Merge pull request #162 from trigaten/pdf_graphs_doc

Make Graphs save as PDFs

src/prompt_systematic_review/experiments/__init__.py

@@ -30,7 +30,5 @@
     graph_gpt_3_5_benchmarks.Experiment,
     run_tomotopy.Experiment,
     topicgpt.Experiment,
-    download_mmlu.Experiment,
-    graph_internal_references.Experiment,
-    graph.Experiment,
+    count_models.Experiment,
 ]

src/prompt_systematic_review/experiments/graph.py

@@ -9,6 +9,7 @@
 import networkx as nx
 import matplotlib.pyplot as plt
 import textwrap
+from prompt_systematic_review.config_data import DataFolderPath
 
 
 class SemanticScholarAPI:
@@ -85,7 +86,13 @@ def visualize_citation_counts(self, paper_references, technique_to_title):
         plt.title("Citation Counts by Technique")
         plt.xticks(rotation=45)
         plt.tight_layout()
-        plt.show()
+        plt.savefig(
+            os.path.join(
+                DataFolderPath, "experiments_output" + os.sep + "paper_graph.pdf"
+            ),
+            format="pdf",
+            bbox_inches="tight",
+        )
 
 
 class Main:

src/prompt_systematic_review/experiments/graph_dataset_citations.py

@@ -87,9 +87,9 @@ def graph_dataset_citations():
     output_dir = os.path.join(DataFolderPath, "experiments_output")
     os.makedirs(output_dir, exist_ok=True)
     # output_file_path = os.path.join(output_dir, "graph_dataset_mentions_output.png")
-    output_file_path = os.path.join(output_dir, "graph_dataset_mentions_output.png")
+    output_file_path = os.path.join(output_dir, "graph_dataset_mentions_output.pdf")
 
-    plt.savefig(output_file_path)
+    plt.savefig(output_file_path, format="pdf", bbox_inches="tight")
 
 
 class Experiment:

src/prompt_systematic_review/experiments/graph_gpt_3_5_benchmarks.py

@@ -48,8 +48,10 @@ def graph_gpt_3_5():
     plt.savefig(
         os.path.join(
             DataFolderPath,
-            "experiments_output" + os.sep + "graph_gpt_3_5_benchmarks_output.png",
-        )
+            "experiments_output" + os.sep + "graph_gpt_3_5_benchmarks_output.pdf",
+        ),
+        format="pdf",
+        bbox_inches="tight",
     )
 
 

src/prompt_systematic_review/experiments/graph_gpt_4_benchmarks200.py

@@ -92,8 +92,10 @@ def graph_gt_4_benchmarks200():
     plt.savefig(
         os.path.join(
             DataFolderPath,
-            "experiments_output" + os.sep + "graph_gpt_4_benchmarks200_output.png",
-        )
+            "experiments_output" + os.sep + "graph_gpt_4_benchmarks200_output.pdf",
+        ),
+        format="pdf",
+        bbox_inches="tight",
     )
 
 

src/prompt_systematic_review/experiments/graph_internal_references.py

@@ -333,7 +333,7 @@ def visualize_graph(self, paper_references, technique_to_title):
         plt.axis("off")
 
         # plt.show()
-        plt.savefig("network_graph.png", format="png", dpi=300)
+        plt.savefig("network_graph.pdf", format="pdf", bbox_inches="tight")
 
     def visualize_citation_counts(self, paper_references, title_to_technique):
         citation_counts = {}

src/prompt_systematic_review/experiments/graph_models.py

@@ -28,9 +28,9 @@ def graph_models(inputFile="model_citation_counts.csv"):
     os.makedirs(output_dir, exist_ok=True)  # Create directory if it doesn't exist
 
     # output_file_path = os.path.join(output_dir, "graph_models_output.png")
-    output_file_path = os.path.join("graph_models_output.png")
+    output_file_path = os.path.join("graph_models_output.pdf")
 
-    plt.savefig(output_file_path)
+    plt.savefig(output_file_path, format="pdf", bbox_inches="tight")
 
 
 class Experiment:

src/prompt_systematic_review/experiments/topicgpt.py

@@ -73,7 +73,9 @@ def run_topic_gpt():
     plt.savefig(
         os.path.join(
             DataFolderPath, "experiments_output" + os.sep + "topic_distribution.png"
-        )
+        ),
+        format="pdf",
+        bbox_inches="tight",
     )
 
 

src/prompt_systematic_review/experiments/visualize_authors.py

@@ -45,8 +45,10 @@ def visualize_authors():
     plt.savefig(
         os.path.join(
             DataFolderPath,
-            "experiments_output" + os.sep + "publication_frequency_by_author.png",
-        )
+            "experiments_output" + os.sep + "publication_frequency_by_author.pdf",
+        ),
+        format="pdf",
+        bbox_inches="tight",
     )
 
     # Display the frequency table
@@ -76,8 +78,10 @@ def visualize_authors():
     plt.savefig(
         os.path.join(
             DataFolderPath,
-            "experiments_output" + os.sep + "publication_count_vs_frequency.png",
-        )
+            "experiments_output" + os.sep + "publication_count_vs_frequency.pdf",
+        ),
+        format="pdf",
+        bbox_inches="tight",
     )