FullTextSuche

FullTextSuche (full-text search) refers to techniques for indexing and querying the textual content of documents to support fast retrieval based on the words they contain. The core idea is to build an index that makes it possible to find documents where the user’s query matches terms, phrases, or patterns, not only document titles or tags.

Under the hood, full-text search relies on an inverted index: each term maps to a list of documents containing it, often with position information for phrase and proximity queries. An indexer tokenizes text, normalizes it (case folding, stemming or lemmatization), and may remove stop words. A query processor parses the user input, retrieves candidate documents, and applies a ranking model such as BM25 or TF-IDF, sometimes augmented with field weights, language analysis, and synonyms.

Supported query features typically include boolean operators, exact-phrase and proximity searches, wildcards, fuzzy matching, and custom analyzers for language-specific stemming and stop-word lists.

Common platforms include search engines and databases: Elasticsearch and Apache Solr (built on Lucene), PostgreSQL's full-text search, MySQL's full-text indexes, and Lucene-based libraries. These systems provide APIs for indexing, updating, and querying, plus ranking, highlighting, and result filtering.

Applications span content management, site search, digital libraries, e-commerce product search, and enterprise document repositories. Trade-offs involve indexing time, update latency, language support, and the quality of ranking, especially for multilingual or highly technical corpora.