Add files via upload

Assignment Code Version 1.1.py

@@ -0,0 +1,74 @@
+import os
+import tarfile
+from tqdm import tqdm
+import torch
+from sentence_transformers import SentenceTransformer
+from sklearn.metrics.pairwise import cosine_similarity
+
+# Function to read citation network dataset
+def read_citation_network(file_path):
+    citations = {}
+    with open(file_path, 'r') as file:
+        for line in file:
+            source, target = line.strip().split('\t')
+            if source not in citations:
+                citations[source] = []
+            citations[source].append(target)
+    return citations
+
+# Function to read abstracts dataset
+def read_abstracts_dataset(abstracts_path):
+    abstracts = {}
+    for folder in os.listdir(abstracts_path):
+        folder_path = os.path.join(abstracts_path, folder)
+        if os.path.isdir(folder_path):
+            for file in os.listdir(folder_path):
+                if file.endswith('.abs'):
+                    paper_id = file.split('.')[0]
+                    with open(os.path.join(folder_path, file), 'r', encoding='latin1') as f:
+                        content = f.read().strip()
+                        abstracts[paper_id] = content
+    return abstracts
+
+# Function to calculate cosine similarity between two vectors
+def calculate_cosine_similarity(vec1, vec2):
+    return cosine_similarity(vec1.reshape(1, -1), vec2.reshape(1, -1))[0][0]
+
+# Function to embed text using Sci-BERT model
+def embed_text(text):
+    return model.encode(text)
+
+# Load Sci-BERT model
+model = SentenceTransformer('allenai/scibert_scivocab_uncased')
+
+# Paths to datasets
+citation_network_file = 'cit-HepTh.txt'
+abstracts_path = 'cit-HepTh-abstracts.tar.gz'
+
+# Read citation network dataset
+citation_network = read_citation_network(citation_network_file)
+
+# Read abstracts dataset
+abstracts = read_abstracts_dataset(abstracts_path)
+
+# Paper IDs for which similarity scores will be calculated
+seed_paper_ids = ['9201001', '9203201', '119203001']
+
+# Calculate similarity scores for seed papers
+for paper_id in seed_paper_ids:
+    if paper_id in citation_network:
+        print(f"Seed Paper ID: {paper_id}")
+        print("References:")
+        for reference_id in citation_network[paper_id]:
+            if reference_id in abstracts:
+                seed_paper_abstract = abstracts[paper_id]
+                reference_abstract = abstracts[reference_id]
+                # Embed abstracts
+                seed_paper_embedding = embed_text(seed_paper_abstract)
+                reference_embedding = embed_text(reference_abstract)
+                # Calculate cosine similarity
+                similarity_score = calculate_cosine_similarity(seed_paper_embedding, reference_embedding)
+                print(f"Reference Paper ID: {reference_id}, Similarity Score: {similarity_score:.4f}")
+        print()
+    else:
+        print(f"No references found for Seed Paper ID: {paper_id}")