GitHub - GentleZhu/ReMine: Integrating Local Context and Global Cohesiveness for Open Information Extraction(WSDM'19) (original) (raw)

ReMine: Integrating Local and Global Cohesiveness for Open Information Extraction

Source code and data for WSDM19' paper "Integrating Local and Global Cohesiveness for Open Information Extraction"

Dependencies

We run all experiments on Ubuntu 16.04.

• python 3.5
• Python library dependencies
• eigen 3.2.5 (already included).

Build

Test with pre-trained model

The result files can be found at results_remine/remine_results.txt

Re-train our model on NYT and twitter corpus(under polishing)

Phrase Extraction Module

$ bash phrase_extraction.sh

Example Segmented Corpus

(background_phrase) [entity_phrase] <relation_phrase>

(Gov. Tim Pawlenty of Minnesota) (the state health department) (this month) (to monitor) [day-to-day operation] the [Minneapolis Veterans Home] [state inspector] (three man) there (the previous month) (because of) [neglect] [medical error]

(The aid group Doctor) (Without Border) [Saturday], more than 275 (wounded people) [Donka Hospital] (in the capital of) [Guinea], (Conakry).

Integrated Optimizer