siavash/Meyar
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

First Version

Browse Source
This commit is contained in:
aliakbar
2026-06-27 07:52:17 +03:30
commit f3d28cea99
97 changed files with 54562 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
notebooks/Step2_ValueAnlyser.ipynb
+301
View File
@@ -0,0 +1,301 @@
{
"cells": [
{
"cell_type": "code",
"execution_count": 1,
"id": "6e0b19d0",
"metadata": {},
"outputs": [],
"source": [
"# this file is loading the articles and the value measures JSON file \n",
"# for each article and value measure it will send to LLM for an analysis\n",
"# user can select which list of measures to be used for the analysis (orginal or opposite)\n",
"# the analysis and th model input are saved into a JSN file"
]
},
{
"cell_type": "code",
"execution_count": 2,
"id": "dd37ecab",
"metadata": {},
"outputs": [],
"source": [
"MAX_NUMBER_OF_DOCUMENTS = 300"
]
},
{
"cell_type": "markdown",
"id": "8edafda4",
"metadata": {},
"source": [
"## Loading Data and Measurs"
]
},
{
"cell_type": "markdown",
"id": "639f75cf",
"metadata": {},
"source": [
"## Loading JSON data"
]
},
{
"cell_type": "code",
"execution_count": 3,
"id": "1a2f9775",
"metadata": {},
"outputs": [],
"source": [
"import json\n",
"with open('Step1_output_list_of_processed_measures.json', encoding=\"utf8\") as f:\n",
" list_of_processed_measures = json.load(f)\n",
" \n",
"list_of_original_measures = [l['output_original'] for l in list_of_processed_measures]\n",
"list_of_opsitive_measures = [l['output_opposite'] for l in list_of_processed_measures]"
]
},
{
"cell_type": "markdown",
"id": "1c94123f",
"metadata": {},
"source": [
"## Analyzer class and implementation"
]
},
{
"cell_type": "code",
"execution_count": 4,
"id": "b6025d8f",
"metadata": {},
"outputs": [],
"source": [
"import joblib\n",
"import time\n",
"import numpy as np\n",
"\n",
"# positive\n",
"PERSONALITY = \"\"\"\n",
"شما یک تحلیلگر علمی و بی‌طرف هستید که در حوزه تحلیل محتوای کیفی تخصص دارید. وظیفه شما سنجش میزان رعایت یک معیار مشخص در یک متن علمی است.\n",
"\n",
"شما بر اساس یک «معیار رفتاری یا محتوایی» که به شما داده می‌شود و یک «متن علمی»، بررسی می‌کنید که تا چه حد این متن با معیار مذکور هماهنگ است. ارزیابی شما بدون تعصب، منصفانه، دقیق و تحلیلی است.\n",
"\n",
"شما در تحلیل خود:\n",
"- ابتدا اجزای متن علمی را با مؤلفه‌های معیار تطبیق می‌دهید.\n",
"- در صورت وجود تطابق یا عدم تطابق، به صورت شفاف و با مثال یا دلایل علمی تحلیل می‌کنید.\n",
"- از قضاوت ارزشی پرهیز می‌کنید و لحن تحلیلی و عینی دارید.\n",
"\"\"\"\n",
"\n",
"INTRO_QUARY = \"\"\"\n",
"اکنون لطفاً متن زیر را بر اساس معیار ارائه‌شده ارزیابی کن. تحلیل خود را با ساختاری علمی، دقیق و بی‌طرف بنویس.\n",
"معیار مورد سنجش:\n",
"\n",
"\"\"\"\n",
" \n",
"QUESTION = \"\"\"\n",
"\n",
"شما در تحلیل خود:\n",
"- ابتدا اجزای متن علمی را با مؤلفه‌های معیار تطبیق می‌دهید.\n",
"- در صورت وجود تطابق یا عدم تطابق، به صورت شفاف و با مثال یا دلایل علمی تحلیل می‌کنید.\n",
"- از قضاوت ارزشی پرهیز می‌کنید و لحن تحلیلی و عینی دارید.\n",
"\"\"\"\n",
"\n",
"\n",
"from LLMInterface import LLMInterface\n",
"class ValueAnalyser(LLMInterface):\n",
" def __init__(self, list_of_measures, personality=PERSONALITY, number_of_jobs=64, number_of_meaures_to_valuate=64):\n",
" super().__init__(personality=personality)\n",
" self.number_of_jobs = number_of_jobs\n",
"\n",
" self.list_of_measures = list_of_measures\n",
" self.list_of_measures = self.list_of_measures[:number_of_meaures_to_valuate]\n",
" \n",
" \n",
" def convert_to_message(self, measure, query):\n",
" intro_text = 'حالا به متن توجه کن: '\n",
" return INTRO_QUARY +'\\n'+ measure +'\\n'+ intro_text +'\\n'+ query +'\\n'+ QUESTION\n",
"\n",
" def get_text_analysis(self, text):\n",
" def func(i):\n",
" measure = self.list_of_measures[i]\n",
" query = self.convert_to_message(measure, text)\n",
" output = self.get_reponse(query)\n",
" return (measure, query, output, -1.)\n",
"\n",
" with joblib.parallel_backend(\"threading\"):\n",
" output_analysis = [joblib.Parallel(n_jobs=self.number_of_jobs)(joblib.delayed(func)(i) for i in range(len(self.list_of_measures)))]\n",
" return output_analysis"
]
},
{
"cell_type": "code",
"execution_count": 6,
"id": "afbf2718",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Resuming from paper index 294 (0‑based)\n",
"Processed 295/300 – بررسی تهدیدها و شناسایی فرصت‌های محیطی صنعت لوازم خانگی در شرایط تحریم اقتصادی، در جهت تبیین مدل اقتصاد مقاومتی (با رویکرد مدیریت استراتژیک)\n",
