| 1 |
<?php |
| 2 |
/** |
| 3 |
* AI Content Analyzer Class |
| 4 |
* |
| 5 |
* Comprehensive AI-powered content analysis with real algorithms for readability, |
| 6 |
* keyword density, semantic analysis, and content optimization. Implements 2025 |
| 7 |
* SEO best practices with industry-standard calculations and recommendations. |
| 8 |
* |
| 9 |
* @package ThinkRank |
| 10 |
* @subpackage SEO |
| 11 |
* @since 1.0.0 |
| 12 |
*/ |
| 13 |
|
| 14 |
declare(strict_types=1); |
| 15 |
|
| 16 |
namespace ThinkRank\SEO; |
| 17 |
|
| 18 |
/** |
| 19 |
* AI Content Analyzer Class |
| 20 |
* |
| 21 |
* Provides comprehensive content analysis using real AI algorithms including |
| 22 |
* readability scoring, keyword analysis, semantic content analysis, and |
| 23 |
* structure optimization with actionable recommendations. |
| 24 |
* |
| 25 |
* @since 1.0.0 |
| 26 |
*/ |
| 27 |
class AI_Content_Analyzer extends Abstract_SEO_Manager { |
| 28 |
|
| 29 |
/** |
| 30 |
* Content analysis algorithms with their specifications |
| 31 |
* |
| 32 |
* @since 1.0.0 |
| 33 |
* @var array |
| 34 |
*/ |
| 35 |
private array $analysis_algorithms = [ |
| 36 |
'readability' => [ |
| 37 |
'flesch_reading_ease' => [ |
| 38 |
'name' => 'Flesch Reading Ease', |
| 39 |
'formula' => '206.835 - (1.015 × ASL) - (84.6 × ASW)', |
| 40 |
'scale' => [90 => 'Very Easy', 80 => 'Easy', 70 => 'Fairly Easy', 60 => 'Standard', 50 => 'Fairly Difficult', 30 => 'Difficult', 0 => 'Very Difficult'], |
| 41 |
'weight' => 40 |
| 42 |
], |
| 43 |
'flesch_kincaid_grade' => [ |
| 44 |
'name' => 'Flesch-Kincaid Grade Level', |
| 45 |
'formula' => '(0.39 × ASL) + (11.8 × ASW) - 15.59', |
| 46 |
'target_range' => [6, 8], |
| 47 |
'weight' => 30 |
| 48 |
], |
| 49 |
'automated_readability_index' => [ |
| 50 |
'name' => 'Automated Readability Index', |
| 51 |
'formula' => '(4.71 × CPW) + (0.5 × SPW) - 21.43', |
| 52 |
'target_range' => [6, 10], |
| 53 |
'weight' => 30 |
| 54 |
] |
| 55 |
], |
| 56 |
'keyword_analysis' => [ |
| 57 |
'density_calculation' => [ |
| 58 |
'optimal_range' => [1.0, 3.0], // 1-3% keyword density |
| 59 |
'warning_threshold' => 5.0, // Above 5% is keyword stuffing |
| 60 |
'minimum_threshold' => 0.5 // Below 0.5% is under-optimized |
| 61 |
], |
| 62 |
'proximity_analysis' => [ |
| 63 |
'ideal_distance' => 100, // Words between keyword occurrences |
| 64 |
'clustering_threshold' => 50 // Words indicating keyword clustering |
| 65 |
], |
| 66 |
'semantic_variations' => [ |
| 67 |
'synonym_weight' => 0.7, |
| 68 |
'related_terms_weight' => 0.5, |
| 69 |
'lsi_keywords_weight' => 0.8 |
| 70 |
] |
| 71 |
], |
| 72 |
'content_structure' => [ |
| 73 |
'heading_hierarchy' => [ |
| 74 |
'h1_count' => ['min' => 1, 'max' => 1], |
| 75 |
'h2_count' => ['min' => 2, 'max' => 10], |
| 76 |
'h3_count' => ['min' => 0, 'max' => 20], |
| 77 |
'proper_nesting' => true |
| 78 |
], |
| 79 |
'paragraph_analysis' => [ |
| 80 |
'ideal_length' => [50, 150], // Words per paragraph |
| 81 |
'max_sentences' => 4, |
| 82 |
'transition_words_ratio' => 0.3 |
| 83 |
], |
| 84 |
'content_length' => [ |
| 85 |
'blog_post' => ['min' => 300, 'optimal' => 1500, 'max' => 3000], |
| 86 |
'product_page' => ['min' => 200, 'optimal' => 800, 'max' => 1500], |
| 87 |
'landing_page' => ['min' => 500, 'optimal' => 2000, 'max' => 4000] |
| 88 |
] |
| 89 |
] |
| 90 |
]; |
| 91 |
|
| 92 |
/** |
| 93 |
* Semantic analysis capabilities |
| 94 |
* |
| 95 |
* @since 1.0.0 |
| 96 |
* @var array |
| 97 |
*/ |
| 98 |
private array $semantic_capabilities = [ |
| 99 |
'topic_modeling' => [ |
| 100 |
'enabled' => true, |
| 101 |
'min_topics' => 3, |
| 102 |
'max_topics' => 10, |
| 103 |
'coherence_threshold' => 0.7 |
| 104 |
], |
| 105 |
'entity_recognition' => [ |
| 106 |
'enabled' => true, |
| 107 |
'entity_types' => ['PERSON', 'ORGANIZATION', 'LOCATION', 'PRODUCT', 'EVENT'], |
| 108 |
'confidence_threshold' => 0.8 |
| 109 |
], |
| 110 |
'sentiment_analysis' => [ |
| 111 |
'enabled' => true, |
| 112 |
'scale' => [-1.0, 1.0], // Negative to positive |
| 113 |
'neutral_range' => [-0.1, 0.1] |
| 114 |
], |
| 115 |
'content_classification' => [ |
| 116 |
'enabled' => true, |
| 117 |
'categories' => ['informational', 'commercial', 'transactional', 'navigational'], |
| 118 |
'confidence_threshold' => 0.6 |
| 119 |
] |
| 120 |
]; |
| 121 |
|
| 122 |
/** |
| 123 |
* Content optimization scoring weights |
| 124 |
* |
| 125 |
* @since 1.0.0 |
| 126 |
* @var array |
| 127 |
*/ |
| 128 |
private array $scoring_weights = [ |
| 129 |
'readability' => 25, |
| 130 |
'keyword_optimization' => 30, |
| 131 |
'content_structure' => 20, |
| 132 |
'semantic_relevance' => 15, |
| 133 |
'technical_seo' => 10 |
| 134 |
]; |
| 135 |
|
| 136 |
/** |
| 137 |
* Constructor |
| 138 |
* |
| 139 |
* @since 1.0.0 |
| 140 |
*/ |
| 141 |
public function __construct() { |
| 142 |
parent::__construct('ai_content_analyzer'); |
| 143 |
} |
| 144 |
|
| 145 |
/** |
| 146 |
* Analyze content with comprehensive AI-powered analysis |
| 147 |
* |
| 148 |
* @since 1.0.0 |
| 149 |
* |
| 150 |
* @param string $content Content to analyze |
| 151 |
* @param array $keywords Target keywords |
| 152 |
* @param array $options Analysis options |
| 153 |
* @return array Comprehensive analysis results |
| 154 |
*/ |
| 155 |
public function analyze_content(string $content, array $keywords = [], array $options = []): array { |
| 156 |
$analysis = [ |
| 157 |
'content_stats' => [], |
| 158 |
'readability' => [], |
| 159 |
'keyword_analysis' => [], |
| 160 |
'semantic_analysis' => [], |
| 161 |
'structure_analysis' => [], |
| 162 |
'optimization_score' => 0, |
| 163 |
'recommendations' => [], |
| 164 |
'ai_confidence' => 0, |
| 165 |
'analysis_timestamp' => current_time('mysql') |
| 166 |
]; |
| 167 |
|
| 168 |
// Basic content statistics |
| 169 |
$analysis['content_stats'] = $this->calculate_content_statistics($content); |
| 170 |
|
| 171 |
// Readability analysis with real algorithms |
| 172 |
$analysis['readability'] = $this->analyze_readability($content, $analysis['content_stats']); |
| 173 |
|
| 174 |
// Keyword analysis and optimization |
| 175 |
if (!empty($keywords)) { |
| 176 |
$analysis['keyword_analysis'] = $this->analyze_keywords($content, $keywords, $analysis['content_stats']); |
| 177 |
} |
| 178 |
|
| 179 |
// Semantic content analysis |
| 180 |
$analysis['semantic_analysis'] = $this->analyze_semantic_content($content, $options); |
| 181 |
|
| 182 |
// Content structure analysis |
| 183 |
$analysis['structure_analysis'] = $this->analyze_content_structure($content, $options); |
| 184 |
|
| 185 |
