Skip to content
Benchmark Generator for Knowledge Extraction
Java
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
data
src
.gitignore
LICENSE
README.md
pom.xml

README.md

BENGAL

Automatic Generation of Benchmarks for Entity Recognition and Linking

This project aims at delivering a framework for generating bechmarking data sets from structure data to the Named Entity Recognition (NER) and Disambiguation (NED a.k.a Entity Linking and Wikification) problems.

We hope you will enjoy using BENGAL!

Support and Feedback

If you need help or you have questions do not hesitate to use the issue tracker.

Setting up BENGAL

  1. Checkout BENGAL from github.
  2. Download the surface forms file which is available at https://hobbitdata.informatik.uni-leipzig.de/bengal/en_surface_forms.tsv.zip.
  3. Download the WordNet dictionary which is available at http://wordnetcode.princeton.edu/3.0/WordNet-3.0.tar.gz.
  4. After extracting the WordNet files, move them into the BENGAL/data directory or provide the path using cli. Note that only the dict folder of WordNet is necessary.
  5. Run mvn clean install in the Bengal directory
  6. to run the program from cli use the jar file with the postfix exec.

Executing BENGAL

Example Usage:

java -cp ./target/BENGAL-1.0-SNAPSHOT-exec.jar org.aksw.simba.bengal.controller.BengalController -st sym -dp /home/WordNet-3.0/dict

CLI Options:

 -dp,--dictpath <arg>           Path to Dictionary's directory, default:
                                data/dict
 -mns,--minsentence <arg>       Minimum number of sentences, default: 3
 -mxs,--maxsentence <arg>       Maximum number of sentences, default: 10
 -n,--numberofdocuments <arg>   Number of documents, default: 1
 -o,--onlyobjectprops           Use only object properties
 -pp,--paraphrase               Use Paraphrasing
 -pr,--pronouns                 Use Pronouns
 -sd,--seed <arg>               Number of Seeds, default: Current System
                                time in milliseconds
 -se,--sparqlendpoint <arg>     Sparql Endpoint, default:
                                http://dbpedia.org/sparql
 -sf,--surfaceforms             Use Surface-Forms
 -sp,--surfaceformpath <arg>    Path to Surface Forms file, default:
                                data/en_surface_forms.tsv
 -st,--selectortype <arg>       Selector Type ('star', 'hybrid', 'path',
                                'sym' or 'summary')
 -wt,--waittime <arg>           Wait time between documents in
                                milliseconds, default: 500

Portuguese Version

The Portuguese version can be found at click here.

You can’t perform that action at this time.