To try out the MARIAN system you can use the MARIAN WWW interface which provides the top-ranked 30 items after searching against the Virginia Tech catalog.
MARIAN (Multiple Access Retrieval of Information with ANnotations) is a system developed by the Virginia Tech Computing Center starting in 1991. It runs on a collection of Pentium and NeXT computers, using one or more threads for each user session. A similarity value is computed for each field (e.g., title, author, subject). The combiner module computes an overall similarity that is used to rank the documents, so users see the top-ranked items only.
Using the NeXT interface, one can call for successive sets of 30 items. Also, one can request circulation data that is obtained from the VTLS computer by way of an expert system analysis module.
In addition there is a Gopher+ interface, one using the curses interface package and access using telnet.
In 1995 the MARIAN system was opened for wider use, after the data was brought into synchronization with the current contents of the VTLS computer.
One precursor of MARIAN is the CODER system. Many of the ideas from it, the interprocess communication approach, and the English lexicon developed for it, are used in MARIAN.
The other precursor of MARIAN is the REVTOLC study --- Retrieval Experiment, Virginia Tech On Line Catalog. A pilot study was done with 300,000 records and 52 users. The full study is reported at length in the 1993 Ph.D. dissertation of Amjad Daoud. Students preferred our approach to VTLS. They also preferred ranked retrieval to standard Boolean retrieval. Details follow: