import nltk import re nltk.download('wordnet') nltk.download('punkt') nltk.download('stopwords') nltk.download('averaged_perceptron_tagger') nltk.download('maxent_ne_chunker') nltk.download('words') nltk.download('brown') from newspaper import Article from newspaper import fulltext import requests import itertools from nltk.tokenize import word_tokenize from sentence_transformers import SentenceTransformer import pandas as pd import numpy as np from pandas import ExcelWriter from torch.utils.data import DataLoader import math from sentence_transformers import models, losses from sentence_transformers import SentencesDataset, LoggingHandler, SentenceTransformer from sentence_transformers.evaluation import EmbeddingSimilarityEvaluator from sentence_transformers.readers import * from nltk.corpus import stopwords stop_words = stopwords.words('english') import matplotlib.pyplot as plt from sklearn.cluster import KMeans from sklearn.decomposition import PCA from sklearn.metrics.pairwise import cosine_similarity import scipy.spatial import networkx as nx from nltk.tokenize import sent_tokenize import scispacy import spacy import en_core_sci_lg import string from nltk.stem.wordnet import WordNetLemmatizer import gradio as gr import inflect from Bio import Entrez from sklearn.cluster import KMeans from sklearn.cluster import AgglomerativeClustering from sklearn.metrics import silhouette_samples, silhouette_score, davies_bouldin_score import json p = inflect.engine() nlp = en_core_sci_lg.load() sp = en_core_sci_lg.load() all_stopwords = sp.Defaults.stop_words word_embedding_model = models.Transformer('cambridgeltl/SapBERT-from-PubMedBERT-fulltext') pooling_model = models.Pooling(word_embedding_model.get_word_embedding_dimension(), pooling_mode_mean_tokens=True, pooling_mode_cls_token=False, pooling_mode_max_tokens=False) embedder = SentenceTransformer(modules=[word_embedding_model, pooling_model]) def search(query): Entrez.email = 'naseer@example.com' handle = Entrez.esearch(db='pubmed', sort='relevance', retmax='5', retmode='xml', term=query) results = Entrez.read(handle) return results def fetch_details(id_list): ids = ','.join(id_list) Entrez.email = 'naseer@example.com' handle_1 = Entrez.efetch(db='pubmed', retmode='xml', id=ids) results_1 = Entrez.read(handle_1) return results_1 def remove_stopwords(sen): sen_new = " ".join([i for i in sen if i not in stop_words]) return sen_new def keyphrase_generator(article_link, model_1, model_2, max_num_keywords): element=[] final_textrank_list=[] document=[] text_doc=[] final_list=[] score_list=[] sum_list=[] model_1 = SentenceTransformer(model_1) model_2 = SentenceTransformer(model_2) url = article_link if (url == False): print("error") html = requests.get(url).text article = fulltext(html) corpus=sent_tokenize(article) indicator_list=['concluded','concludes','in a study', 'concluding','conclude','in sum','in a recent study','therefore','thus','so','hence', 'as a result','accordingly','consequently','in short','proves that','shows that','suggests that','demonstrates that','found that','observed that', 'indicated that','suggested that','demonstrated that'] count_dict={} for l in corpus: c=0 for l2 in indicator_list: if l.find(l2)!=-1:#then it is a substring c=1 break if c:# count_dict[l]=1 else: count_dict[l]=0 for sent, score in count_dict.items(): score_list.append(score) clean_sentences_new = pd.Series(corpus).str.replace("[^a-zA-Z]", " ").tolist() corpus_embeddings = model_1.encode(clean_sentences_new) sim_mat = np.zeros([len(clean_sentences_new), len(clean_sentences_new)]) for i in range(len(clean_sentences_new)): len_embeddings=(len(corpus_embeddings[i])) for j in range(len(clean_sentences_new)): if i != j: if(len_embeddings == 1024): sim_mat[i][j] = cosine_similarity(corpus_embeddings[i].reshape(1,1024), corpus_embeddings[j].reshape(1,1024))[0,0] elif(len_embeddings == 768): sim_mat[i][j] = cosine_similarity(corpus_embeddings[i].reshape(1,768), corpus_embeddings[j].reshape(1,768))[0,0] nx_graph = nx.from_numpy_array(sim_mat) scores = nx.pagerank(nx_graph) sentences=((scores[i],s) for i,s in enumerate(corpus)) for elem in sentences: element.append(elem[0]) for sc, lst in zip(score_list, element): ########## taking the scores from both the lists sum1=sc+lst sum_list.append(sum1) x=sorted(((sum_list[i],s) for i,s in enumerate(corpus)), reverse=True) for elem in x: final_textrank_list.append(elem[1]) a=int((10*len(final_textrank_list))/100.0) if(a<5): total=5 else: total=int(a) for i in range(total): document.append(final_textrank_list[i]) doc=" ".join(document) for i in document: doc_1=nlp(i) text_doc.append([X.text for X in doc_1.ents]) entity_list = [item for sublist in text_doc for item in sublist] entity_list = [word for word in entity_list if not word in all_stopwords] entity_list = [word_entity for word_entity in entity_list if(p.singular_noun(word_entity) == False)] entity_list=list(dict.fromkeys(entity_list)) doc_embedding = model_2.encode([doc]) candidates=entity_list