// Calculate overall optimization score |
| 186 |
$analysis['optimization_score'] = $this->calculate_optimization_score($analysis); |
| 187 |
|
| 188 |
// Generate AI-powered recommendations |
| 189 |
$analysis['recommendations'] = $this->generate_optimization_recommendations($analysis); |
| 190 |
|
| 191 |
// Calculate AI confidence score |
| 192 |
$analysis['ai_confidence'] = $this->calculate_ai_confidence($analysis); |
| 193 |
|
| 194 |
return $analysis; |
| 195 |
} |
| 196 |
|
| 197 |
/** |
| 198 |
* Analyze readability using real algorithms |
| 199 |
* |
| 200 |
* @since 1.0.0 |
| 201 |
* |
| 202 |
* @param string $content Content to analyze |
| 203 |
* @param array $stats Content statistics |
| 204 |
* @return array Readability analysis results |
| 205 |
*/ |
| 206 |
public function analyze_readability(string $content, array $stats): array { |
| 207 |
$readability = [ |
| 208 |
'flesch_reading_ease' => 0, |
| 209 |
'flesch_kincaid_grade' => 0, |
| 210 |
'automated_readability_index' => 0, |
| 211 |
'average_sentence_length' => 0, |
| 212 |
'average_syllables_per_word' => 0, |
| 213 |
'readability_score' => 0, |
| 214 |
'reading_level' => '', |
| 215 |
'recommendations' => [] |
| 216 |
]; |
| 217 |
|
| 218 |
if ($stats['word_count'] === 0 || $stats['sentence_count'] === 0) { |
| 219 |
return $readability; |
| 220 |
} |
| 221 |
|
| 222 |
// Calculate metrics |
| 223 |
$readability['average_sentence_length'] = $stats['word_count'] / $stats['sentence_count']; |
| 224 |
$readability['average_syllables_per_word'] = $stats['syllable_count'] / $stats['word_count']; |
| 225 |
$characters_per_word = $stats['character_count'] / $stats['word_count']; |
| 226 |
|
| 227 |
// Flesch Reading Ease (real algorithm) |
| 228 |
$readability['flesch_reading_ease'] = 206.835 |
| 229 |
- (1.015 * $readability['average_sentence_length']) |
| 230 |
- (84.6 * $readability['average_syllables_per_word']); |
| 231 |
|
| 232 |
// Flesch-Kincaid Grade Level (real algorithm) |
| 233 |
$readability['flesch_kincaid_grade'] = (0.39 * $readability['average_sentence_length']) |
| 234 |
+ (11.8 * $readability['average_syllables_per_word']) |
| 235 |
- 15.59; |
| 236 |
|
| 237 |
// Automated Readability Index (real algorithm) |
| 238 |
$readability['automated_readability_index'] = (4.71 * $characters_per_word) |
| 239 |
+ (0.5 * $readability['average_sentence_length']) |
| 240 |
- 21.43; |
| 241 |
|
| 242 |
// Determine reading level |
| 243 |
$readability['reading_level'] = $this->determine_reading_level($readability['flesch_reading_ease']); |
| 244 |
|
| 245 |
// Calculate overall readability score (0-100) |
| 246 |
$readability['readability_score'] = $this->calculate_readability_score($readability); |
| 247 |
|
| 248 |
// Generate readability recommendations |
| 249 |
$readability['recommendations'] = $this->generate_readability_recommendations($readability); |
| 250 |
|
| 251 |
return $readability; |
| 252 |
} |
| 253 |
|
| 254 |
/** |
| 255 |
* Analyze keywords with density and optimization metrics |
| 256 |
* |
| 257 |
* @since 1.0.0 |
| 258 |
* |
| 259 |
* @param string $content Content to analyze |
| 260 |
* @param array $keywords Target keywords |
| 261 |
* @param array $stats Content statistics |
| 262 |
* @return array Keyword analysis results |
| 263 |
*/ |
| 264 |
public function analyze_keywords(string $content, array $keywords, array $stats): array { |
| 265 |
$keyword_analysis = [ |
| 266 |
'primary_keywords' => [], |
| 267 |
'secondary_keywords' => [], |
| 268 |
'keyword_density' => [], |
| 269 |
'keyword_distribution' => [], |
| 270 |
'semantic_keywords' => [], |
| 271 |
'optimization_score' => 0, |
| 272 |
'recommendations' => [] |
| 273 |
]; |
| 274 |
|
| 275 |
$content_lower = strtolower($content); |
| 276 |
$words = str_word_count($content_lower, 1); |
| 277 |
|
| 278 |
foreach ($keywords as $index => $keyword) { |
| 279 |
$keyword_lower = strtolower($keyword); |
| 280 |
$keyword_data = [ |
| 281 |
'keyword' => $keyword, |
| 282 |
'occurrences' => 0, |
| 283 |
'density' => 0, |
| 284 |
'positions' => [], |
| 285 |
'distribution_score' => 0, |
| 286 |
'optimization_status' => 'under_optimized' |
| 287 |
]; |
| 288 |
|
| 289 |
// Count keyword occurrences |
| 290 |
$keyword_data['occurrences'] = substr_count($content_lower, $keyword_lower); |
| 291 |
|
| 292 |
// Calculate keyword density |
| 293 |
if ($stats['word_count'] > 0) { |
| 294 |
$keyword_data['density'] = ($keyword_data['occurrences'] / $stats['word_count']) * 100; |
| 295 |
} |
| 296 |
|
| 297 |
// Find keyword positions for distribution analysis |
| 298 |
$keyword_data['positions'] = $this->find_keyword_positions($content_lower, $keyword_lower); |
| 299 |
|
| 300 |
// Calculate distribution score |
| 301 |
$keyword_data['distribution_score'] = $this->calculate_keyword_distribution($keyword_data['positions'], $stats['word_count']); |
| 302 |
|
| 303 |
// Determine optimization status |
| 304 |
$keyword_data['optimization_status'] = $this->determine_keyword_optimization_status($keyword_data['density']); |
| 305 |
|
| 306 |
// Categorize as primary or secondary |
| 307 |
if ($index === 0 || $keyword_data['density'] >= 1.0) { |
| 308 |
$keyword_analysis['primary_keywords'][] = $keyword_data; |
| 309 |
} else { |
| 310 |
$keyword_analysis['secondary_keywords'][] = $keyword_data; |
| 311 |
} |
| 312 |
|
| 313 |
$keyword_analysis['keyword_density'][$keyword] = $keyword_data['density']; |
| 314 |
} |
| 315 |
|
| 316 |
// Find semantic keywords |
| 317 |
$keyword_analysis['semantic_keywords'] = $this->find_semantic_keywords($content, $keywords); |
| 318 |
|
| 319 |
// Calculate overall keyword optimization score |
| 320 |
$keyword_analysis['optimization_score'] = $this->calculate_keyword_optimization_score($keyword_analysis); |
| 321 |
|
| 322 |
// Generate keyword recommendations |
| 323 |
$keyword_analysis['recommendations'] = $this->generate_keyword_recommendations($keyword_analysis); |
| 324 |
|
| 325 |
return $keyword_analysis; |
| 326 |
} |
| 327 |
|
| 328 |
/** |
| 329 |
* Analyze semantic content with topic modeling and entity recognition |
| 330 |
* |
| 331 |
* @since 1.0.0 |
| 332 |
* |
| 333 |
* @param string $content Content to analyze |
| 334 |
* @param array $options Analysis options |
| 335 |
* @return array Semantic analysis results |
| 336 |
*/ |
| 337 |
public function analyze_semantic_content(string $content, array $options = []): array { |
| 338 |
$semantic_analysis = [ |
| 339 |
'topic_clusters' => [], |
| 340 |
'entities' => [], |
| 341 |
'sentiment' => [], |
| 342 |
'content_classification' => [], |
| 343 |
