Unlike other information retrieval systems, Terrier learns from empirical data and adapts to the user’s information needs and queries. The framework offers a modular API for querying, allowing the development of applications as diverse as experimenting with standard test collections, or the rapid deployment of Web, intranet and desktop search engines.
Terrier implements state-of-the-art indexing and retrieval functionalities, such as DFR and BM25F, as well as term dependence proximity models such as Markov Random Fields, and provides an ideal platform for the rapid development and evaluation of large-scale retrieval applications.
Terrier has an outstanding performance with respect to other current public technologies that aim to provide similar retrieval facilities and is readily deployable on large-scale collections of documents. Search results can be viewed in a handy desktop search application, or online from a JSP web interface.