Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
|
@@ -30,22 +30,6 @@ STANDARD_COLORS = {
|
|
| 30 |
'WORK OF ART': '#DDA0DD' # Plum
|
| 31 |
}
|
| 32 |
|
| 33 |
-
# Entity definitions for glossary
|
| 34 |
-
ENTITY_DEFINITIONS = {
|
| 35 |
-
'DATE': 'Absolute or relative dates or periods',
|
| 36 |
-
'EVENT': 'Named hurricanes, battles, wars, sports events, etc.',
|
| 37 |
-
'FAC': 'Facilities - Buildings, airports, highways, bridges, etc.',
|
| 38 |
-
'GPE': 'Geopolitical entities - Countries, cities, states',
|
| 39 |
-
'LANG': 'Any named language',
|
| 40 |
-
'LOC': 'Non-GPE locations - Mountain ranges, bodies of water',
|
| 41 |
-
'MISC': 'Miscellaneous entities - Things that don\'t fit elsewhere',
|
| 42 |
-
'NORP': 'Nationalities or religious or political groups',
|
| 43 |
-
'ORG': 'Organizations - Companies, agencies, institutions, etc.',
|
| 44 |
-
'PER': 'People, including fictional characters',
|
| 45 |
-
'PRODUCT': 'Objects, vehicles, foods, etc. (Not services)',
|
| 46 |
-
'Work of Art': 'Titles of books, songs, movies, paintings, etc.'
|
| 47 |
-
}
|
| 48 |
-
|
| 49 |
# Additional colours for custom entities
|
| 50 |
CUSTOM_COLOR_PALETTE = [
|
| 51 |
'#FF9F43', '#10AC84', '#EE5A24', '#0FBC89', '#5F27CD',
|
|
@@ -746,40 +730,15 @@ def process_text(text, standard_entities, custom_entities_str, confidence_thresh
|
|
| 746 |
|
| 747 |
return summary, legend_html + highlighted_html, results_html
|
| 748 |
|
| 749 |
-
# Create colored checkbox labels
|
| 750 |
-
def create_colored_checkbox_labels():
|
| 751 |
-
"""Create checkbox labels with color indicators"""
|
| 752 |
-
labels = []
|
| 753 |
-
for entity in STANDARD_ENTITIES:
|
| 754 |
-
colour_key = "WORK OF ART" if entity == "Work of Art" else entity.upper()
|
| 755 |
-
colour = STANDARD_COLORS.get(colour_key, '#CCCCCC')
|
| 756 |
-
# Create label with color dot
|
| 757 |
-
labels.append(f"β {entity}")
|
| 758 |
-
return labels
|
| 759 |
-
|
| 760 |
# Create Gradio interface
|
| 761 |
def create_interface():
|
| 762 |
-
|
| 763 |
-
custom_css = """
|
| 764 |
-
/* Color the checkbox labels based on entity type */
|
| 765 |
-
"""
|
| 766 |
-
for i, entity in enumerate(STANDARD_ENTITIES):
|
| 767 |
-
colour_key = "WORK OF ART" if entity == "Work of Art" else entity.upper()
|
| 768 |
-
colour = STANDARD_COLORS.get(colour_key, '#CCCCCC')
|
| 769 |
-
custom_css += f"""
|
| 770 |
-
label[for="checkbox-{i}"] span:first-child {{
|
| 771 |
-
color: {colour} !important;
|
| 772 |
-
font-weight: bold;
|
| 773 |
-
}}
|
| 774 |
-
"""
|
| 775 |
-
|
| 776 |
-
with gr.Blocks(title="Hybrid NER + GLiNER Tool", theme=gr.themes.Soft(), css=custom_css) as demo:
|
| 777 |
gr.Markdown("""
|
| 778 |
-
# π―
|
| 779 |
|
| 780 |
Combine common NER categories with your own custom entity types! This tool uses both traditional NER models and GLiNER for comprehensive entity extraction.