'semantic_keywords' => [], |
| 344 |
'coherence_score' => 0, |
| 345 |
'relevance_score' => 0 |
| 346 |
]; |
| 347 |
|
| 348 |
// Topic modeling and clustering |
| 349 |
if ($this->semantic_capabilities['topic_modeling']['enabled']) { |
| 350 |
$semantic_analysis['topic_clusters'] = $this->extract_topic_clusters($content); |
| 351 |
} |
| 352 |
|
| 353 |
// Named entity recognition |
| 354 |
if ($this->semantic_capabilities['entity_recognition']['enabled']) { |
| 355 |
$semantic_analysis['entities'] = $this->recognize_entities($content); |
| 356 |
} |
| 357 |
|
| 358 |
// Sentiment analysis |
| 359 |
if ($this->semantic_capabilities['sentiment_analysis']['enabled']) { |
| 360 |
$semantic_analysis['sentiment'] = $this->analyze_sentiment($content); |
| 361 |
} |
| 362 |
|
| 363 |
// Content classification |
| 364 |
if ($this->semantic_capabilities['content_classification']['enabled']) { |
| 365 |
$semantic_analysis['content_classification'] = $this->classify_content($content); |
| 366 |
} |
| 367 |
|
| 368 |
// Extract semantic keywords |
| 369 |
$semantic_analysis['semantic_keywords'] = $this->extract_semantic_keywords($content); |
| 370 |
|
| 371 |
// Calculate coherence score |
| 372 |
$semantic_analysis['coherence_score'] = $this->calculate_content_coherence($semantic_analysis); |
| 373 |
|
| 374 |
// Calculate relevance score |
| 375 |
$semantic_analysis['relevance_score'] = $this->calculate_semantic_relevance($semantic_analysis); |
| 376 |
|
| 377 |
return $semantic_analysis; |
| 378 |
} |
| 379 |
|
| 380 |
/** |
| 381 |
* Analyze content structure and organization |
| 382 |
* |
| 383 |
* @since 1.0.0 |
| 384 |
* |
| 385 |
* @param string $content Content to analyze |
| 386 |
* @param array $options Analysis options |
| 387 |
* @return array Structure analysis results |
| 388 |
*/ |
| 389 |
public function analyze_content_structure(string $content, array $options = []): array { |
| 390 |
$structure_analysis = [ |
| 391 |
'heading_structure' => [], |
| 392 |
'paragraph_analysis' => [], |
| 393 |
'list_usage' => [], |
| 394 |
'image_analysis' => [], |
| 395 |
'link_analysis' => [], |
| 396 |
'content_flow' => [], |
| 397 |
'structure_score' => 0, |
| 398 |
'recommendations' => [] |
| 399 |
]; |
| 400 |
|
| 401 |
// Analyze heading hierarchy |
| 402 |
$structure_analysis['heading_structure'] = $this->analyze_heading_structure($content); |
| 403 |
|
| 404 |
// Analyze paragraph structure |
| 405 |
$structure_analysis['paragraph_analysis'] = $this->analyze_paragraph_structure($content); |
| 406 |
|
| 407 |
// Analyze list usage |
| 408 |
$structure_analysis['list_usage'] = $this->analyze_list_usage($content); |
| 409 |
|
| 410 |
// Analyze images |
| 411 |
$structure_analysis['image_analysis'] = $this->analyze_image_usage($content); |
| 412 |
|
| 413 |
// Analyze internal and external links |
| 414 |
$structure_analysis['link_analysis'] = $this->analyze_link_structure($content); |
| 415 |
|
| 416 |
// Analyze content flow and transitions |
| 417 |
$structure_analysis['content_flow'] = $this->analyze_content_flow($content); |
| 418 |
|
| 419 |
// Calculate structure score |
| 420 |
$structure_analysis['structure_score'] = $this->calculate_structure_score($structure_analysis); |
| 421 |
|
| 422 |
// Generate structure recommendations |
| 423 |
$structure_analysis['recommendations'] = $this->generate_structure_recommendations($structure_analysis); |
| 424 |
|
| 425 |
return $structure_analysis; |
| 426 |
} |
| 427 |
|
| 428 |
/** |
| 429 |
* Validate SEO settings (implements interface) |
| 430 |
* |
| 431 |
* @since 1.0.0 |
| 432 |
* |
| 433 |
* @param array $settings Settings array to validate |
| 434 |
* @return array Validation results |
| 435 |
*/ |
| 436 |
public function validate_settings(array $settings): array { |
| 437 |
$validation = [ |
| 438 |
'valid' => true, |
| 439 |
'errors' => [], |
| 440 |
'warnings' => [], |
| 441 |
'suggestions' => [], |
| 442 |
'score' => 100 |
| 443 |
]; |
| 444 |
|
| 445 |
// Validate analysis algorithms settings |
| 446 |
if (isset($settings['readability_enabled']) && !is_bool($settings['readability_enabled'])) { |
| 447 |
$validation['errors'][] = 'Readability analysis setting must be boolean'; |
| 448 |
$validation['valid'] = false; |
| 449 |
} |
| 450 |
|
| 451 |
if (isset($settings['keyword_analysis_enabled']) && !is_bool($settings['keyword_analysis_enabled'])) { |
| 452 |
$validation['errors'][] = 'Keyword analysis setting must be boolean'; |
| 453 |
$validation['valid'] = false; |
| 454 |
} |
| 455 |
|
| 456 |
if (isset($settings['semantic_analysis_enabled']) && !is_bool($settings['semantic_analysis_enabled'])) { |
| 457 |
$validation['errors'][] = 'Semantic analysis setting must be boolean'; |
| 458 |
$validation['valid'] = false; |
| 459 |
} |
| 460 |
|
| 461 |
// Validate scoring weights |
| 462 |
if (isset($settings['scoring_weights']) && is_array($settings['scoring_weights'])) { |
| 463 |
$total_weight = array_sum($settings['scoring_weights']); |
| 464 |
if ($total_weight !== 100) { |
| 465 |
$validation['warnings'][] = 'Scoring weights should total 100%'; |
| 466 |
} |
| 467 |
} |
| 468 |
|
| 469 |
// Validate keyword density thresholds |
| 470 |
if (isset($settings['keyword_density_threshold'])) { |
| 471 |
$threshold = $settings['keyword_density_threshold']; |
| 472 |
if (!is_numeric($threshold) || $threshold < 0 || $threshold > 10) { |
| 473 |
$validation['errors'][] = 'Keyword density threshold must be between 0 and 10'; |
| 474 |
$validation['valid'] = false; |
| 475 |
} |
| 476 |
} |
| 477 |
|
| 478 |
// Validate readability targets |
| 479 |
if (isset($settings['target_reading_level'])) { |
| 480 |
$valid_levels = ['elementary', 'middle_school', 'high_school', 'college', 'graduate']; |
| 481 |
if (!in_array($settings['target_reading_level'], $valid_levels, true)) { |
| 482 |
$validation['errors'][] = 'Invalid target reading level specified'; |
| 483 |
$validation['valid'] = false; |
| 484 |
} |
| 485 |
} |
| 486 |
|
| 487 |
// Calculate validation score |
| 488 |
$validation['score'] = $this->calculate_validation_score($validation); |
| 489 |
|
| 490 |
return $validation; |
| 491 |
} |
| 492 |
|
| 493 |
/** |
| 494 |
* Get output data for frontend rendering (implements interface) |
| 495 |
* |
| 496 |
* @since 1.0.0 |
| 497 |
* |
| 498 |
* @param string $context_type The context type |
| 499 |
* @param int|null $context_id Optional. Context ID |
| 500 |
* @return array Output data ready for frontend rendering |
| 501 |
*/ |
| 502 |
public function get_output_data(string $context_type, ?int $context_id): array { |
| 503 |
