import streamlit as st import pandas as pd from top2vec import Top2Vec from wordcloud import WordCloud @st.cache_resource def load_model(): name = '2024-02-14_22_19_09_reduced__model_medline0[test,mincount_250,topicmergedelta_0.15,ngram_True,ngramvocab_args_250_.4_80M_npmi_connector].json' with open(name, 'rb') as f: return Top2Vec.load(f) @st.cache_data def load_data(): df = pd.read_pickle('medline_0_df_nlp.pkl.bz2', compression='bz2') return df # Notification of loading start_container = st.empty() if 'first' not in st.session_state: st.session_state['first'] = True if st.session_state['first'] == True: start_container.text('Loading model and data... This may take two to three minutes on the initial run.') st.session_state['first'] = False # Load the model and data model = load_model() df = load_data() # Remove loading notification start_container.empty() # User interface with st.sidebar.form('Explore Topics', clear_on_submit=False): st.write('Input a search term(s) separated by a space and receive the topics that are most semantically similar.') txt_et = st.text_area('Keyword(s) or Phrase(s)') num_et = st.number_input('Number of topics to return', min_value=1, max_value=model.get_num_topics(reduced=True), # Number of topics in the reduced model value=10) # Default value is 10 submitted_et = st.form_submit_button("Submit") if submitted_et: txt_et_list = txt_et.split() topic_words, word_scores, topic_scores, topic_nums = model.search_topics(txt_et_list, num_topics=num_et, reduced=True) for i, topic in enumerate(topic_nums): st.subheader('Topic Number: ' + str(topic)) st.write('Search Cosine Similarity: ' + str(topic_scores[i])) st.write('Topic Words:', ', '.join(topic_words[i][:10])) # Top documents for each topic with st.expander('Top Documents in Topic'): _, document_ids = model.search_documents_by_topic(topic_num=topic, reduced=True, num_docs=10) for id in document_ids: st.markdown(f'[{df.at[id, "TI"]}](https://pubmed.ncbi.nlm.nih.gov/{df.at[id, "PMID"]})') # Wordcloud new_dict = {} for ii, words in enumerate(topic_words[i]): new_dict[words] = word_scores[i][ii] st.image(WordCloud(width=850, height=250, background_color='white', prefer_horizontal=0.7).generate_from_frequencies(new_dict).to_image())