Add SemQuestTree
What
SemQuestTree — a multi-table semantic retriever in pure Java 8, zero third-party dependencies, MIT. Knowledge is written as a tree of criteria (event-level atomic assertions); a decision model judges content layer by layer as it descends the tree — add places content into leaves, quest retrieves with natural-language queries. Every hit carries its full decision path (which criteria fired, per-layer probability, token cost).
Why it fits (criteria c/d)
Existing Search entries (Lucene, Solr, Elasticsearch, OpenSearch) are inverted-index or vector-similarity engines. A criteria-tree retriever where the model itself does the routing is a different retrieval paradigm — qualifying under (c) "absolutely unique in its approach" / (d) "a niche product that fills a gap": small-to-medium knowledge bases that must stay human-readable and steerable.
- Zero dependencies (JDK 8+ only), no infrastructure to run
- Multi-label placement without top-k truncation
- Ships with a visualization server and HTTP API; runs out of the box with a built-in mock backend (no credentials needed)
- English README, API reference and troubleshooting docs
Disclosure
I am the author.
Summary by cubic
Adds SemQuestTree to the search engines list in README_SOURCE.md.
SemQuestTree is a criteria-tree semantic retriever for Java 8 with no dependencies; it routes content through a decision model rather than using inverted-index or vector similarity, and every hit carries a readable decision path. The PR author is also the project's maintainer.