$settings = $this->get_settings($context_type, $context_id); |
| 504 |
|
| 505 |
$output = [ |
| 506 |
'analysis_results' => [], |
| 507 |
'recommendations' => [], |
| 508 |
'optimization_score' => 0, |
| 509 |
'enabled' => $settings['enabled'] ?? true |
| 510 |
]; |
| 511 |
|
| 512 |
if (!$output['enabled']) { |
| 513 |
return $output; |
| 514 |
} |
| 515 |
|
| 516 |
// Get content for analysis |
| 517 |
$content = $this->extract_content_for_analysis($context_type, $context_id); |
| 518 |
$keywords = $this->extract_target_keywords($context_type, $context_id); |
| 519 |
|
| 520 |
if (!empty($content)) { |
| 521 |
// Perform comprehensive analysis |
| 522 |
$analysis_options = [ |
| 523 |
'readability_enabled' => $settings['readability_enabled'] ?? true, |
| 524 |
'keyword_analysis_enabled' => $settings['keyword_analysis_enabled'] ?? true, |
| 525 |
'semantic_analysis_enabled' => $settings['semantic_analysis_enabled'] ?? true, |
| 526 |
'structure_analysis_enabled' => $settings['structure_analysis_enabled'] ?? true |
| 527 |
]; |
| 528 |
|
| 529 |
$output['analysis_results'] = $this->analyze_content($content, $keywords, $analysis_options); |
| 530 |
$output['recommendations'] = $output['analysis_results']['recommendations'] ?? []; |
| 531 |
$output['optimization_score'] = $output['analysis_results']['optimization_score'] ?? 0; |
| 532 |
|
| 533 |
// Store analysis results in database |
| 534 |
$this->store_analysis_results($context_type, $context_id, $output['analysis_results']); |
| 535 |
} |
| 536 |
|
| 537 |
return $output; |
| 538 |
} |
| 539 |
|
| 540 |
/** |
| 541 |
* Get default settings for a context type (implements interface) |
| 542 |
* |
| 543 |
* @since 1.0.0 |
| 544 |
* |
| 545 |
* @param string $context_type The context type to get defaults for |
| 546 |
* @return array Default settings array |
| 547 |
*/ |
| 548 |
public function get_default_settings(string $context_type): array { |
| 549 |
$defaults = [ |
| 550 |
'enabled' => true, |
| 551 |
'readability_enabled' => true, |
| 552 |
'keyword_analysis_enabled' => true, |
| 553 |
'semantic_analysis_enabled' => true, |
| 554 |
'structure_analysis_enabled' => true, |
| 555 |
'auto_analyze' => true, |
| 556 |
'target_reading_level' => 'high_school', |
| 557 |
'keyword_density_threshold' => 3.0, |
| 558 |
'min_content_length' => 300, |
| 559 |
'scoring_weights' => $this->scoring_weights |
| 560 |
]; |
| 561 |
|
| 562 |
// Context-specific defaults |
| 563 |
switch ($context_type) { |
| 564 |
case 'post': |
| 565 |
$defaults['min_content_length'] = 300; |
| 566 |
$defaults['target_reading_level'] = 'high_school'; |
| 567 |
break; |
| 568 |
case 'page': |
| 569 |
$defaults['min_content_length'] = 200; |
| 570 |
$defaults['target_reading_level'] = 'middle_school'; |
| 571 |
break; |
| 572 |
case 'product': |
| 573 |
$defaults['min_content_length'] = 150; |
| 574 |
$defaults['target_reading_level'] = 'middle_school'; |
| 575 |
$defaults['keyword_density_threshold'] = 2.5; |
| 576 |
break; |
| 577 |
} |
| 578 |
|
| 579 |
return $defaults; |
| 580 |
} |
| 581 |
|
| 582 |
/** |
| 583 |
* Get settings schema definition (implements interface) |
| 584 |
* |
| 585 |
* @since 1.0.0 |
| 586 |
* |
| 587 |
* @param string $context_type The context type to get schema for |
| 588 |
* @return array Settings schema definition |
| 589 |
*/ |
| 590 |
public function get_settings_schema(string $context_type): array { |
| 591 |
return [ |
| 592 |
'enabled' => [ |
| 593 |
'type' => 'boolean', |
| 594 |
'title' => 'Enable AI Content Analysis', |
| 595 |
'description' => 'Enable AI-powered content analysis and optimization', |
| 596 |
'default' => true |
| 597 |
], |
| 598 |
'readability_enabled' => [ |
| 599 |
'type' => 'boolean', |
| 600 |
'title' => 'Enable Readability Analysis', |
| 601 |
'description' => 'Analyze content readability using Flesch-Kincaid algorithms', |
| 602 |
'default' => true |
| 603 |
], |
| 604 |
'keyword_analysis_enabled' => [ |
| 605 |
'type' => 'boolean', |
| 606 |
'title' => 'Enable Keyword Analysis', |
| 607 |
'description' => 'Analyze keyword density and optimization', |
| 608 |
'default' => true |
| 609 |
], |
| 610 |
'semantic_analysis_enabled' => [ |
| 611 |
'type' => 'boolean', |
| 612 |
'title' => 'Enable Semantic Analysis', |
| 613 |
'description' => 'Perform topic modeling and entity recognition', |
| 614 |
'default' => true |
| 615 |
], |
| 616 |
'structure_analysis_enabled' => [ |
| 617 |
'type' => 'boolean', |
| 618 |
'title' => 'Enable Structure Analysis', |
| 619 |
'description' => 'Analyze content structure and organization', |
| 620 |
'default' => true |
| 621 |
], |
| 622 |
'target_reading_level' => [ |
| 623 |
'type' => 'string', |
| 624 |
'title' => 'Target Reading Level', |
| 625 |
'description' => 'Target reading level for content optimization', |
| 626 |
'enum' => ['elementary', 'middle_school', 'high_school', 'college', 'graduate'], |
| 627 |
'default' => 'high_school' |
| 628 |
], |
| 629 |
'keyword_density_threshold' => [ |
| 630 |
'type' => 'number', |
| 631 |
'title' => 'Keyword Density Threshold', |
| 632 |
'description' => 'Maximum keyword density percentage (1-10%)', |
| 633 |
'minimum' => 1, |
| 634 |
'maximum' => 10, |
| 635 |
'default' => 3.0 |
| 636 |
], |
| 637 |
'min_content_length' => [ |
| 638 |
'type' => 'integer', |
| 639 |
'title' => 'Minimum Content Length', |
| 640 |
'description' => 'Minimum word count for content analysis', |
| 641 |
'minimum' => 50, |
| 642 |
'default' => 300 |
| 643 |
] |
| 644 |
]; |
| 645 |
} |
| 646 |
|
| 647 |
/** |
| 648 |
* Calculate content statistics |
| 649 |
* |
| 650 |
* @since 1.0.0 |
| 651 |
* |
| 652 |
* @param string $content Content to analyze |
| 653 |
* @return array Content statistics |
| 654 |
*/ |
| 655 |
private function calculate_content_statistics(string $content): array { |
| 656 |
$clean_content = wp_strip_all_tags($content); |
| 657 |
|
| 658 |
$stats = [ |
| 659 |
'character_count' => strlen($clean_content), |
| 660 |
'character_count_no_spaces' => strlen(str_replace(' ', '', $clean_content)), |
| 661 |
'word_count' => str_word_count($clean_content), |
| 662 |
'sentence_count' => 0, |
| 663 |
'paragraph_count' => 0, |
| 664 |
'syllable_count' => 0, |
| 665 |
'complex_words' => 0, |
| 666 |
'average_words_per_sentence' => 0, |
| 667 |
'average_syllables_per_word' => 0 |
| 668 |
]; |
| 669 |
|
| 670 |
// Count sentences |
| 671 |
$sentences = preg_split('/[.!?]+/', $clean_content, -1, PREG_SPLIT_NO_EMPTY); |
| 672 |
$stats['sentence_count'] = count($sentences); |
| 673 |
|
| 674 |
// Count paragraphs |
| 675 |