|
| 781 |
|
| 782 |
-
## π€
|
| 783 |
|
| 784 |
### How to use:
|
| 785 |
1. **π Enter your text** in the text area below
|
|
@@ -812,14 +771,6 @@ def create_interface():
|
|
| 812 |
with gr.Column():
|
| 813 |
gr.Markdown("### π― Common Entity Types")
|
| 814 |
|
| 815 |
-
# Add the inline help text
|
| 816 |
-
gr.HTML("""
|
| 817 |
-
<div style="background-color: #e3f2fd; padding: 10px; border-radius: 5px; margin-bottom: 10px; font-size: 13px;">
|
| 818 |
-
<strong>Quick Guide:</strong> PER = People | ORG = Organizations | LOC = Locations |
|
| 819 |
-
GPE = Countries/Cities | FAC = Facilities | DATE = Dates | EVENT = Named Events
|
| 820 |
-
</div>
|
| 821 |
-
""")
|
| 822 |
-
|
| 823 |
# Model selector
|
| 824 |
model_dropdown = gr.Dropdown(
|
| 825 |
choices=ner_manager.model_names,
|
|
@@ -828,39 +779,11 @@ def create_interface():
|
|
| 828 |
info="Choose which model to use for common NER"
|
| 829 |
)
|
| 830 |
|
| 831 |
-
#
|
| 832 |
-
glossary_html = """
|
| 833 |
-
<details style="margin: 10px 0; padding: 10px; background-color: #f8f9fa; border-radius: 8px; border: 1px solid #ddd;">
|
| 834 |
-
<summary style="cursor: pointer; font-weight: bold; padding: 5px; color: #1976d2;">
|
| 835 |
-
βΉοΈ Detailed Entity Type Definitions (Click to expand)
|
| 836 |
-
</summary>
|
| 837 |
-
<div style="margin-top: 10px; padding: 10px;">
|
| 838 |
-
<dl style="margin: 0; font-size: 14px;">
|
| 839 |
-
"""
|
| 840 |
-
|
| 841 |
-
for entity, definition in ENTITY_DEFINITIONS.items():
|
| 842 |
-
colour_key = "WORK OF ART" if entity == "Work of Art" else entity.upper()
|
| 843 |
-
colour = STANDARD_COLORS.get(colour_key, '#CCCCCC')
|
| 844 |
-
glossary_html += f"""
|
| 845 |
-
<div style="margin-bottom: 8px;">
|
| 846 |
-
<dt style="font-weight: bold; display: inline; color: {colour};">{entity}:</dt>
|
| 847 |
-
<dd style="display: inline; margin-left: 5px;">{definition}</dd>
|
| 848 |
-
</div>
|
| 849 |
-
"""
|
| 850 |
-
|
| 851 |
-
glossary_html += """
|
| 852 |
-
</dl>
|
| 853 |
-
</div>
|
| 854 |
-
</details>
|
| 855 |
-
"""
|
| 856 |
-
gr.HTML(glossary_html)
|
| 857 |
-
|
| 858 |
-
# Common entities checkboxes with colored labels
|
| 859 |
standard_entities = gr.CheckboxGroup(
|
| 860 |
-
choices=
|
| 861 |
-
value=[
|
| 862 |
-
label="Select Common Entities"
|
| 863 |
-
elem_id="standard-entities-checkbox"
|
| 864 |
)
|
| 865 |
|
| 866 |
# Select/Deselect All button
|
|
@@ -874,7 +797,7 @@ def create_interface():
|
|
| 874 |
return [], "βοΈ Select All"
|
| 875 |
else:
|
| 876 |
# If none selected, select all
|
| 877 |
-
return
|
| 878 |
|
| 879 |
select_all_btn.click(
|
| 880 |
fn=toggle_all_entities,
|
|
@@ -883,7 +806,7 @@ def create_interface():
|
|
| 883 |
)
|
| 884 |
|
| 885 |
with gr.Column():
|
| 886 |
-
gr.Markdown("### β¨ Custom Entity Types
|
| 887 |
custom_entities = gr.Textbox(
|
| 888 |
label="Custom Entities (comma-separated)",
|
| 889 |
placeholder="e.g. relationships, occupations, skills, emotions",
|
|
@@ -895,8 +818,6 @@ def create_interface():
|
|
| 895 |
- emotions, actions, objects
|
| 896 |
- medical conditions, treatments
|
| 897 |
- financial terms, business roles
|
| 898 |
-
|
| 899 |
-
*GLiNER model will extract these custom entity types from your text*
