Semantic and similarity search

The exact search (= classic full-text search) only makes information visible when the corresponding search terms are known. But if one is not aware of the information, one doesn't know the wording, and the information remains concealed.

The synonym search (= semantic search) finds synonyms of the entered query text, within the query language and across other languages. For example, the query "bicycle" will also find "bike", "push-bike", "bicyclette", "Fahrrad", "Velo", "velocipede", "Drahtesel," etc.; a key feature when looking for information written by other people (using different terms for the same thing).

Collocated terms (or sets of words that are used together) are also recognized and handled, e.g. "United Nations Secretariat", "common sense", "Europäische Union", "office fédérale de la statistique".

Finally, the similarity search finds similar documents by comparing the content of a query text in natural language (e.g. an incoming e-mail) to the content of the existing documents; no literal conformity to the search terms is needed (see SNVworld).

The cross-lingual similarity search is central to an automatic categorization of documents and matching documents of similar content. It works even if the majority of expressions differ.

InfoCodex is the only system currently available on the market that performs a true cross-lingual semantic search and similarity search. No translation of queries is necessary to find documents in other languages.