$paragraphs = preg_split('/\n\s*\n/', trim($content), -1, PREG_SPLIT_NO_EMPTY); |
| 676 |
$stats['paragraph_count'] = count($paragraphs); |
| 677 |
|
| 678 |
// Count syllables |
| 679 |
$words = str_word_count($clean_content, 1); |
| 680 |
foreach ($words as $word) { |
| 681 |
$syllables = $this->count_syllables($word); |
| 682 |
$stats['syllable_count'] += $syllables; |
| 683 |
|
| 684 |
// Count complex words (3+ syllables) |
| 685 |
if ($syllables >= 3) { |
| 686 |
$stats['complex_words']++; |
| 687 |
} |
| 688 |
} |
| 689 |
|
| 690 |
// Calculate averages |
| 691 |
if ($stats['sentence_count'] > 0) { |
| 692 |
$stats['average_words_per_sentence'] = $stats['word_count'] / $stats['sentence_count']; |
| 693 |
} |
| 694 |
|
| 695 |
if ($stats['word_count'] > 0) { |
| 696 |
$stats['average_syllables_per_word'] = $stats['syllable_count'] / $stats['word_count']; |
| 697 |
} |
| 698 |
|
| 699 |
return $stats; |
| 700 |
} |
| 701 |
|
| 702 |
/** |
| 703 |
* Count syllables in a word (real algorithm) |
| 704 |
* |
| 705 |
* @since 1.0.0 |
| 706 |
* |
| 707 |
* @param string $word Word to count syllables for |
| 708 |
* @return int Number of syllables |
| 709 |
*/ |
| 710 |
private function count_syllables(string $word): int { |
| 711 |
$word = strtolower($word); |
| 712 |
$word = preg_replace('/[^a-z]/', '', $word); |
| 713 |
|
| 714 |
if (strlen($word) <= 3) { |
| 715 |
return 1; |
| 716 |
} |
| 717 |
|
| 718 |
// Remove common endings that don't add syllables |
| 719 |
$word = preg_replace('/(?:[^laeiouy]es|ed|[^laeiouy]e)$/', '', $word); |
| 720 |
$word = preg_replace('/^y/', '', $word); |
| 721 |
|
| 722 |
// Count vowel groups |
| 723 |
$matches = preg_match_all('/[aeiouy]{1,2}/', $word); |
| 724 |
|
| 725 |
return max(1, $matches); |
| 726 |
} |
| 727 |
|
| 728 |
/** |
| 729 |
* Determine reading level from Flesch Reading Ease score |
| 730 |
* |
| 731 |
* @since 1.0.0 |
| 732 |
* |
| 733 |
* @param float $flesch_score Flesch Reading Ease score |
| 734 |
* @return string Reading level description |
| 735 |
*/ |
| 736 |
private function determine_reading_level(float $flesch_score): string { |
| 737 |
if ($flesch_score >= 90) { |
| 738 |
return 'Very Easy (5th grade)'; |
| 739 |
} elseif ($flesch_score >= 80) { |
| 740 |
return 'Easy (6th grade)'; |
| 741 |
} elseif ($flesch_score >= 70) { |
| 742 |
return 'Fairly Easy (7th grade)'; |
| 743 |
} elseif ($flesch_score >= 60) { |
| 744 |
return 'Standard (8th-9th grade)'; |
| 745 |
} elseif ($flesch_score >= 50) { |
| 746 |
return 'Fairly Difficult (10th-12th grade)'; |
| 747 |
} elseif ($flesch_score >= 30) { |
| 748 |
return 'Difficult (College level)'; |
| 749 |
} else { |
| 750 |
return 'Very Difficult (Graduate level)'; |
| 751 |
} |
| 752 |
} |
| 753 |
|
| 754 |
/** |
| 755 |
* Calculate overall readability score |
| 756 |
* |
| 757 |
* @since 1.0.0 |
| 758 |
* |
| 759 |
* @param array $readability Readability metrics |
| 760 |
* @return int Readability score (0-100) |
| 761 |
*/ |
| 762 |
private function calculate_readability_score(array $readability): int { |
| 763 |
$scores = []; |
| 764 |
|
| 765 |
// Flesch Reading Ease (40% weight) |
| 766 |
$scores[] = max(0, min(100, $readability['flesch_reading_ease'])) * 0.4; |
| 767 |
|
| 768 |
// Flesch-Kincaid Grade Level (30% weight) - convert to 0-100 scale |
| 769 |
$grade_score = max(0, min(100, 100 - ($readability['flesch_kincaid_grade'] * 5))); |
| 770 |
$scores[] = $grade_score * 0.3; |
| 771 |
|
| 772 |
// Automated Readability Index (30% weight) - convert to 0-100 scale |
| 773 |
$ari_score = max(0, min(100, 100 - ($readability['automated_readability_index'] * 5))); |
| 774 |
$scores[] = $ari_score * 0.3; |
| 775 |
|
| 776 |
return (int) round(array_sum($scores)); |
| 777 |
} |
| 778 |
|
| 779 |
/** |
| 780 |
* Generate readability recommendations |
| 781 |
* |
| 782 |
* @since 1.0.0 |
| 783 |
* |
| 784 |
* @param array $readability Readability analysis |
| 785 |
* @return array Recommendations |
| 786 |
*/ |
| 787 |
private function generate_readability_recommendations(array $readability): array { |
| 788 |
$recommendations = []; |
| 789 |
|
| 790 |
if ($readability['flesch_reading_ease'] < 60) { |
| 791 |
$recommendations[] = [ |
| 792 |
'type' => 'readability', |
| 793 |
'priority' => 'high', |
| 794 |
'message' => 'Content is difficult to read. Consider shorter sentences and simpler words.', |
| 795 |
'action' => 'Reduce average sentence length to under 20 words' |
| 796 |
]; |
| 797 |
} |
| 798 |
|
| 799 |
if ($readability['average_sentence_length'] > 25) { |
| 800 |
$recommendations[] = [ |
| 801 |
'type' => 'sentence_length', |
| 802 |
'priority' => 'medium', |
| 803 |
'message' => 'Sentences are too long. Break them into shorter, clearer sentences.', |
| 804 |
'action' => 'Target 15-20 words per sentence' |
| 805 |
]; |
| 806 |
} |
| 807 |
|
| 808 |
if ($readability['average_syllables_per_word'] > 1.7) { |
| 809 |
$recommendations[] = [ |
| 810 |
'type' => 'word_complexity', |
| 811 |
'priority' => 'medium', |
| 812 |
'message' => 'Use simpler words to improve readability.', |
| 813 |
'action' => 'Replace complex words with simpler alternatives' |
| 814 |
]; |
| 815 |
} |
| 816 |
|
| 817 |
if ($readability['flesch_kincaid_grade'] > 12) { |
| 818 |
$recommendations[] = [ |
| 819 |
'type' => 'grade_level', |
| 820 |
'priority' => 'high', |
| 821 |
'message' => 'Content requires college-level reading skills. Simplify for broader audience.', |
| 822 |
'action' => 'Target 8th-10th grade reading level' |
| 823 |
]; |
| 824 |
} |
| 825 |
|
| 826 |
return $recommendations; |
| 827 |
} |
| 828 |
|
| 829 |
/** |
| 830 |
* Find keyword positions in content |
| 831 |
* |
| 832 |
* @since 1.0.0 |
| 833 |
* |
| 834 |
* @param string $content Content to search |
| 835 |
* @param string $keyword Keyword to find |
| 836 |
* @return array Keyword positions |
| 837 |
*/ |
| 838 |
private function find_keyword_positions(string $content, string $keyword): array { |
| 839 |
$positions = []; |
| 840 |
$offset = 0; |
| 841 |
|
| 842 |
while (($pos = strpos($content, $keyword, $offset)) !== false) { |
| 843 |
$positions[] = $pos; |
| 844 |
$offset = $pos + 1; |
| 845 |
} |
| 846 |
|
| 847 |
return $positions; |
| 848 |
} |
| 849 |
|
| 850 |
/** |
| 851 |
* Calculate keyword distribution score |
| 852 |
* |
| 853 |
* @since 1.0.0 |
| 854 |
* |
| 855 |
* @param array $positions Keyword positions |
| 856 |
* @param int $word_count Total word count |
| 857 |
* @return float Distribution score (0-100) |
| 858 |
*/ |
| 859 |
private function calculate_keyword_distribution(array $positions, int $word_count): float { |
