Triplehop MatchPoint search software was acquired by Oracle last June. Support continues for existing customers; a migration path may available as Oracle integrates Triplehop's technology into its own enterprise search efforts.This looks like a good match Oracle Text
was a fairly good traditional information retrieval
system. It used techniques for indexing, query processing retrieval and ranking that have proved successful in the TREC
tests over the years. But it wasn't very good at handling the kind of short generic and vague queries we often see in enterprise search. MatchPoint
contains many of the attributes of modern search engines: it has a robot crawler, database integration, and can index files on shared servers and workstations, duplicate detection, over 200 file formats, incremental indexing and simple search syntax. But it goes beyond that: add-ons index video and audio multimedia files, the queries can include Booleans, parentheses, wild cards, and fuzzy matching, as well as "Search By Concept" and results clustering. While they claim a form of faceted metadata results, I've never seen it in action so reserve judgment.
So for Oracle, it seems a good match. They get a system that matches, at least in theory, all the enterprise search software features, while still including many of the information retrieval features that have seemed important in the past.