Compounds
How AI makes the difference
In cross-linguistic data retrieval, the diversity of languages in terms of vocabulary, grammar or syntax poses a major challenge. How can terms from different languages be meaningfully related to each other without having to translate the content? Conventional search solutions use thesauri or synonym lists for this purpose, which usually have to be maintained in a time-consuming manner, e.g., to link technical terms or synonyms. This manual processing is not only time-consuming and labor-intensive, but also always lags behind, especially with large, constantly growing data. New synonyms are added. Terms change.
If you want to search across languages in data, the hurdle of different language syntax is added. While in one language terms are a combination of several word stems, in another language it is only apparent from the context that the word group is a generic term. In German, the term "Zylinderkopfdichtung" (cylinder head gasket) is made up of the individual words "Zylinder" (cylinder), "Kopf" (head) and "Dichtung" (gasket), each of which has its own independent meaning. It is only in this genuine combination that it becomes clear that this is a special term. In English, there are hardly any such word combinations (or compounds) and it is only from the context of the content that the reader realizes that the terms "cylinder head gasket" do not make sense independently of each other. For a search solution, these syntactic subtleties are a major challenge, and even more so for cross-linguistic searches.
INTERGATOR Patent Search uses specially trained AI models that are compatible with each other. Without having to translate the search terms, the artificial intelligence recognizes the cross-linguistic similarity and recognizes that the German "Zylinderkopfdichtung" is analogous to the English "cylinder head gasket" and therefore lists patents in other languages in addition to the German ones. The AI does not use maintained lists and thesauri. Instead, it uses artificial intelligence statistics and is thus independent of upstream translations. The technology is applicable to different languages and INTERGATOR uses a specially trained model for each language.