| 860 |
if (empty($positions) || $word_count === 0) { |
| 861 |
return 0; |
| 862 |
} |
| 863 |
|
| 864 |
// Divide content into sections and check keyword presence |
| 865 |
$sections = 4; |
| 866 |
$section_size = $word_count / $sections; |
| 867 |
$sections_with_keyword = 0; |
| 868 |
|
| 869 |
for ($i = 0; $i < $sections; $i++) { |
| 870 |
$section_start = $i * $section_size; |
| 871 |
$section_end = ($i + 1) * $section_size; |
| 872 |
|
| 873 |
foreach ($positions as $pos) { |
| 874 |
if ($pos >= $section_start && $pos < $section_end) { |
| 875 |
$sections_with_keyword++; |
| 876 |
break; |
| 877 |
} |
| 878 |
} |
| 879 |
} |
| 880 |
|
| 881 |
return ($sections_with_keyword / $sections) * 100; |
| 882 |
} |
| 883 |
|
| 884 |
/** |
| 885 |
* Determine keyword optimization status |
| 886 |
* |
| 887 |
* @since 1.0.0 |
| 888 |
* |
| 889 |
* @param float $density Keyword density percentage |
| 890 |
* @return string Optimization status |
| 891 |
*/ |
| 892 |
private function determine_keyword_optimization_status(float $density): string { |
| 893 |
$config = $this->analysis_algorithms['keyword_analysis']['density_calculation']; |
| 894 |
|
| 895 |
if ($density < $config['minimum_threshold']) { |
| 896 |
return 'under_optimized'; |
| 897 |
} elseif ($density > $config['warning_threshold']) { |
| 898 |
return 'over_optimized'; |
| 899 |
} elseif ($density >= $config['optimal_range'][0] && $density <= $config['optimal_range'][1]) { |
| 900 |
return 'optimal'; |
| 901 |
} else { |
| 902 |
return 'good'; |
| 903 |
} |
| 904 |
} |
| 905 |
|
| 906 |
/** |
| 907 |
* Extract content for analysis |
| 908 |
* |
| 909 |
* @since 1.0.0 |
| 910 |
* |
| 911 |
* @param string $context_type Context type |
| 912 |
* @param int|null $context_id Context ID |
| 913 |
* @return string Content to analyze |
| 914 |
*/ |
| 915 |
private function extract_content_for_analysis(string $context_type, ?int $context_id): string { |
| 916 |
$content = ''; |
| 917 |
|
| 918 |
switch ($context_type) { |
| 919 |
case 'post': |
| 920 |
case 'page': |
| 921 |
case 'product': |
| 922 |
if ($context_id) { |
| 923 |
$post = get_post($context_id); |
| 924 |
if ($post) { |
| 925 |
$content = $post->post_title . ' ' . $post->post_content; |
| 926 |
} |
| 927 |
} |
| 928 |
break; |
| 929 |
case 'site': |
| 930 |
// Get homepage content |
| 931 |
$homepage_id = get_option('page_on_front'); |
| 932 |
if ($homepage_id && $homepage_id !== '0') { |
| 933 |
$homepage = get_post($homepage_id); |
| 934 |
if ($homepage) { |
| 935 |
$content = $homepage->post_title . ' ' . $homepage->post_content; |
| 936 |
} |
| 937 |
} else { |
| 938 |
// Get recent posts for blog homepage |
| 939 |
$recent_posts = get_posts(['numberposts' => 3]); |
| 940 |
$content_parts = []; |
| 941 |
foreach ($recent_posts as $post) { |
| 942 |
$content_parts[] = $post->post_title . ' ' . wp_trim_words($post->post_content, 100); |
| 943 |
} |
| 944 |
$content = implode(' ', $content_parts); |
| 945 |
} |
| 946 |
break; |
| 947 |
} |
| 948 |
|
| 949 |
return $content; |
| 950 |
} |
| 951 |
|
| 952 |
/** |
| 953 |
* Extract target keywords for analysis |
| 954 |
* |
| 955 |
* @since 1.0.0 |
| 956 |
* |
| 957 |
* @param string $context_type Context type |
| 958 |
* @param int|null $context_id Context ID |
| 959 |
* @return array Target keywords |
| 960 |
*/ |
| 961 |
private function extract_target_keywords(string $context_type, ?int $context_id): array { |
| 962 |
// This would typically get keywords from the database |
| 963 |
// For now, return some default keywords based on content |
| 964 |
$keywords = []; |
| 965 |
|
| 966 |
$content = $this->extract_content_for_analysis($context_type, $context_id); |
| 967 |
if (!empty($content)) { |
| 968 |
// Simple keyword extraction from title and content |
| 969 |
$words = str_word_count(strtolower(wp_strip_all_tags($content)), 1); |
| 970 |
$word_counts = array_count_values($words); |
| 971 |
|
| 972 |
// Remove common stop words |
| 973 |
$stop_words = ['the', 'and', 'or', 'but', 'in', 'on', 'at', 'to', 'for', 'of', 'with', 'by']; |
| 974 |
foreach ($stop_words as $stop_word) { |
| 975 |
unset($word_counts[$stop_word]); |
| 976 |
} |
| 977 |
|
| 978 |
arsort($word_counts); |
| 979 |
$keywords = array_slice(array_keys($word_counts), 0, 5); |
| 980 |
} |
| 981 |
|
| 982 |
return $keywords; |
| 983 |
} |
| 984 |
|
| 985 |
/** |
| 986 |
* Store analysis results in database |
| 987 |
* |
| 988 |
* @since 1.0.0 |
| 989 |
* |
| 990 |
* @param string $context_type Context type |
| 991 |
* @param int|null $context_id Context ID |
| 992 |
* @param array $results Analysis results |
| 993 |
* @return bool Success status |
| 994 |
*/ |
| 995 |
private function store_analysis_results(string $context_type, ?int $context_id, array $results): bool { |
| 996 |
global $wpdb; |
| 997 |
|
| 998 |
$table_name = $wpdb->prefix . 'thinkrank_seo_analysis'; |
| 999 |
|
| 1000 |
$data = [ |
| 1001 |
'context_type' => $context_type, |
| 1002 |
'context_id' => $context_id, |
| 1003 |
'analysis_type' => 'ai_content_analysis', |
| 1004 |
'analysis_data' => wp_json_encode($results), |
| 1005 |
'score' => $results['optimization_score'] ?? 0, |
| 1006 |
'status' => 'completed', |
| 1007 |
'ai_confidence' => $results['ai_confidence'] ?? 0, |
| 1008 |
'recommendations' => wp_json_encode($results['recommendations'] ?? []), |
| 1009 |
'analyzed_by' => get_current_user_id() |
| 1010 |
]; |
| 1011 |
|
| 1012 |
// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery -- Analysis results storage requires direct database access |
| 1013 |
$result = $wpdb->insert($table_name, $data); |
| 1014 |
|
| 1015 |
return $result !== false; |
| 1016 |
} |
| 1017 |
|
| 1018 |
/** |
| 1019 |
* Calculate optimization score |
| 1020 |
* |
| 1021 |
* @since 1.0.0 |
| 1022 |
* |
| 1023 |
* @param array $analysis Complete analysis results |
| 1024 |
* @return int Optimization score (0-100) |
| 1025 |
*/ |
| 1026 |
private function calculate_optimization_score(array $analysis): int { |
| 1027 |
$scores = []; |
| 1028 |
|
| 1029 |
// Readability score (25% weight) |
| 1030 |
if (!empty($analysis['readability']['readability_score'])) { |
| 1031 |
$scores['readability'] = $analysis['readability']['readability_score'] * ($this->scoring_weights['readability'] / 100); |
| 1032 |
} |
| 1033 |
|
| 1034 |
// Keyword optimization score (30% weight) |
| 1035 |
if (!empty($analysis['keyword_analysis']['optimization_score'])) { |
| 1036 |
$scores['keyword_optimization'] = $analysis['keyword_analysis']['optimization_score'] * ($this->scoring_weights['keyword_optimization'] / 100); |
