The CSTR database is currently made available via a number of interfaces over the internet. These interfaces include any WWW or Mosaic client, gophers, FTP, and some POSTGRES interfaces. For those interested the WWW URL address is http://tr-ftp.CS.Berkeley.EDU/. FTP users can access the technical report database using anonymous@tr-ftp.CS.Berkeley.EDU.
We have also produced a graphical user interface to the electronic library. This interface has the capability of communicating with information servers using the Z39.50 communication protocol, and also to interact with POSTGRES using the libpq interface. The preliminary version of this user interface for the electronic library of technical reports is based on the Lassen Text Browser interface developed for the Sequoia 2000 project. This interface was designed to separate user interactions from the underlying search and retrieval engine, and to operate in a distributed environment where these components may be running on separate machines of a local or wide-area network. The interactive elements of this interface are based on the Tcl language and Tk toolkit interpreters[\protect\citeauthoryearOusterhout1991][\protect\citeauthoryearOusterhout1990], and thus provide a high degree of flexibility in laying out and modifying the interface elements. Because the interface is based on interpreted code, it can be changed virtually ``on-the-fly'' for testing and improvements, but still provides remarkably fast and efficient screen manipulation.
In addition, the Berkeley group is developing a visually-oriented browser for the electronic library called ``Tioga''[\protect\citeauthoryearStonebraker et al.1993]. This interface, developed for the Sequoia Project, provides a set of visualization tools for interacting with the database in new ways. In Tioga, the user links together boxes and arrows representing database or processing functions and data flows (rather like the paradigm used in many visual programming languages, and in such systems as Khoros). The builtin tools of Tioga permit the user to present graphically the results of these operations on the data. The user might, for example, generate a plot of term frequency for some selected terms vs. the date that the report was generated. The ad hoc sorts of data analysis available in Tioga make it a very powerful tool for a variety of context. We plan to give this browser the capability to generate requests in the Z39.50 standard protocol, and to add a variety of text analysis tools to its set of of ``ingredients''