Extracting temporal meta-information from knowledge Bases

This repository contains the first version of our work titled "Extracting temporal meta-information from knowledge Bases". Since this is a work in progress, several improvements will be added later. The goal here was only to test our approach at the expense of inefficiency of running times. The algorithm that implements the reasoning to obtain temporal information can be found in the inferTempMetafactsipynb file. We use jupyter notebook to facilitate the reproduction of the work. Files generated are stored in outputs folder.

To run the inferTempMetafacts.ipynb, you must:

• create a local base with the YAGO files (Download) - YAGO files used. In addtion, a database script to load YAGO into a Postgres database is also provided in the download link. In this work, we only use some YAGO (.tsv) themes to recreate a portion of the YAGO database. These files are:

- Taxonomy: yagoSchema, yagoTaxonomy, and yagoTypes.
- CORE: yagoFacts, yagoLiteralFacts, and yagoDateFacts.
- Meta: yagoMetaFacts -- temporal and geospatial meta facts of yagoFacts

• update the database.ini file with the information from your database.
• create the folder outputs.

To run the search_movies.ipynb, update the APIkeys.json file (the file can be found at lib folder) with your IMDb key.

• The meta-facts obtained can be found at the formatted folder (all files are formatted. We've compressed the files because of the size.). Files generated before being formatted are stored in "outputs".
• The file that combines the contents of the actedIn.csv file with the IMDb information can be found at imdb folder.

• Tirthajyoti Sarkar - Step guide to search IMDb information.