STAN

The web is full of high quality data in the form of tables that could be integrated to generate new knowledge. However web tables are often lacking in semantics. A crucial problem is the correct relationships detection between a table and an ontology in order to define the semantics. In this context STAN is a tool for semi-automatic semantic annotation of tables that lets non technicians carry out the annotation process.

System Requirements

Linux OS
Git
Java
Eclipse (download at https://www.eclipse.org/downloads)
50GB available disk space
4GB memory required (8GB advised)

Check out the repository and configure your local machine

#!bash
$ git clone https://github.com/brando91/STAN.git
$ cd STAN
$ ./build-and-test.sh

Create the required index

#!bash
$ ./stan/scripts/build-index.sh dbpedia

This will take around 7 hours to complete.

Run STAN

#!bash
$ ./launch-stan.sh
Open a browser at http://localhost:8081

stan's People

Contributors

Stargazers

Watchers

stan's Issues

Logs refinements

Labelling algorithm knows property types

During the indicization the types of dbpedia properties (Datatype, ObjectType..) are lost. We need to store them.

Entire table annotation API

API to annotate an entire table

Separate STAN experiments in a new codebase

repo cleanup and better organization

lanciafiamme senza pietà su tutto il codice non usato

About page

Create a page with some info about STAN.

The route will be http://stan.disco.unimib.it/about

Separate algorithm experiments in a new codebase

Snow flake export

Composed star export

Hidden subject export

Property exploration in ABSTAT

Create a link from each suggested property to ABSTAT to help users better understand the usage.

document and validate the installation requirements

add a validation step in the travis build?

Previous tables resume

Now it is possible to resume the last table uploaded. Maybe it is interesting to give the possibility to resume one of the last n tables uploaded.

Upload preview

Preview of the table to upload modifying the delimeter character

Fix logging configuration

it.disco.unimib.labeller.benchmark.HttpConnector uses libraries from HttpClient 4.5 and when HttpConnectorTest runs the debug logs are not handled.
Even reading this https://hc.apache.org/httpcomponents-client-4.5.x/logging.html i still can't figure out how to handle it.

Property type filtering

When selecting a property the object type options should be filtered in order to match the type of the property selected (Datatype, ObjectType, ...)

SubProperty annotation

Give the possibility to specify a custom property as subproperty of an existing one

"How to contribute" wiki page

We need a page (the repo wiki?) that explains clearly how to contribute to the repo. For example, do we require unit test for implemented features? Do we expect pull requests from the students? etc..