# Cluster Search Readme #


Cluster Search is running on Cliopatria powered by Prolog. Below the instructions for setting up cluster search on a mac or linux machine.

## Setup webserver ##


Install latest SWI-Prolog development release: http://www.swi-prolog.org/download/devel

Clone the Cliopatria webserver code in a suitable directory:
```bash
$ cd git
$ git clone git@github.com:ClioPatria/ClioPatria.git
```

Create a folder for the clustersearch project:
```bash
$ cd ..
$ mkdir clustersearch
$ cd clustersearch
```

Now configure this directory as a ClioPatria project by running:
```bash
$ ../git/ClioPatria/configure
```

You can now run the web server using the command below (typing 'halt.' will stop it):
```bash
$ ./run.pl
```
   
## Setup cluster_search package ##

Start the webserver and install the required cpacks:
```bash
$ ./run.pl
?- cpack_install(cluster_search).
```

Congratulations! If everything went well you now have a server with cluster_search functionallity running at: localhost:3020/clustersearch

## Load data ##

Create a folder for the rdf data:
```bash
$ cd ..
$ ls
  ClioPatria  clustersearch
$ mkdir rdf
$ cd rdf
```

Download the data and move it to the folder, here we download the iconclass vocabulary:
```bash
$ mkdir iconclass
$ cd iconclass
$ wget http://iconclass.org/data/iconclass.20121019.nt.gz
```
If you have access to the eculture server, this is how to clone the rijksmuseum collection data (replacing USER with the appropriate username):
```bash
$ cd ..
$ git clone USER@eculture.cs.vu.nl:/home/git/commit/rma.git rijksmuseum
```

Loading the data after starting the server:
```bash
$ cd ..
$ ./run.pl
?- rdf_load('../rdf/iconclass/iconclass.20121019.nt.gz', [graph(iconclass)]).
```
For the Rijksmuseum daata there are multiple subsets to choose from:
```bash
?- rdf_library:rdf_attach_library('../rdf/rijksmuseum/rdf/void.ttl').
?- rdf_library:rdf_list_library.
```
A list of datasets to load is shown. You can for example choose to load the complete collection: 
```bash
?- rdf_library:rdf_load_library('RMA-collection',[]).
```
With the data loaded you should be able to cluster some artworks!