candidate_embeddings = model_2.encode(candidates) distances = cosine_similarity(doc_embedding, candidate_embeddings) top_n = max_num_keywords keyword_list = [candidates[index] for index in distances.argsort()[0][-top_n:]] keywords = '\n'.join(keyword_list) c_len=(len(keyword_list)) keyword_embeddings = embedder.encode(keyword_list) data_embeddings = embedder.encode(keyword_list) for num_clusters in range(1, top_n): clustering_model = KMeans(n_clusters=num_clusters) clustering_model.fit(keyword_embeddings) cluster_assignment = clustering_model.labels_ clustered_sentences = [[] for i in range(num_clusters)] for sentence_id, cluster_id in enumerate(cluster_assignment): clustered_sentences[cluster_id].append(keyword_list[sentence_id]) cl_sent_len=(len(clustered_sentences)) list_cluster=list(clustered_sentences) a=len(list_cluster) cluster_list_final.append(list_cluster) if (c_len==cl_sent_len and c_len>=3) or cl_sent_len==1: silhouette_avg = 0 silhouette_score_list.append(silhouette_avg) elif c_len==cl_sent_len==2: silhouette_avg = 1 silhouette_score_list.append(silhouette_avg) else: silhouette_avg = silhouette_score(keyword_embeddings, cluster_assignment) silhouette_score_list.append(silhouette_avg) res_dict = dict(zip(silhouette_score_list, cluster_list_final)) cluster_items=res_dict[max(res_dict)] for i in cluster_items: z=' OR '.join(i) comb.append("("+z+")") comb_list.append(comb) combinations = [] for subset in itertools.combinations(comb, 2): combinations.append(subset) f1_list=[] for s in combinations: final = ' AND '.join(s) f1_list.append("("+final+")") f_1=' OR '.join(f1_list) final_list.append(f_1) ncbi_url='https://eutils.ncbi.nlm.nih.gov/entrez/eutils/' last_url='esearch.fcgi?db=pubmed'+'&term='+f_1 search_rettype = '&rettype=json' overall_url=ncbi_url+last_url+search_rettype+'&sort=relevance' r = requests.get(overall_url) root = ET.fromstring(r.text) levels = root.findall('.//Id') name_list=[] for level in levels: name = level.text name_list.append(name) name_1 = ','.join(name_list) fetch_url='efetch.fcgi?db=pubmed' search_id='&id='+name_1 ret_type='&rettype=text' ret_mode='&retmode=xml' ret_max='&retmax=10' ret_sort='&sort=relevance' return_url=ncbi_url+fetch_url+search_id+ret_type+ret_mode+ret_max+ret_sort r_1 = requests.get(return_url) root_1 = ET.fromstring(r_1.text) levels_1 = root_1.findall('.//ArticleTitle') for level in levels_1: name = level.text title_list.append(name) return title_list gr.Interface(keyphrase_generator, inputs=[gr.inputs.Textbox(lines=1, placeholder="Provide article web link here",default="", label="Article web link"), gr.inputs.Dropdown(choices=['sentence-transformers/all-mpnet-base-v2', 'sentence-transformers/all-mpnet-base-v1', 'sentence-transformers/all-distilroberta-v1', 'sentence-transformers/gtr-t5-large', 'pritamdeka/S-Bluebert-snli-multinli-stsb', 'pritamdeka/S-Biomed-Roberta-snli-multinli-stsb', 'sentence-transformers/stsb-mpnet-base-v2', 'sentence-transformers/stsb-roberta-base-v2', 'sentence-transformers/stsb-distilroberta-base-v2', 'sentence-transformers/sentence-t5-large', 'sentence-transformers/sentence-t5-base'], type="value", default='sentence-transformers/all-mpnet-base-v1', label="Select any SBERT model for TextRank from the list below"), gr.inputs.Dropdown(choices=['sentence-transformers/paraphrase-mpnet-base-v2', 'sentence-transformers/all-mpnet-base-v1', 'sentence-transformers/paraphrase-distilroberta-base-v1', 'sentence-transformers/paraphrase-xlm-r-multilingual-v1', 'sentence-transformers/paraphrase-multilingual-mpnet-base-v2', 'sentence-transformers/paraphrase-albert-small-v2', 'sentence-transformers/paraphrase-albert-base-v2', 'sentence-transformers/paraphrase-MiniLM-L12-v2', 'sentence-transformers/paraphrase-MiniLM-L6-v2', 'sentence-transformers/all-MiniLM-L12-v2', 'sentence-transformers/all-distilroberta-v1', 'sentence-transformers/paraphrase-TinyBERT-L6-v2', 'sentence-transformers/paraphrase-MiniLM-L3-v2', 'sentence-transformers/all-MiniLM-L6-v2'], type="value", default='sentence-transformers/all-mpnet-base-v1', label="Select any SBERT model for keyphrases from the list below"), gr.inputs.Slider(minimum=5, maximum=30, step=1, default=10, label="Max Keywords")], outputs=gr.outputs.Textbox(type="auto", label="Stuff"), theme="peach", title="Scientific Article Keyphrase Generator", description="Generates the keyphrases from an article which best describes the article.", article= "The work is based on a part of the paper provided here." "\t It uses the TextRank algorithm with SBERT to first find the top sentences and then extracts the keyphrases from those sentences using scispaCy and SBERT." "\t The list of SBERT models required in the textboxes can be found in SBERT Pre-trained models hub." "\t The default model names are provided which can be changed from the list of pretrained models. " "\t The value of output keyphrases can be changed. The default value is 10, minimum is 5 and a maximum value of 30.").launch(share=True,debug=True)