|
| 900 |
""")
|
| 901 |
|
| 902 |
analyse_btn = gr.Button("π Analyse Text", variant="primary", size="lg")
|
|
@@ -914,15 +835,9 @@ def create_interface():
|
|
| 914 |
gr.Markdown("### π Detailed Results")
|
| 915 |
results_output = gr.HTML(label="Entity Results")
|
| 916 |
|
| 917 |
-
# Function to process with colored labels
|
| 918 |
-
def process_with_colored_labels(text, colored_labels, custom_entities, confidence_threshold, selected_model, progress=gr.Progress()):
|
| 919 |
-
# Remove the color dots from labels
|
| 920 |
-
standard_entities = [label.replace("β ", "") for label in colored_labels]
|
| 921 |
-
return process_text(text, standard_entities, custom_entities, confidence_threshold, selected_model, progress)
|
| 922 |
-
|
| 923 |
# Connect the button to the processing function
|
| 924 |
analyse_btn.click(
|
| 925 |
-
fn=
|
| 926 |
inputs=[
|
| 927 |
text_input,
|
| 928 |
standard_entities,
|
|
@@ -938,21 +853,21 @@ def create_interface():
|
|
| 938 |
examples=[
|
| 939 |
[
|
| 940 |
"John Smith works at Google in New York. He graduated from Stanford University in 2015 and specialises in artificial intelligence research. His wife Sarah is a doctor at Mount Sinai Hospital.",
|
| 941 |
-
[
|
| 942 |
"relationships, occupations, educational background",
|
| 943 |
0.3,
|
| 944 |
"entities_spacy_en_core_web_trf"
|
| 945 |
],
|
| 946 |
[
|
| 947 |
"The meeting between CEO Jane Doe and the board of directors at Microsoft headquarters in Seattle discussed the Q4 financial results and the new AI strategy for 2024.",
|
| 948 |
-
[
|
| 949 |
"corporate roles, business events, financial terms",
|
| 950 |
0.4,
|
| 951 |
"entities_flair_ner-ontonotes-large"
|
| 952 |
],
|
| 953 |
[
|
| 954 |
"Dr. Emily Watson published a research paper on machine learning algorithms at MIT. She collaborates with her colleague Prof. David Chen on natural language processing projects.",
|
| 955 |
-
[
|
| 956 |
"academic titles, research topics, collaborations",
|
| 957 |
0.3,
|
| 958 |
"entities_gliner_knowledgator/modern-gliner-bi-large-v1.0"
|
|
@@ -966,7 +881,6 @@ def create_interface():
|
|
| 966 |
model_dropdown
|
| 967 |
]
|
| 968 |
)
|
| 969 |
-
|
| 970 |
# Add model information links
|
| 971 |
gr.HTML("""
|
| 972 |
<hr style="margin-top: 40px; margin-bottom: 20px;">
|
|
|
|
| 30 |
'WORK OF ART': '#DDA0DD' # Plum
|
| 31 |
}
|
| 32 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 33 |
# Additional colours for custom entities
|
| 34 |
CUSTOM_COLOR_PALETTE = [
|
| 35 |
'#FF9F43', '#10AC84', '#EE5A24', '#0FBC89', '#5F27CD',
|
|
|
|
| 730 |
|
| 731 |
return summary, legend_html + highlighted_html, results_html
|
| 732 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 733 |
# Create Gradio interface
|
| 734 |
def create_interface():
|
| 735 |
+
with gr.Blocks(title="Hybrid NER + GLiNER Tool", theme=gr.themes.Soft()) as demo:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 736 |
gr.Markdown("""
|
| 737 |
+
# π― Named Entity Recognition (NER) Explorer Tool
|
| 738 |
|
| 739 |
Combine common NER categories with your own custom entity types! This tool uses both traditional NER models and GLiNER for comprehensive entity extraction.
