Keyword Recommendation System Using Dynamic Standards

The patent segments the keyword recommendation process into multiple independent modules: a keyword database storing diverse keyword types (product names, categories, attributes, synonyms), a recommendation engine that applies different recommendation standards (PPC, CTR, query volume, relevance), and a selection interface. This segmentation allows the system to handle multiple recommendation scenarios without becoming unmanageably complex.

The patent implements a universal keyword recommendation system that can serve multiple functions: recommending exact product matches, suggesting related keywords when no exact match exists, providing tail keywords for long-tail search optimization, and supporting different recommendation criteria (PPC, CTR, query volume). This multi-functionality resolves the contradiction by making the system adaptable to various advertiser needs while maintaining a unified architecture.

The patent implements dynamic recommendation standards that can be adjusted in real-time based on advertiser selection. The system allows advertisers to choose from multiple recommendation criteria (PPC, CTR, query volume, relevance) and dynamically switch between them. The recommendation engine adapts its behavior based on the selected standard, providing precise recommendations while maintaining ease of operation through a user-friendly interface that presents clear options.

The patent enables advertisers to self-select their preferred recommendation standard from a presented list of options. The system provides default recommendations but allows advertisers to independently choose the criterion that best suits their campaign goals (e.g., selecting CTR for brand awareness or PPC for conversion optimization). This self-service approach improves precision by allowing advertiser control while maintaining ease of operation through intuitive selection interfaces.

The patent applies local quality by differentiating between different types of keyword recommendations within the same system. It recommends highly-related keywords for main product categories while simultaneously suggesting tail keywords for long-tail search optimization. The system assigns different relevance thresholds to different keyword types: high relevance for product-name matches, moderate relevance for category keywords, and lower relevance for tail keywords. This resolves the contradiction by providing both quantity and reliability through differentiated recommendation quality.

The patent implements partial action by recommending keywords at different relevance levels based on advertiser needs. When an advertiser selects to receive more keyword options, the system partially expands the recommendation set to include less highly-related but potentially valuable tail keywords. This allows advertisers to receive exactly the quantity of keywords they need without forcing low-relevance keywords on those who only want precise matches, thus maintaining reliability while increasing quantity when requested.