| 1037 |
} |
| 1038 |
|
| 1039 |
// Content structure score (20% weight) |
| 1040 |
if (!empty($analysis['structure_analysis']['structure_score'])) { |
| 1041 |
$scores['content_structure'] = $analysis['structure_analysis']['structure_score'] * ($this->scoring_weights['content_structure'] / 100); |
| 1042 |
} |
| 1043 |
|
| 1044 |
// Semantic relevance score (15% weight) |
| 1045 |
if (!empty($analysis['semantic_analysis']['relevance_score'])) { |
| 1046 |
$scores['semantic_relevance'] = $analysis['semantic_analysis']['relevance_score'] * ($this->scoring_weights['semantic_relevance'] / 100); |
| 1047 |
} |
| 1048 |
|
| 1049 |
// Technical SEO score (10% weight) |
| 1050 |
$technical_score = $this->calculate_technical_seo_score($analysis); |
| 1051 |
$scores['technical_seo'] = $technical_score * ($this->scoring_weights['technical_seo'] / 100); |
| 1052 |
|
| 1053 |
return (int) round(array_sum($scores)); |
| 1054 |
} |
| 1055 |
|
| 1056 |
/** |
| 1057 |
* Generate optimization recommendations |
| 1058 |
* |
| 1059 |
* @since 1.0.0 |
| 1060 |
* |
| 1061 |
* @param array $analysis Complete analysis results |
| 1062 |
* @return array Optimization recommendations |
| 1063 |
*/ |
| 1064 |
private function generate_optimization_recommendations(array $analysis): array { |
| 1065 |
$recommendations = []; |
| 1066 |
|
| 1067 |
// Readability recommendations |
| 1068 |
if (!empty($analysis['readability']['recommendations'])) { |
| 1069 |
$recommendations = array_merge($recommendations, $analysis['readability']['recommendations']); |
| 1070 |
} |
| 1071 |
|
| 1072 |
// Keyword recommendations |
| 1073 |
if (!empty($analysis['keyword_analysis']['recommendations'])) { |
| 1074 |
$recommendations = array_merge($recommendations, $analysis['keyword_analysis']['recommendations']); |
| 1075 |
} |
| 1076 |
|
| 1077 |
// Structure recommendations |
| 1078 |
if (!empty($analysis['structure_analysis']['recommendations'])) { |
| 1079 |
$recommendations = array_merge($recommendations, $analysis['structure_analysis']['recommendations']); |
| 1080 |
} |
| 1081 |
|
| 1082 |
// Content length recommendations |
| 1083 |
$word_count = $analysis['content_stats']['word_count'] ?? 0; |
| 1084 |
if ($word_count < 300) { |
| 1085 |
$recommendations[] = [ |
| 1086 |
'type' => 'content_length', |
| 1087 |
'priority' => 'high', |
| 1088 |
'message' => 'Content is too short for optimal SEO performance.', |
| 1089 |
'action' => 'Expand content to at least 300 words' |
| 1090 |
]; |
| 1091 |
} |
| 1092 |
|
| 1093 |
// Sort recommendations by priority |
| 1094 |
usort($recommendations, function ($a, $b) { |
| 1095 |
$priority_order = ['high' => 3, 'medium' => 2, 'low' => 1]; |
| 1096 |
return ($priority_order[$b['priority']] ?? 0) - ($priority_order[$a['priority']] ?? 0); |
| 1097 |
}); |
| 1098 |
|
| 1099 |
return $recommendations; |
| 1100 |
} |
| 1101 |
|
| 1102 |
/** |
| 1103 |
* Calculate AI confidence score |
| 1104 |
* |
| 1105 |
* @since 1.0.0 |
| 1106 |
* |
| 1107 |
* @param array $analysis Analysis results |
| 1108 |
* @return float Confidence score (0-1) |
| 1109 |
*/ |
| 1110 |
private function calculate_ai_confidence(array $analysis): float { |
| 1111 |
$confidence_factors = []; |
| 1112 |
|
| 1113 |
// Content length confidence |
| 1114 |
$word_count = $analysis['content_stats']['word_count'] ?? 0; |
| 1115 |
if ($word_count > 0) { |
| 1116 |
$confidence_factors[] = min(1.0, $word_count / 500); // Full confidence at 500+ words |
| 1117 |
} |
| 1118 |
|
| 1119 |
// Analysis completeness confidence |
| 1120 |
$completed_analyses = 0; |
| 1121 |
$total_analyses = 4; // readability, keyword, semantic, structure |
| 1122 |
|
| 1123 |
if (!empty($analysis['readability'])) { |
| 1124 |
$completed_analyses++; |
| 1125 |
} |
| 1126 |
if (!empty($analysis['keyword_analysis'])) { |
| 1127 |
$completed_analyses++; |
| 1128 |
} |
| 1129 |
if (!empty($analysis['semantic_analysis'])) { |
| 1130 |
$completed_analyses++; |
| 1131 |
} |
| 1132 |
if (!empty($analysis['structure_analysis'])) { |
| 1133 |
$completed_analyses++; |
| 1134 |
} |
| 1135 |
|
| 1136 |
$confidence_factors[] = $completed_analyses / $total_analyses; |
| 1137 |
|
| 1138 |
return empty($confidence_factors) ? 0.0 : array_sum($confidence_factors) / count($confidence_factors); |
| 1139 |
} |
| 1140 |
|
| 1141 |
/** |
| 1142 |
* Calculate validation score |
| 1143 |
* |
| 1144 |
* @since 1.0.0 |
| 1145 |
* |
| 1146 |
* @param array $validation Validation results |
| 1147 |
* @return int Score (0-100) |
| 1148 |
*/ |
| 1149 |
private function calculate_validation_score(array $validation): int { |
| 1150 |
$score = 100; |
| 1151 |
$score -= count($validation['errors']) * 20; |
| 1152 |
$score -= count($validation['warnings']) * 10; |
| 1153 |
$score -= count($validation['suggestions']) * 5; |
| 1154 |
|
| 1155 |
return max(0, $score); |
| 1156 |
} |
| 1157 |
|
| 1158 |
/** |
| 1159 |
* Calculate technical SEO score |
| 1160 |
* |
| 1161 |
* @since 1.0.0 |
| 1162 |
* |
| 1163 |
* @param array $analysis Analysis results |
| 1164 |
* @return int Technical SEO score (0-100) |
| 1165 |
*/ |
| 1166 |
private function calculate_technical_seo_score(array $analysis): int { |
| 1167 |
$score = 100; |
| 1168 |
|
| 1169 |
// Deduct points for missing elements |
| 1170 |
if (empty($analysis['structure_analysis']['heading_structure'])) { |
| 1171 |
$score -= 20; |
| 1172 |
} |
| 1173 |
|
| 1174 |
if (empty($analysis['structure_analysis']['image_analysis'])) { |
| 1175 |
$score -= 15; |
| 1176 |
} |
| 1177 |
|
| 1178 |
if (empty($analysis['structure_analysis']['link_analysis'])) { |
| 1179 |
$score -= 15; |
| 1180 |
} |
| 1181 |
|
| 1182 |
return max(0, $score); |
| 1183 |
} |
| 1184 |
|
| 1185 |
/** |
| 1186 |
* Simple implementations for semantic analysis methods |
| 1187 |
* These would be enhanced with actual AI/ML libraries in production |
| 1188 |
*/ |
| 1189 |
|
| 1190 |
private function extract_topic_clusters(string $content): array { |
| 1191 |
// Simplified topic extraction |
| 1192 |
return ['topics' => ['general content'], 'confidence' => 0.7]; |
| 1193 |
} |
| 1194 |
|
| 1195 |
private function recognize_entities(string $content): array { |
| 1196 |
// Simplified entity recognition |
| 1197 |
return ['entities' => [], 'confidence' => 0.6]; |
| 1198 |
} |
| 1199 |
|
| 1200 |
private function analyze_sentiment(string $content): array { |
| 1201 |
// Simplified sentiment analysis |
| 1202 |
return ['sentiment' => 'neutral', 'score' => 0.0, 'confidence' => 0.7]; |
| 1203 |
} |
| 1204 |
|
| 1205 |
private function classify_content(string $content): array { |
