Judgment Document Retrieval via Information Code Association

The patent segments the retrieval process into two distinct dimensions: vertical retrieval based on classification codes (cause of action, document type, court level) and horizontal retrieval based on association relationships (similar cases, related parties, referenced documents). This segmentation allows each dimension to operate independently while contributing to comprehensive results, resolving the contradiction between simplicity and comprehensiveness.

The patent introduces a horizontal dimension to the traditional vertical retrieval model. By adding association relationships as a new dimension alongside classification hierarchies, the system expands from single-dimension keyword matching to multi-dimensional retrieval, enabling comprehensive results while maintaining interface simplicity through unified search input.

The patent performs preliminary action by pre-establishing association relationships between judgment documents during the indexing phase. Association rules, similarity algorithms, and reference relationships are computed and stored in advance, so that during retrieval, the system can directly leverage these pre-computed connections without requiring users to iteratively input multiple keywords, achieving both accuracy and simplicity.

The patent introduces classification codes and association relationships as intermediaries between user queries and judgment documents. Instead of direct keyword matching, the system uses these intermediaries to bridge the gap, enabling accurate retrieval through single keyword input while the intermediary structures handle the complexity of multi-dimensional matching and association.

The patent merges vertical retrieval results (from classification code matching) and horizontal retrieval results (from association relationships) into a unified result set. By combining these two retrieval streams, the system maintains the speed advantage of direct matching while supplementing it with associatively related documents, achieving both speed and completeness simultaneously.

The patent creates a universal retrieval framework that handles both exact matching (vertical) and associative matching (horizontal) through the same system architecture. The retrieval engine is designed to execute multiple retrieval strategies in parallel or sequence, making the system multi-functional in terms of retrieval modes while presenting a unified interface to users.