designed loss function. In this demo, we show how EVIN can extract named events on-the-fly based on a user's specific interests. EVIN provides a GUI that allows users to query the system, to browse the extracted events along a timeline visualization, and to explore details about events and the associated news.

For scientific works about EVIN, cite this paper.

Experimental Data

The experimental data can be downloaded here.

The experimental data including the large news corpus used for knowledge base population can be downloaded here (1.4 GB).

If you have any question regarding the data please contact Erdal Kuzey.

Web Demo

EVIN web demo presents the grouping and chaining capabilities of EVIN.

The demo uses a small corpus to show the proof of the concept. Therefore, please check the examples provided in the tutorial. For usage, you can check the video demo.

Watch the video demo showing the functionality of EVIN. The demo paper is here.

Further Information

EVIN is part of the YAGO-NAGA project at the Max Planck Institute for Informatics in Saarbrücken/Germany. It is developed by the Databases and Information Systems Group.

The main people behind these projects are Erdal Kuzey and Gerhard Weikum.

 

For questions and comments, please contact Erdal Kuzey.