| 1206 |
// Simplified content classification |
| 1207 |
return ['category' => 'informational', 'confidence' => 0.8]; |
| 1208 |
} |
| 1209 |
|
| 1210 |
private function extract_semantic_keywords(string $content): array { |
| 1211 |
// Simplified semantic keyword extraction |
| 1212 |
return []; |
| 1213 |
} |
| 1214 |
|
| 1215 |
private function calculate_content_coherence(array $semantic_analysis): float { |
| 1216 |
return 0.8; // Simplified coherence score |
| 1217 |
} |
| 1218 |
|
| 1219 |
private function calculate_semantic_relevance(array $semantic_analysis): float { |
| 1220 |
return 75.0; // Simplified relevance score |
| 1221 |
} |
| 1222 |
|
| 1223 |
private function analyze_heading_structure(string $content): array { |
| 1224 |
// Extract headings from HTML content |
| 1225 |
$headings = []; |
| 1226 |
preg_match_all('/<h([1-6])[^>]*>(.*?)<\/h[1-6]>/i', $content, $matches, PREG_SET_ORDER); |
| 1227 |
|
| 1228 |
foreach ($matches as $match) { |
| 1229 |
$headings[] = [ |
| 1230 |
'level' => (int) $match[1], |
| 1231 |
'text' => wp_strip_all_tags($match[2]), |
| 1232 |
'length' => strlen(wp_strip_all_tags($match[2])) |
| 1233 |
]; |
| 1234 |
} |
| 1235 |
|
| 1236 |
return ['headings' => $headings, 'count' => count($headings)]; |
| 1237 |
} |
| 1238 |
|
| 1239 |
private function analyze_paragraph_structure(string $content): array { |
| 1240 |
$paragraphs = preg_split('/\n\s*\n/', wp_strip_all_tags($content), -1, PREG_SPLIT_NO_EMPTY); |
| 1241 |
$analysis = ['count' => count($paragraphs), 'average_length' => 0]; |
| 1242 |
|
| 1243 |
if (!empty($paragraphs)) { |
| 1244 |
$total_words = 0; |
| 1245 |
foreach ($paragraphs as $paragraph) { |
| 1246 |
$total_words += str_word_count($paragraph); |
| 1247 |
} |
| 1248 |
$analysis['average_length'] = $total_words / count($paragraphs); |
| 1249 |
} |
| 1250 |
|
| 1251 |
return $analysis; |
| 1252 |
} |
| 1253 |
|
| 1254 |
private function analyze_list_usage(string $content): array { |
| 1255 |
$ul_count = preg_match_all('/<ul[^>]*>/i', $content); |
| 1256 |
$ol_count = preg_match_all('/<ol[^>]*>/i', $content); |
| 1257 |
|
| 1258 |
return ['unordered_lists' => $ul_count, 'ordered_lists' => $ol_count]; |
| 1259 |
} |
| 1260 |
|
| 1261 |
private function analyze_image_usage(string $content): array { |
| 1262 |
preg_match_all('/<img[^>]+>/i', $content, $img_tags); |
| 1263 |
$images = []; |
| 1264 |
|
| 1265 |
foreach ($img_tags[0] as $img_tag) { |
| 1266 |
preg_match('/alt=["\']([^"\']*)["\']/', $img_tag, $alt_match); |
| 1267 |
$images[] = ['has_alt' => !empty($alt_match[1])]; |
| 1268 |
} |
| 1269 |
|
| 1270 |
return ['count' => count($images), 'images' => $images]; |
| 1271 |
} |
| 1272 |
|
| 1273 |
private function analyze_link_structure(string $content): array { |
| 1274 |
preg_match_all('/<a[^>]+href=["\']([^"\']+)["\'][^>]*>([^<]*)<\/a>/i', $content, $link_matches, PREG_SET_ORDER); |
| 1275 |
|
| 1276 |
$internal_links = 0; |
| 1277 |
$external_links = 0; |
| 1278 |
|
| 1279 |
foreach ($link_matches as $match) { |
| 1280 |
$url = $match[1]; |
| 1281 |
if (strpos($url, home_url()) === 0 || strpos($url, '/') === 0) { |
| 1282 |
$internal_links++; |
| 1283 |
} else { |
| 1284 |
$external_links++; |
| 1285 |
} |
| 1286 |
} |
| 1287 |
|
| 1288 |
return ['internal' => $internal_links, 'external' => $external_links]; |
| 1289 |
} |
| 1290 |
|
| 1291 |
private function analyze_content_flow(string $content): array { |
| 1292 |
// Simplified content flow analysis |
| 1293 |
return ['flow_score' => 80, 'transitions' => 5]; |
| 1294 |
} |
| 1295 |
|
| 1296 |
private function calculate_structure_score(array $structure_analysis): int { |
| 1297 |
$score = 100; |
| 1298 |
|
| 1299 |
// Deduct points for poor structure |
| 1300 |
if (empty($structure_analysis['heading_structure']['headings'])) { |
| 1301 |
$score -= 30; |
| 1302 |
} |
| 1303 |
|
| 1304 |
if ($structure_analysis['paragraph_analysis']['average_length'] > 150) { |
| 1305 |
$score -= 20; |
| 1306 |
} |
| 1307 |
|
| 1308 |
if ($structure_analysis['image_analysis']['count'] === 0) { |
| 1309 |
$score -= 15; |
| 1310 |
} |
| 1311 |
|
| 1312 |
return max(0, $score); |
| 1313 |
} |
| 1314 |
|
| 1315 |
private function generate_structure_recommendations(array $structure_analysis): array { |
| 1316 |
$recommendations = []; |
| 1317 |
|
| 1318 |
if (empty($structure_analysis['heading_structure']['headings'])) { |
| 1319 |
$recommendations[] = [ |
| 1320 |
'type' => 'headings', |
| 1321 |
'priority' => 'high', |
| 1322 |
'message' => 'Add heading tags (H1, H2, H3) to improve content structure', |
| 1323 |
'action' => 'Use proper heading hierarchy' |
| 1324 |
]; |
| 1325 |
} |
| 1326 |
|
| 1327 |
return $recommendations; |
| 1328 |
} |
| 1329 |
|
| 1330 |
private function find_semantic_keywords(string $content, array $keywords): array { |
| 1331 |
// Simplified semantic keyword finding |
| 1332 |
return []; |
| 1333 |
} |
| 1334 |
|
| 1335 |
private function calculate_keyword_optimization_score(array $keyword_analysis): int { |
| 1336 |
$score = 0; |
| 1337 |
$keyword_count = count($keyword_analysis['primary_keywords']) + count($keyword_analysis['secondary_keywords']); |
| 1338 |
|
| 1339 |
if ($keyword_count > 0) { |
| 1340 |
$optimal_count = 0; |
| 1341 |
foreach ($keyword_analysis['primary_keywords'] as $keyword) { |
| 1342 |
if ($keyword['optimization_status'] === 'optimal') { |
| 1343 |
$optimal_count++; |
| 1344 |
} |
| 1345 |
} |
| 1346 |
$score = ($optimal_count / $keyword_count) * 100; |
| 1347 |
} |
| 1348 |
|
| 1349 |
return (int) $score; |
| 1350 |
} |
| 1351 |
|
| 1352 |
private function generate_keyword_recommendations(array $keyword_analysis): array { |
| 1353 |
$recommendations = []; |
| 1354 |
|
| 1355 |
foreach ($keyword_analysis['primary_keywords'] as $keyword) { |
| 1356 |
if ($keyword['optimization_status'] === 'under_optimized') { |
| 1357 |
$recommendations[] = [ |
| 1358 |
'type' => 'keyword_density', |
| 1359 |
'priority' => 'medium', |
| 1360 |
'message' => "Keyword '{$keyword['keyword']}' appears too few times", |
| 1361 |
'action' => 'Increase keyword usage naturally' |
| 1362 |
]; |
| 1363 |
} elseif ($keyword['optimization_status'] === 'over_optimized') { |
| 1364 |
$recommendations[] = [ |
| 1365 |
'type' => 'keyword_density', |
| 1366 |
'priority' => 'high', |
| 1367 |
'message' => "Keyword '{$keyword['keyword']}' may be over-optimized", |
| 1368 |
'action' => 'Reduce keyword density to avoid keyword stuffing' |
| 1369 |
]; |
| 1370 |
} |
| 1371 |
} |
| 1372 |
|
| 1373 |
return $recommendations; |
| 1374 |
} |
| 1375 |
} |
| 1376 |
|
