modules/text_analysis/semantic_analysis.py

import streamlit as st
import spacy
import networkx as nx
import matplotlib.pyplot as plt
from collections import Counter
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.metrics.pairwise import cosine_similarity

# ... (mantén las definiciones de POS_COLORS, POS_TRANSLATIONS, y ENTITY_LABELS como están)

def identify_key_concepts(doc):
    word_freq = Counter([token.lemma_.lower() for token in doc if token.pos_ in ['NOUN', 'VERB'] and not token.is_stop])
    return word_freq.most_common(10)  # Top 10 conceptos clave

def create_concept_graph(text, concepts):
    vectorizer = TfidfVectorizer()
    tfidf_matrix = vectorizer.fit_transform([text])
    concept_vectors = vectorizer.transform([c[0] for c in concepts])
    similarity_matrix = cosine_similarity(concept_vectors, concept_vectors)

    G = nx.Graph()
    for i, (concept, weight) in enumerate(concepts):
        G.add_node(concept, weight=weight)
        for j in range(i+1, len(concepts)):
            if similarity_matrix[i][j] > 0.1:
                G.add_edge(concept, concepts[j][0], weight=similarity_matrix[i][j])

    return G

def visualize_concept_graph(G, lang):
    fig, ax = plt.subplots(figsize=(15, 10))
    pos = nx.spring_layout(G, k=0.5, iterations=50)
    
    node_sizes = [G.nodes[node]['weight'] * 100 for node in G.nodes()]
    nx.draw_networkx_nodes(G, pos, node_size=node_sizes, node_color='lightblue', alpha=0.8, ax=ax)
    nx.draw_networkx_labels(G, pos, font_size=10, font_weight="bold", ax=ax)
    nx.draw_networkx_edges(G, pos, width=1, alpha=0.5, ax=ax)
    
    edge_labels = nx.get_edge_attributes(G, 'weight')
    nx.draw_networkx_edge_labels(G, pos, edge_labels=edge_labels, font_size=8, ax=ax)

    title = {
        'es': "Relaciones entre Conceptos Clave",
        'en': "Key Concept Relations",
        'fr': "Relations entre Concepts Clés"
    }
    ax.set_title(title[lang], fontsize=16)
    ax.axis('off')

    plt.tight_layout()
    return fig

def perform_semantic_analysis(text, nlp, lang):
    doc = nlp(text)

    # Identificar conceptos clave
    key_concepts = identify_key_concepts(doc)

    # Crear y visualizar grafo de conceptos
    concept_graph = create_concept_graph(text, key_concepts)
    relations_graph = visualize_concept_graph(concept_graph, lang)
    
    return {
        'key_concepts': key_concepts,
        'relations_graph': relations_graph
    }

__all__ = ['perform_semantic_analysis', 'ENTITY_LABELS', 'POS_TRANSLATIONS']