基于PHP的Boolean Retrieval（Binary Operation）实现及结果格式需求咨询

Let's refine your existing code to handle the Boolean query "One AND Two AND NOT Three" and output results in the requested "Term Doc..." format. Here's a step-by-step solution tailored to your needs:

Step 1: Preprocess the Corpus #

First, we'll clean up document processing to create a clear mapping of document IDs to their unique, lowercase terms — this makes retrieval faster and avoids case sensitivity issues.

Step 2: Parse & Map Query Terms #

We’ll break down your query into terms and operators, then build a lookup table showing which documents contain each term.

Step 3: Execute Boolean Logic #

We’ll compute the intersection of documents containing "One" and "Two", then exclude any documents that include "Three".

Full Working Code #

<?php
$path = "Korpus";

// 1. Build a map of document IDs to their unique, lowercase terms
$docTerms = [];
foreach (glob("$path/Doc*.txt") as $file) {
    // Extract document ID from filename (e.g., Doc1.txt → 1)
    preg_match('/Doc(\d+)\.txt/', $file, $matches);
    $docId = $matches[1];
    
    // Process content: lowercase, split by spaces, remove duplicates/empty strings
    $content = strtolower(trim(file_get_contents($file)));
    $terms = array_unique(explode(" ", $content));
    $terms = array_filter($terms, fn($term) => !empty($term));
    
    $docTerms[$docId] = $terms;
}

// 2. Parse the Boolean query into terms and operators
$query = "One AND Two AND NOT Three";
$tokens = array_filter(explode(" ", strtolower($query)));

$terms = [];
foreach ($tokens as $token) {
    if (!in_array(strtoupper($token), ['AND', 'OR', 'NOT'])) {
        $terms[] = $token;
    }
}

// 3. Map each query term to the documents that contain it
$termToDocs = [];
foreach ($terms as $term) {
    $termToDocs[$term] = [];
    foreach ($docTerms as $docId => $docTermsList) {
        if (in_array($term, $docTermsList)) {
            $termToDocs[$term][] = $docId;
        }
    }
}

// 4. Execute the query logic: (One AND Two) AND NOT Three
$andResult = array_intersect($termToDocs['one'], $termToDocs['two']);
$finalResult = array_diff($andResult, $termToDocs['three']);

// 5. Output in the requested format
echo "Term\tDocuments\n";
echo "------------------------\n";
foreach ($termToDocs as $term => $docs) {
    echo ucfirst($term) . "\t" . (empty($docs) ? "None" : implode(", ", $docs)) . "\n";
}

echo "\nQuery: \"One AND Two AND NOT Three\"\n";
echo "Matching Documents: " . (empty($finalResult) ? "None" : implode(", ", $finalResult)) . "\n";
?>

Key Improvements & Notes #

• Case Insensitivity: All text is converted to lowercase so "One" and "one" are treated as the same term.
• Cleaner Data: Duplicate terms and empty strings (from extra spaces) are filtered out to avoid retrieval errors.
• Readable Output: First prints each term with its associated documents, then the final query result in a clear format.
• Flexibility: If your corpus uses different filename patterns or needs punctuation removal, you can adjust the regex or add a step to strip special characters from content.

内容的提问来源于stack exchange，提问作者lowspec