"Processed 296/300 – ارائه مدل مدیریت تکنولوژی در شرایط اقتصاد مقاومتی در صنایع با تکنولوژی پیچیده؛ مطالعه موردی: صنعت مخابرات\n",
"Processed 297/300 – بررسی فرصت های ناشی از کلان روند های اقتصاد دیجیتال به منظور تدوین الگوی توسعه صنعتی ایران در راستای سیاست های کلی اقتصاد مقاومتی\n",
"Processed 298/300 – نقش و کارکرد اداره امور مالیاتی در راستای اصلاح نظام درآمدی دولت با تأکید بر افزایش سهم درآمدهای مالیاتی مطرح در سیاست‌های کلی اقتصاد مقاومتی\n",
"Processed 299/300 – بررسی عدالت اقتصادی از منظر قرآن(در راستای تحقق اقتصاد مقاومتی)\n",
"Processed 300/300 – نقش دولت در مدیریت واردات در راستای تحقق سیاست‌های کلی اقتصاد مقاومتی\n",
"Processing complete.\n"
]
}
],
"source": [
"import json\n",
"import time\n",
"from typing import List, Dict, Any\n",
"\n",
"analyzer = ValueAnalyser(list_of_measures=list_of_original_measures)\n",
"\n",
"# Load the source data\n",
"with open('Results.json', encoding='utf8') as f:\n",
" data = json.load(f)\n",
"\n",
"TOTAL_PAPERS = len(data['results'])\n",
"\n",
"CHECKPOINT_FILE = 'checkpoint.txt'\n",
"OUTPUT_FILE = 'step2-original.json'\n",
"\n",
"def load_checkpoint() -> int:\n",
" \"\"\"Return the last successfully processed index (0‑based), or 0 if none.\"\"\"\n",
" try:\n",
" with open(CHECKPOINT_FILE, 'r') as f:\n",
" return int(f.read().strip())\n",
" except FileNotFoundError:\n",
" return 0\n",
"\n",
"def save_checkpoint(index: int):\n",
" \"\"\"Save the last completed index.\"\"\"\n",
" with open(CHECKPOINT_FILE, 'w') as f:\n",
" f.write(str(index))\n",
"\n",
"def load_existing_results() -> List[Dict[str, Any]]:\n",
" \"\"\"Load previously saved results, or return empty list if file is missing/corrupt.\"\"\"\n",
" try:\n",
" with open(OUTPUT_FILE, 'r', encoding='utf8') as f:\n",
" return json.load(f)\n",
" except (FileNotFoundError, json.JSONDecodeError):\n",
" return []\n",
"\n",
"def save_results(results: List[Dict[str, Any]]):\n",
" \"\"\"Overwrite the output file with the full list of results.\"\"\"\n",
" with open(OUTPUT_FILE, 'w', encoding='utf8') as f:\n",
" json.dump(results, f, ensure_ascii=False, indent=2)\n",
"\n",
"# Resume from last checkpoint\n",
"start_index = load_checkpoint()\n",
"all_results = load_existing_results()\n",
"\n",
"# If you want to process a fixed range (e.g., 10 to 20), adjust start_index accordingly\n",
"START = 0\n",
"END = TOTAL_PAPERS # or 20, or any limit\n",
"\n",
"if start_index < START:\n",
" start_index = START\n",
"\n",
"print(f\"Resuming from paper index {start_index} (0‑based)\")\n",
"\n",
"# Process each remaining paper\n",
"for i in range(start_index, END):\n",
" article = data['results'][i]\n",
" retries = 3\n",
" success = False\n",
"\n",
" for attempt in range(retries):\n",
" try:\n",
" # Simulate network call – your actual analysis may involve HTTP requests\n",
" res = analyzer.get_text_analysis('\\n'.join(article))\n",
" \n",
" result_entry = {\n",
" \"title\": article['title'],\n",
" \"year\": article[\"year\"],\n",
" \"grade\": article[\"grade\"],\n",
" \"authors\": article[\"authors\"],\n",
" \"universities\": article[\"universities\"],\n",
" 'link': article['link'],\n",
" 'abstract': article['abstract'],\n",
" \"response\": res\n",
" }\n",
" \n",
" all_results.append(result_entry)\n",
" save_results(all_results) # save after each success\n",
" save_checkpoint(i + 1) # next index to start from\n",
" print(f\"Processed {i+1}/{END} – {article['title']}\")\n",
" success = True\n",
" break # exit retry loop\n",
" \n",
" except Exception as e:\n",
" print(f\"Error on paper {i} (attempt {attempt+1}/{retries}): {e}\")\n",
" if attempt < retries - 1:\n",
" time.sleep(5) # wait before retry\n",
" else:\n",
" print(f\"Failed to process paper {i} after {retries} attempts. Skipping.\")\n",
" # Optionally, save checkpoint as i+1 to skip this paper entirely\n",
" save_checkpoint(i + 1)\n",
" \n",
" if not success:\n",
" # Log the failure but continue with next paper\n",
" continue\n",
" \n",
" # Small delay between papers to avoid overwhelming the server\n",
" time.sleep(1)\n",
"\n",
"print(\"Processing complete.\")"
]
},
{
"cell_type": "code",
"execution_count": 7,
"id": "baad4652-cc0b-4f0b-8e89-c1ec8bc95b3d",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"300\n"
]
}
],
"source": [
"print(TOTAL_PAPERS)"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "f8764ea0-1301-4696-982d-e5f6120e959c",
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3 (ipykernel)",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.12.13"
}
},
"nbformat": 4,
"nbformat_minor": 5
}