|
| 740 |
|
| 741 |
+
## π€ Overlapping entities are automatically shared with split-colour highlighting!
|
| 742 |
|
| 743 |
### How to use:
|
| 744 |
1. **π Enter your text** in the text area below
|
|
|
|
| 771 |
with gr.Column():
|
| 772 |
gr.Markdown("### π― Common Entity Types")
|
| 773 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 774 |
# Model selector
|
| 775 |
model_dropdown = gr.Dropdown(
|
| 776 |
choices=ner_manager.model_names,
|
|
|
|
| 779 |
info="Choose which model to use for common NER"
|
| 780 |
)
|
| 781 |
|
| 782 |
+
# Common entities with select all functionality
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 783 |
standard_entities = gr.CheckboxGroup(
|
| 784 |
+
choices=STANDARD_ENTITIES,
|
| 785 |
+
value=['PER', 'ORG', 'LOC', 'MISC'], # Default selection
|
| 786 |
+
label="Select Common Entities"
|
|
|
|
| 787 |
)
|
| 788 |
|
| 789 |
# Select/Deselect All button
|
|
|
|
| 797 |
return [], "βοΈ Select All"
|
| 798 |
else:
|
| 799 |
# If none selected, select all
|
| 800 |
+
return STANDARD_ENTITIES, "π Deselect All"
|
| 801 |
|
| 802 |
select_all_btn.click(
|
| 803 |
fn=toggle_all_entities,
|
|
|
|
| 806 |
)
|
| 807 |
|
| 808 |
with gr.Column():
|
| 809 |
+
gr.Markdown("### β¨ Custom Entity Types")
|
| 810 |
custom_entities = gr.Textbox(
|
| 811 |
label="Custom Entities (comma-separated)",
|
| 812 |
placeholder="e.g. relationships, occupations, skills, emotions",
|
|
|
|
| 818 |
- emotions, actions, objects
|
| 819 |
- medical conditions, treatments
|
| 820 |
- financial terms, business roles
|
|
|
|
|
|
|
| 821 |
""")
|
| 822 |
|
| 823 |
analyse_btn = gr.Button("π Analyse Text", variant="primary", size="lg")
|
|
|
|
| 835 |
gr.Markdown("### π Detailed Results")
|
| 836 |
results_output = gr.HTML(label="Entity Results")
|
| 837 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 838 |
# Connect the button to the processing function
|
| 839 |
analyse_btn.click(
|
| 840 |
+
fn=process_text,
|
| 841 |
inputs=[
|
| 842 |
text_input,
|
| 843 |
standard_entities,
|
|
|
|
| 853 |
examples=[
|
| 854 |
[
|
| 855 |
"John Smith works at Google in New York. He graduated from Stanford University in 2015 and specialises in artificial intelligence research. His wife Sarah is a doctor at Mount Sinai Hospital.",
|
| 856 |
+
["PER", "ORG", "LOC", "DATE"],
|
| 857 |
"relationships, occupations, educational background",
|
| 858 |
0.3,
|
| 859 |
"entities_spacy_en_core_web_trf"
|
| 860 |
],
|
| 861 |
[
|
| 862 |
"The meeting between CEO Jane Doe and the board of directors at Microsoft headquarters in Seattle discussed the Q4 financial results and the new AI strategy for 2024.",
|
| 863 |
+
["PER", "ORG", "LOC", "DATE"],
|
| 864 |
"corporate roles, business events, financial terms",
|
| 865 |
0.4,
|
| 866 |
"entities_flair_ner-ontonotes-large"
|
| 867 |
],
|
| 868 |
[
|
| 869 |
"Dr. Emily Watson published a research paper on machine learning algorithms at MIT. She collaborates with her colleague Prof. David Chen on natural language processing projects.",
|
| 870 |
+
["PER", "ORG", "Work of Art"],
|
| 871 |
"academic titles, research topics, collaborations",
|
| 872 |
0.3,
|
| 873 |
"entities_gliner_knowledgator/modern-gliner-bi-large-v1.0"
|
|
|
|
| 881 |
model_dropdown
|
| 882 |
]
|
| 883 |
)
|
|
|
|
| 884 |
# Add model information links
|
| 885 |
gr.HTML("""
|
| 886 |
<hr style="margin-top: 40px; margin-bottom: 20px;">
|