SEED (Sub-topics)

SEED (Sub-topics) is a list of sub-topics generated from a seed query (the starting keyword) that forms the foundation for building topical clusters and knowledge graphs. Sub-topics are discovered from three sources: LLM predictions, SERP data (People Also Ask, Related Searches), and competitor analysis of what the top 10 results cover. Each sub-topic becomes a potential node in the knowledge graph or a candidate for a standalone article.

In the semantic audit pipeline, SEED sub-topics serve as input for the keyword expansion tool, which generates 20–30 sub-queries from each sub-topic. The quality of seeds determines the quality of the entire downstream pipeline—poor seed quality compromises the entire downstream analysis.

For example, the seed query 'inheritance law' might produce sub-topics like: forced heirship, wills, intestate succession, estate division, inheritance tax, and renunciation of inheritance.

In practice, you should generate sub-topics from at least two sources (LLM + SERP). Always validate these against domain expertise, as LLMs can miss niche industry topics.