{
 "cells": [
  {
   "cell_type": "code",
   "id": "536d2f280afb463a",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2025-04-10T06:29:09.658974Z",
     "start_time": "2025-04-10T06:29:09.655915Z"
    }
   },
   "source": [
    "import pandas as pd\n",
    "from sklearn.feature_extraction.text import TfidfVectorizer\n",
    "from sklearn.metrics.pairwise import cosine_similarity\n",
    "import matplotlib.pyplot as plt"
   ],
   "outputs": [],
   "execution_count": 10
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2025-04-10T06:29:09.733105Z",
     "start_time": "2025-04-10T06:29:09.664030Z"
    }
   },
   "cell_type": "code",
   "source": [
    "# Step 1: Load Data\n",
    "bugs_df = pd.read_csv(\"cassandra/cassandra_bugs.csv\")\n",
    "duplicates_df = pd.read_csv(\"cassandra/cassandra_bugs-combined.csv\")"
   ],
   "id": "3114abe328d17fc7",
   "outputs": [],
   "execution_count": 11
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2025-04-10T06:29:09.754972Z",
     "start_time": "2025-04-10T06:29:09.740325Z"
    }
   },
   "cell_type": "code",
   "source": [
    "# Step 2: Preprocessing\n",
    "# Merge Summary and Description\n",
    "bugs_df[\"text\"] = bugs_df[[\"Summary\", \"Description\"]].fillna(\"\").agg(\" \".join, axis=1)\n",
    "\n",
    "# Map issue ID to text\n",
    "issue_text_map = dict(zip(bugs_df[\"Issue id\"], bugs_df[\"text\"]))\n",
    "\n",
    "# Clean duplicate ID column: handle comma-separated IDs\n",
    "duplicates_df[\"Duplicate id\"] = duplicates_df[\"Duplicate id\"].astype(str).str.split(r\",\\s*\")\n",
    "duplicates_df = duplicates_df.explode(\"Duplicate id\")\n",
    "\n",
    "# Convert to integers and drop invalid rows\n",
    "duplicates_df[\"Issue id\"] = pd.to_numeric(duplicates_df[\"Issue id\"], errors='coerce')\n",
    "duplicates_df[\"Duplicate id\"] = pd.to_numeric(duplicates_df[\"Duplicate id\"], errors='coerce')\n",
    "duplicates_df = duplicates_df.dropna().astype(int)\n",
    "\n",
    "# Filter for valid IDs present in text map\n",
    "duplicates_df = duplicates_df[\n",
    "    (duplicates_df[\"Issue id\"].isin(issue_text_map)) &\n",
    "    (duplicates_df[\"Duplicate id\"].isin(issue_text_map))\n",
    "]"
   ],
   "id": "3b450db57ab204bf",
   "outputs": [],
   "execution_count": 12
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2025-04-10T06:29:10.006779Z",
     "start_time": "2025-04-10T06:29:09.761677Z"
    }
   },
   "cell_type": "code",
   "source": [
    "# Step 3: TF-IDF Vectorization\n",
    "all_ids = list(issue_text_map.keys())\n",
    "all_texts = [issue_text_map[iid] for iid in all_ids]\n",
    "\n",
    "vectorizer = TfidfVectorizer(stop_words=\"english\", max_features=10000)\n",
    "tfidf_matrix = vectorizer.fit_transform(all_texts)\n",
    "\n",
    "# Index mapping\n",
    "id_to_index = {iid: idx for idx, iid in enumerate(all_ids)}"
   ],
   "id": "d6fdcb8b7717a316",
   "outputs": [],
   "execution_count": 13
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2025-04-10T06:29:10.425883Z",
     "start_time": "2025-04-10T06:29:10.015635Z"
    }
   },
   "cell_type": "code",
   "source": [
    "# Step 4: IR-based Retrieval\n",
    "results = []\n",
    "\n",
    "for _, row in duplicates_df.iterrows():\n",
    "    qid = row[\"Issue id\"]\n",
    "    true_duplicate = row[\"Duplicate id\"]\n",
    "\n",
    "    if qid not in id_to_index or true_duplicate not in id_to_index:\n",
    "        continue\n",
    "\n",
    "    q_idx = id_to_index[qid]\n",
    "    q_vec = tfidf_matrix[q_idx]\n",
    "\n",
    "    similarities = cosine_similarity(q_vec, tfidf_matrix).flatten()\n",
    "    top_indices = similarities.argsort()[::-1][1:11]  # Exclude self\n",
    "    top_ids = [all_ids[i] for i in top_indices]\n",
    "    top_scores = similarities[top_indices]\n",
    "\n",
    "    found = true_duplicate in top_ids\n",
    "\n",
    "    results.append({\n",
    "        \"query_id\": qid,\n",
    "        \"true_duplicate\": true_duplicate,\n",
    "        \"found_in_top_10\": found,\n",
    "        \"top_10_ranked_ids\": top_ids,\n",
    "        \"top_10_scores\": top_scores.tolist()\n",
    "    })\n"
   ],
   "id": "90fbf1690fdd9326",
   "outputs": [],
   "execution_count": 14
  },
  {
   "metadata": {
    "collapsed": true,
    "ExecuteTime": {
     "end_time": "2025-04-10T06:29:10.484890Z",
     "start_time": "2025-04-10T06:29:10.432510Z"
    }
   },
   "cell_type": "code",
   "source": [
    "# Step 5: Evaluate Recall@10\n",
    "recall_at_10 = sum(r[\"found_in_top_10\"] for r in results) / len(results)\n",
    "print(f\"Recall@10: {recall_at_10:.2f} over {len(results)} queries\")\n",
    "\n",
    "# Step 6: Score Distribution Plot\n",
    "all_scores = [score for r in results for score in r[\"top_10_scores\"]]\n",
    "\n",
    "plt.figure(figsize=(8, 5))\n",
    "plt.hist(all_scores, bins=20, edgecolor='black')\n",
    "plt.title(\"Distribution of Top-10 Cosine Similarity Scores\")\n",
    "plt.xlabel(\"Cosine Similarity\")\n",
    "plt.ylabel(\"Frequency\")\n",
    "plt.grid(True)\n",
    "plt.tight_layout()\n",
    "plt.show()\n"
   ],
   "id": "initial_id",
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Recall@10: 0.61 over 300 queries\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 800x500 with 1 Axes>"
      ],
      "image/png": "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"
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "execution_count": 15
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}