simplify data population

task_specific_data_population.py

@@ -7,7 +7,7 @@ import json_repair
 class DataPopulation:
     def __init__(self, api_key):
         # Set the API key during initialization
-        self.client = OpenAI(api_key=api_key)
+        self.client = OpenAI(api_key=openai_api_key)
         self.conversation = [
             {
                 "role": "system",
@@ -27,7 +27,6 @@ class DataPopulation:
             }
         ]
 
-
     def fetch_huggingface_dataset(self, dataset_name):
         """Fetch the dataset from Hugging Face."""
         return load_dataset(dataset_name)
@@ -62,41 +61,24 @@ class DataPopulation:
         self.conversation.append({
             "role": "user",
             "content": (
-                f"Given the task: '{task}' and the following page details:\n{relevant_page_details}\n\n"
-                "Instructions:\n"
-                "1. Ensure each page contains the following keys: 'page_name', 'features', and 'user_data' (even if empty).\n"
-                "2. Update 'user_data' with essential information relevant to the task. Do not modify the 'features' section.\n"
-                "3. Maintain the exact input structure, making changes only to 'user_data' where required.\n"
-                "4. Confirm that both 'features' and 'user_data' are dictionaries.\n"
-                "5. Verify that every page includes exactly the three required keys: 'page_name', 'features', and 'user_data'."
-                "6. Make sure 'user_data' must exist as a key! "
+                f"Given the task: '{task}' and the following task-relevant page details:\n{relevant_page_details}\n\n"
+                f"Update each page's 'user_data' field with essential information for task-completion."
+                f"Do not modify the other fields"
+                "Output should maintain the exact input structure, making changes only to 'user_data' where required."
+                "'user_data' must exists as a key in each page.\n"
             )
         })
         response_content = self.gpt4_chat(self.conversation)
         return response_content
 
-    def ask_to_update_sitemap(self, sitemap, new_page):
-        """Update the sitemap with the new page and adjust links."""
-        self.conversation.append({
-            "role": "user",
-            "content": (
-                f"Given the current sitemap:\n{sitemap}\nand a new page: '{new_page}' (currently only has a 'uid'), update the sitemap by:\n"
-                "- Adding necessary details to '{new_page}', including 'links_to'.\n"
-                "- Updating 'links_to' in other relevant pages to include '{new_page}'.\n"
-                "Ensure the output retains the exact structure of the input."
-            )
-        })
-        response_sitemap = self.gpt4_chat(self.conversation)
-        return json_repair.loads(response_sitemap)
-
     def ask_to_update_user_state(self, task, user_state):
         """Update the user state based on the task."""
         self.conversation.append({
             "role": "user",
             "content": (
-                f"Given the task: '{task}' and the current user state:\n{user_state}\n\n"
-                "Update the user state to reflect any changes necessary for completing the task. "
-                "Only modify values that are required; maintain the exact structure of the input."
+                f"Given the task: '{task}', default user state:\n{user_state}, and user_data in chat history.\n\n"
+                "Initialize the user state values to reflect any initial status necessary for completing the task. "
+                "Output should maintain the exact input structure."
             )
         })
         response_content = self.gpt4_chat(self.conversation)