Add datasets

nlmapsweb/config/default.py

@@ -58,3 +58,22 @@ REPLACE_DUPLICATE_NAMES = True
 REPLACE_DUPLICATE_NAMES_THRESHOLD = 3
 
 WTF_CSRF_ENABLED = False
+
+# Example for downloadable datasets. name, url_path, license and license_url
+# are required fields.
+DATASETS = [
+    {
+        'name': 'NLMaps Web',
+        'url_path': '/static/datasets/nlmapsweb.zip',
+        'license': 'CC BY-SA',
+        'license_url': 'CC BY-SA',
+        'original_url': 'https://www.cl.uni-heidelberg.de/statnlpgroup/nlmaps/',
+        'comment': 'Collected by different annotators'
+    },
+    {
+        'name': 'NLMaps Web 2',
+        'url_path': '/static/datasets/nlmapsweb.zip',
+        'license': 'CC BY-SA 4.0',
+        'license_url': 'https://creativecommons.org/licenses/by-sa/4.0/',
+    }
+]

nlmapsweb/config/production.py

@@ -23,3 +23,50 @@ MODELS = [
 ]
 CURRENT_MODEL = 'n3_arc_ratio05_prod.yaml'
 FALLBACK_MODEL = 'n3.yaml'
+
+DATASETS = [
+    {
+        'name': 'NLMaps 2.1',
+        'url_path': '/static/datasets/nlmaps_v2.1.zip',
+        'license': 'CC BY-NC-SA 4.0',
+        'license_url':  'https://creativecommons.org/licenses/by-nc-sa/4.0/',
+        'original_url': 'https://www.cl.uni-heidelberg.de/statnlpgroup/nlmaps/',
+        'comment': 'Modification of NLMaps v2, published by Carolin Lawrence and Stefan Riezler.',
+    },
+    {
+        'name': 'NLMaps 3a',
+        'url_path': '/static/datasets/nlmaps_v3a.zip',
+        'license': 'CC BY-SA 4.0',
+        'license_url': 'https://creativecommons.org/licenses/by-sa/4.0/',
+    },
+    {
+        'name': 'NLMaps 3b',
+        'url_path': '/static/datasets/nlmaps_v3b.zip',
+        'license': 'CC BY-SA 4.0',
+        'license_url': 'https://creativecommons.org/licenses/by-sa/4.0/',
+    },
+    {
+        'name': 'NLMaps 3',
+        'url_path': '/static/datasets/nlmaps_v3.zip',
+        'license': 'CC BY-NC-SA 4.0',
+        'license_url':  'https://creativecommons.org/licenses/by-nc-sa/4.0/',
+    },
+    {
+        'name': 'NLMaps 3 (no noise)',
+        'url_path': '/static/datasets/nlmaps_v3_no_noise.zip',
+        'license': 'CC BY-NC-SA 4.0',
+        'license_url':  'https://creativecommons.org/licenses/by-nc-sa/4.0/',
+    },
+    {
+        'name': 'NLMaps 4 (raw)',
+        'url_path': '/static/datasets/nlmaps_v4_raw.zip',
+        'license': 'CC BY-SA 4.0',
+        'license_url':  'https://creativecommons.org/licenses/by-sa/4.0/',
+    },
+    {
+        'name': 'NLMaps 4',
+        'url_path': '/static/datasets/nlmaps_v4.zip',
+        'license': 'CC BY-SA 4.0',
+        'license_url':  'https://creativecommons.org/licenses/by-sa/4.0/',
+    },
+]

nlmapsweb/static/css/datasets.css

+.dataset-comment {
+    font-style: italic;
+}

nlmapsweb/templates/datasets.html

+{% extends 'base.html' %}
+
+{% set page = 'datasets' %}
+
+{% block content %}
+<div class="row block">
+    <div class="block-header">
+        <span class="block-title">Datasets</span>
+    </div>
+    <div class="block-body">
+        {% if datasets %}
+        <p>These are the datasets that were published together with the Master’s thesis on “An Online Learning System for Parsing and Answering Geographical Queries in Natural Language”, which is available <a href="https://gitlab.cl.uni-heidelberg.de/will/nlmaps-ma/-/raw/thesis/thesis/thesis.pdf">here</a>. The thesis also explains how they were created.</p>
+        <div class="container-vertical-flex">
+            {% for dataset in datasets %}
+            <div class="dataset container-vertical-flex">
+              <strong class="dataset-name"><a href="{{ dataset.url_path }}">{{ dataset.name }}</a></strong>
+                <span class="dataset-license">License: <a href="{{ dataset.license_url }}">{{ dataset.license }}</a></span>
+                {% if dataset.original_url %}
+                <span><a href="{{ dataset.original_url }}">Original dataset</a></span>
+                {% endif %}
+                {% if dataset.comment %}
+                <p class="dataset-comment">{{ dataset.comment }}</p>
+                {% endif %}
+            </div>
+            {% endfor %}
+        </div>
+        {% else %}
+        <p>No datasets available.</p>
+        {% endif %}
+    </div>
+</div>
+{% endblock %}

nlmapsweb/views/__init__.py

@@ -4,6 +4,7 @@ from .errors import (page_forbidden, page_not_found, internal_server_error,
                      mt_server_error)
 from .feedback import (create_feedback, delete_feedback_piece, export_feedback,
                        feedback_piece, list_feedback, update_parse_taggings)
+from .info import datasets
 from .legal import legal_notice
 from .login import login, logout
 from .parse_logs import parse_logs

nlmapsweb/views/info.py

+from flask import current_app, render_template
+
+
+@current_app.route('/datasets')
+def datasets():
+    available_datasets = current_app.config.get('DATASETS', [])
+    return render_template('datasets.html', datasets=available_datasets)