Describes the history and features of eight open-source search engines, freeWAIS-sf (aging code and hard to install, but good for searching email and public domain etexts); Harvest (powerful gathering features for frequently-changing data stores, good with structured documents); ht://Dig (tricky to configure, no phrase searching, automatic stemming and match word highlighting); Isearch (weak documentation and support, easy to install, dated interface, Z39.50 support); MPS Information Server (zippy indexing of both text and structured data, Z39.50 support, Perl API, limited documentation); SWISH-E (simple to install engine, CGIs in Perl and PHP still beta, good for HTML pages, recognizes new META tags, sorts results by field; WebGlimpse (easy to install and configure, requires commercial version for customized output); Yaz/Zebra (mainly Z39.50, no Perl API, mainly a toolkit to index and respond to distributed client queries). Article also points out that chaotic information is less than helpful and encourages organization, structure and vocabulary control.
FS Consulting, developers of MPS, took part in the TExt Retrieval Conference organized by the US National Institute of Standards and Technology to encourage research in information retrieval from large text collections. The conference set various tasks for search engines and provided a venue for the developers to test them. The following links are to articles written about using the search engine for the TREC tasks.
ScienceServer
Technology Overview
A description of the MPS Information Server in the ScienceServer library journal delivery and search system.
A description of the MPS Information Server in the ScienceServer library journal delivery and search system.