meta data for this page
- Deutsch (de)
- English (en)
5.6 Search Types
INTERGATOR Patent Search offers three basic search types, which differ in the way search terms are processed and in how hits are determined. This allows the platform to support semantic, Boolean, and hybrid search strategies.
Overview of the three search types:
| Search Type | Description | Example | Typical Results |
|---|---|---|---|
| Semantic | Analyzes meanings and contextual relationships instead of mere character strings. The AI recognizes thematically related concepts, synonyms, and paraphrases. | “project report” | Hits containing *minutes*, *documentation*, *record*, or *memo*, even if “project report” does not appear exactly |
| Boolean | Uses a purely lexical search with terms, field references, and Boolean operators. The search is not semantic, but based on the entered search syntax. | `ti:solar AND ab:inverter` | Hits containing the specified search expression in the defined fields |
| Hybrid | Combines semantic search with additional Boolean or syntax filters. Semantic search remains the basis, while Boolean elements act as additional restrictive filters. | Free text + syntax filter | Results that are semantically relevant and also match defined lexical or bibliographic criteria |
Functional Overview
- Semantic Search:
Semantic search is based on an AI language model specifically trained for patents. Search terms, text passages, or complete descriptions are interpreted in terms of meaning and linked to similar concepts. This makes it possible to find documents even if the exact search terms do not appear in them.
- Boolean Search:
Boolean search uses a classic lexical search logic with terms, field names, and Boolean operators. In INTERGATOR Patent Search, it is performed via the BOOL perspective. In this perspective, the search is not semantic. The Boolean query is ranking and therefore also influences the order of the hits.
- Hybrid Search:
Hybrid search combines semantic search with additional Boolean or syntax filters. Within a semantic search, these Boolean elements act as filters, but are not ranking. This makes it possible to narrow down semantically identified hits using bibliographic or lexical criteria.
AI Patent Model
For semantic search, INTERGATOR Patent Search uses its own AI language models, trained on millions of German and English patent documents. These models run entirely locally – there is no access to external AI cloud services. This enables:
- a transparent and traceable semantic search
- cross-lingual search
- continuous further development and adaptation to specific subject areas
Use in Practice
Which search type is applied depends on the selected search approach:
- An input in natural language generally produces a semantic search.
- Additional Boolean or syntax expressions within a semantic search result in a hybrid search.
- An exclusively Boolean search is carried out via the BOOL perspective.
Note
The three search types complement each other effectively:
- Semantic search is particularly suitable for content-based and context-related searches.
- Boolean search is suitable for precise, controlled queries using search syntax.
- Hybrid search combines both approaches and is especially useful when semantically identified hits need to be narrowed down further in a targeted way.