Search Engine

The MathWebSearch system (MWS) is a content-based search engine for mathematical formulae. It indexes MathML formulae, using a technique derived from automated theorem proving: term indexing.

The MWS search engine is optimized for fast query responses and interactive applications. Unification queries form the basis for expressive query languages with well-defined semantics. Any corpus whose formulae can referenced by URIs and converted to Content MathML can be indexed by MWS.

Project Information

MWS is developed by the KWARC group at Jacobs University. The source code is available in the MWS GitHub Repository. For details and news about the project, see the MWS Developer Portal.

Applications

The MWS search engine can be employed by various front-ends. Here are some web-based ones.

• TeMaSearch - combined text and Math search for Zentralblatt Math
• ZBLSearch - a content-oriented search engine for the formulae in the 3.3 million reviews and abstracts in the Zentralblatt Math data base. Queries in LaTeX extended with search variables.
• XLSearch - a search engine for spreadsheet formulae. Queries in Excel + variables
• FlatSearch - a search engine for formal
• SentidoSearch - a multi-format input front-end for search queries based on the Sentido system. It includes a generic panel-based content formula editor.
• Mediawiki integrations: Physikerwelt, (Wikipedia is coming).

We are currently developing:

• ETA May 2013: NTCIR-11 - a formula search engine for the NTCIR-11 Math Pilot Task, ranging over 100 000 articles from arXiv.org. Queries in LaTeX + variables.
• ETA June 2013: MizarATS - a search engine for applicable theorems in the Mizar system. This already runs as part of the mizarwiki system, but we are working on a standalone version. Queries in Mizar mathematical language format.

Our applications are largely restricted by our server capacities and in particular the RAM (we would like to offer a formula search engine for all of the arXiv.org, but would need another server with about 128GB RAM for that. If you want to donate server space or propose another application, contact Michael Kohlhase.