oeg-upm / tada-hdt-entity

Add classes to entity columns in tabular data using HDT as the knowledge source

Annotate entity and textual columns using HDT

TADA HDT Entity

This library offers several techniques to annotate entity and textual properties. It user HDT-compressed knowledge base as the training set to annotated a given table.

requirements

• easy-logger
• tabular-parser
• hdt-cpp
• getopt

Compile

We offer two options to compile:

Make

make

CMake

mkdir build
cd build
cmake ..
make

Tests

inside the build directory

./tests

Known Issues

1. If all the tests fail, it could be related to the index file. Updating the file test.ttl would results in an error when loading the test.hdt. To resolved this just delete the index file and run the tests again. An updated index file would then be generated and the tests should pass.

Technical documentation

The technical documentation can be found on https://oeg-upm.github.io/tada-hdt-entity

tada-hdt-entity's Issues

Add distdir to makefile for installation

So make install would use the DISTDIR

segmentation fault

a single function to combine both techniques for labelling property columns

A single function that when called would try different techniques progressively (e.g., if a technique returns results then no need to fetch more predictions)

Check if the class destructor also free up the space

cache for single session

To eliminate duplicate queries.

annotate textual properties for the heuristic technique

sample option

Add an option to use a sample of the data.

new semaphore settings with docker

Support sample option for property annotation

Currently, the sample option only works for annotating entity columns, but not for property columns.

sample for property permissive performance

It takes a lot of time even with sample option as all subjects will be taken into account.